Description:FORM 2
THE PATENTS ACT, 1970
(39 of 1970)
&
THE PATENT RULES, 2003
COMPLETE SPECIFICATION
(See Section 10 and Rule 13)
Title of Invention:
A SYSTEM AND A METHOD FOR CONTROLLED GENERATION OF SYNTHETIC FEATURES

APPLICANT:
QURE.AI TECHNOLOGIES PRIVATE LIMITED
An Indian entity having address as:
6th Floor, 606, Wing E, Times Square, Andheri-Kurla Road, Marol, Andheri (E), Marol Naka, Mumbai, Mumbai, Maharashtra, India, 400059

The following specification particularly describes the invention and the manner in which it is to be performed.

CROSS-REFERENCE TO RELATED APPLICATIONS AND PRIORITY
[0001] The present application does not claim priority from the Indian patent application.
TECHNICAL FIELD
[0002] The present disclosure relates to the field of medical image processing. More particularly, the disclosure pertains to a method for controlled generation of synthetic features in radiographic images using generative models.
BACKGROUND
[0003] This section is intended to introduce the reader to various aspects of art, which may be related to various aspects of the present disclosure that are described or claimed below. This discussion is believed to be helpful in providing the reader with background information to facilitate a better understanding of the various aspects of the present disclosure. Accordingly, it should be understood that these statements in this background section are to be read in this light, and not as admissions of prior art. Similarly, a problem mentioned in the background section or associated with the subject matter of the background section should not be assumed to have been previously recognized in the prior art. The subject matter in the background section merely represents different approaches, which in and of themselves may also correspond to implementations of the claimed technology.
[0004] Radiological image datasets are increasingly used for training and validating computer-aided diagnosis (CAD) systems, especially in the domain of thoracic imaging. Among various pathologies, pulmonary nodules are critical features that must be accurately detected, localized, and characterized due to their diagnostic relevance to lung cancer. However, obtaining large, balanced, and annotated datasets containing nodules with diverse appearances and perceptual characteristics is a known challenge. Annotated data with subtle or borderline nodules is particularly scarce, leading to performance gaps in CAD models, especially in edge-case scenarios.
[0005] Various approaches have been proposed to mitigate the scarcity of labeled data. Traditional data augmentation techniques such as rotation, cropping, flipping, and contrast variation are commonly used to artificially increase dataset diversity. However, these techniques apply global transformations that do not generate new clinically meaningful variations and fail to replicate the complex, localized patterns of real pulmonary nodules.
[0006] Handcrafted methods involving manual or rule-based insertion of synthetic nodules into existing radiographic images have also been explored. While these approaches offer a degree of control over visual characteristics, they often lack realism, time consuming and labor-intensive activity, possess intra-observer variability and may not integrate naturally with anatomical background structures, thereby limiting their utility for training robust models.
[0007] Synthetic data generation methods, particularly those using deep generative models such as Generative Adversarial Networks (GANs) and diffusion models, have been investigated to address this scarcity. These models are capable of generating realistic image patches that mimic true medical conditions. However, such models often lack controllability in terms of specific clinical attributes of interest, such as nodule location, shape, calcification, and perceptibility (subtlety). In many conventional approaches, the generation process is not transparent, nor is it easily customizable across different feature dimensions.
[0008] Diffusion-based generative models have demonstrated improved fidelity in medical image synthesis and offer advantages over GANs in terms of image quality and stability. However, current diffusion-based methods suffer from long training times and high computational cost. Moreover, such models often lack mechanisms for dynamic control of nodule attributes, making it difficult to tailor the synthetic data to specific diagnostic needs.
[0009] Moreover, methods for estimating perceptual subtlety in medical images often rely on heuristics or static thresholds, which do not reflect complex visual similarities in local patterns or textures.
[0010] The need therefore exists for a system and method that enables controlled generation of synthetic radiological features, such as lung nodules, with fine-grained modulation of both feature characteristics and perceptibility. It is further desirable for such a system to provide modular control interfaces and perceptual feedback mechanisms for enhancing dataset realism and diversity, especially for training robust AI models in the healthcare domain.
SUMMARY
[0011] Before the present system and method are described, it is to be understood that this disclosure is not limited to the system and its arrangement as described, as there can be multiple possible embodiments which are not expressly illustrated in the present disclosure.
[0012] The present disclosure overcomes one or more shortcomings of the prior art and provides additional advantages discussed throughout the present disclosure. Additional features and advantages are realized through the techniques of the present disclosure. Other embodiments and aspects of the disclosure are described in detail herein and are considered a part of the claimed disclosure.
[0013] It is also to be understood that the terminology used in the description is for the purpose of describing the versions or embodiments only and is not intended to limit the scope of the present application.
[0014] This summary is not intended to identify essential features of the claimed subject matter nor is it intended for use in classifying or limiting the scope of the claimed subject matter.
[0015] According to embodiments illustrated herein, a method for controlled generation of synthetic features in an image is disclosed. In one embodiment, the method may involve a step of receiving one or more input parameters. The one or more parameters may comprise a feature mask, one or more feature characteristics, a subtlety control value. Further, the method may involve a step of retrieving one or more adapters based on the one or more input parameters. Further, the method may involve a step of generating the controlled synthetic features by providing the feature mask and the one or more adapters to a generative diffusion model. Further, the method may involve step of providing the controlled synthetic features on the image.
[0016] According to embodiments illustrated herein, a system for controlled generation of synthetic features in the image is disclosed. In one embodiment, the system may involve a memory and a processor. Further, the processor may be coupled with the memory. Furthermore, the processor may be configured to execute programmed instructions stored in the memory. Furthermore, the processor is configured to receive the one or more input parameters. In one embodiment, the one or more parameters may comprise the feature mask, the one or more feature characteristics, and the subtlety control value. Furthermore, the processor may be configured to retrieve the one or more adapters based on the one or more input parameters. Furthermore, the processor may be configured to generate the controlled synthetic features by providing the feature mask and the one or more adapters to the generative diffusion model. Furthermore, the processor may be configured to provide the controlled synthetic features on the image.
[0017] According to embodiments illustrated herein, provided a non-transitory computer-readable storage medium for controlled generation of synthetic features in the image is disclosed. The non-transitory computer-readable storage medium having stored thereon, a set of computer-executable instructions causing a computer comprising one or more processors to perform various steps. The step may involve receiving the one or more input parameters. The one or more parameters may comprise the feature mask, the one or more feature characteristics, and the subtlety control value. Further, the step may involve retrieving the one or more adapters based on the one or more input parameters. Further, the step may involve generating the controlled synthetic features by providing the feature mask and the one or more adapters to the generative diffusion model. Furthermore, the step may involve providing the controlled synthetic features on the image.
[0018] The foregoing summary is illustrative only and is not intended to be in any way limiting. In addition to the illustrative aspects, embodiments, and features described above, further aspects, embodiments, and features will become apparent by reference to the drawings and the following detailed description.
BRIEF DESCRIPTION OF DRAWINGS
[0019] The detailed description is described with reference to the accompanying figures. In the Figures, the left-most digit(s) of a reference number identifies the Figure in which the reference number first appears. The same numbers are used throughout the drawings to refer like features and components.
[0020] Figure 1 illustrates a block diagram describing a system (100) for controlled generation of synthetic features in an image, in accordance with an embodiment of a present subject matter.
[0021] Figure 2 illustrates a block diagram showing an overview of an application server (101) for controlled generation of synthetic features in the image, in accordance with the embodiment of the present subject matter.
[0022] Figure 3 illustrates a flowchart describing a method (300) for controlled generation of synthetic features in the image, in accordance with an embodiment of the present subject matter.
[0023] Figure 4 illustrates a block diagram (400) of an exemplary computer system (401) for implementing embodiments consistent with the present subject matter.
[0024] It should be noted that the accompanying figures are intended to present illustrations of exemplary embodiments of the present disclosure. These figures are not intended to limit the scope of the present disclosure. It should also be noted that accompanying figures are not necessarily drawn to scale.
DETAILED DESCRIPTION
[0025] Reference throughout the specification to “various embodiments,” “some embodiments,” “one embodiment,” or “an embodiment” means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. Thus, appearances of the phrases “in various embodiments,” “in some embodiments,” “in one embodiment,” or “in an embodiment” in places throughout the specification are not necessarily all referring to the same embodiment. Furthermore, the features, structures or characteristics may be combined in any suitable manner in one or more embodiments.
[0026] The words "comprising," "having," "containing," and "including," and other forms thereof, are intended to be equivalent in meaning and be open ended in that an item or items following any one of these words is not meant to be an exhaustive listing of such item or items or meant to be limited to only the listed item or items. It must also be noted that, the singular forms "a," "an," and "the" include plural references unless the context clearly dictates otherwise. Although any methods similar or equivalent to those described herein can be used in the practice or testing of embodiments of the present disclosure, the exemplary methods are described. The disclosed embodiments are merely exemplary of the disclosure, which may be embodied in various forms.
[0027] The terminology “one or more synthetic nodules”, “synthetic nodules”, and “nodules” has the same meaning and are used alternatively throughout the specification. The terminology “feature mask”, “nodule mask”, and “mask” has the same meaning and are used alternatively throughout the specification. The terminology “one or more adapters”, “adapters”, and “adapter modules” has the same meaning and are used alternatively throughout the specification. The terminology “radiographic image” and “image” has the same meaning and are used alternatively throughout the specification unless specified otherwise.
[0028] An objective of the present disclosure is to provide a system and method for controlled generation of synthetic features, such as lung nodules, in medical images using a generative diffusion model.
[0029] Another objective of the present disclosure is to enable the generation of synthetic image regions with precise control over nodule characteristics including calcification, homogeneity, and border irregularity, thereby improving the quality and utility of synthetic training data for machine learning applications.
[0030] Yet another objective of the present disclosure is to provide a subtlety control mechanism that dynamically adjusts the visual perceptibility of the generated nodule using a control parameter derived from a perceptual score.
[0031] Yet another objective of the present disclosure is to support the generation of annotated synthetic image regions that can be used to train, validate, and benchmark machine learning models in the domain of lung cancer detection.
[0032] Yet another objective of the present disclosure is to improve the efficiency of data generation pipelines in medical imaging applications by enabling the reuse and composability of modular characteristics and subtlety control components.
[0033] Yet another objective of the present disclosure is to assist in the development of robust and explainable AI systems by systematically varying and annotating clinically relevant radiological features in synthetic data.
[0034] Referring to Figure 1 is a block diagram that illustrates the system (100) controlled generation of synthetic features in an image, in accordance with at least one embodiment of the present subject matter. The system (100) typically comprises an application server (101), a database server (102), a communication network (103), and a user computing device (104). The application server (101), the database server (102), and the user computing device (104) are typically communicatively coupled with each other via the communication network (103). In an embodiment, the application server (101) may communicate with the database server (102), and the user computing device (104) using one or more protocols such as, but not limited to, Hypertext Transfer Protocol (HTTP), Transmission Control Protocol/Internet Protocol (TCP/IP), Wireless Application Protocol (WAP), RF mesh, Bluetooth Low Energy (BLE), and the like, to communicate with one another.
[0035] In an embodiment, the database server (102) may refer to a computing device that is configured to store, manage, and serves structured and unstructured data required for the generation and projection of synthetic features in medical images. The database server (102) may be adapted to maintain large volumes of medical imaging datasets, including chest radiographs, annotated masks, synthetic feature characteristics, adapter module configurations, and corresponding metadata. In an embodiment, the database server (102) may be configured to receive feature-specific data and projection parameters from the application server (101), and to provide rapid retrieval of stored feature masks, model parameters, and adapter modules. The database server (102) may further support indexing, tagging, and retrieval of subtlety estimation metrics associated with perceptual realism of generated features.
[0036] In an exemplary embodiment, the database server (102) may be configured to support training datasets, ground truth records, and outputs of the generative diffusion model, enabling further fine-tuning or validation of synthetic feature generation. The database server (102) may also provide a centralized data access layer for real-time system operations and audit logging. In one embodiment, the database server (102) may include or interface with one or more database management technologies such as relational databases, NoSQL systems, distributed data lakes, or cloud-based storage platforms, optimized for high-throughput medical image processing.
[0037] In yet another embodiment, the database server (102) may be configured to implement secure access protocols, role-based data access, and periodic synchronization with federated or remote systems to ensure data integrity, security, and scalability for synthetic feature generation in clinical or research workflows.
[0038] A person with ordinary skills in art will understand that the scope of the disclosure is not limited to the database server (102) as a separate entity. In an embodiment, the functionalities of the database server (102) can be integrated into the application server (101) or into the user computing device (104).
[0039] In an embodiment, the application server (101) may refer to a computing device or a software framework hosting an application or a software service hosting an application configured to perform synthetic feature generation in radiographic images. In an embodiment, the application server (101) may be implemented to execute procedures including receiving one or more input parameters, retrieving adapter modules, invoking a generative diffusion model, and controlling the generation and projection of synthetic features. In an embodiment, the application server (101) may be realized through various types of application servers such as, but are not limited to, a Java application server, a .NET framework application server, a Base4 application server, a PHP framework application server, or any other application server framework.
[0040] In an embodiment, the application server (101) may be configured to manage the flow of data between the user computing device (104), the database server (102), and one or more AI-based generative models. The application server (101) may receive one or more input parameters including a feature mask, one or more feature characteristics, and a subtlety control value from the user computing device (104). The application server (101) may further retrieve one or more adapters based on at least one of the feature characteristics, the subtlety control value, or a combination thereof, from the database server (102).
[0041] In an embodiment, the application server (101) may be configured to provide the feature mask and the retrieved one or more adapters as input to a generative diffusion model to generate one or more controlled synthetic features in the image. The generated features may correspond to one or more synthetic nodules, and may be defined by their location, size, shape, texture, calcification, and border attributes as specified in the received feature characteristics.
[0042] In an embodiment, the application server (101) may be configured to apply one or more adapter modules selected from a set comprising Low-Rank Adaptation (LoRA) modules, adapter modules, prompt-tuning modules, hypernetwork modules, or prefix-tuning modules. The selected adapters may include one or more feature adapters and one or more subtlety adapters that are trained on the corresponding characteristics and subtlety estimation metrics.
[0043] In an embodiment, the application server (101) may interact with the generative diffusion model to control the perceptibility level of the synthetic features using the subtlety control value. The subtlety control value may correspond to one or more subtlety estimation metrics, including but not limited to Optimal Transport, Maximum Mean Discrepancy, Kullback–Leibler Divergence, Jensen–Shannon Divergence, Energy Distance, or Earth Mover’s Distance.
[0044] In an embodiment, the application server (101) may be configured to annotate the generated synthetic features on the radiographic image with corresponding metadata including the one or more feature characteristics and the subtlety control value. The annotated output may be transmitted to the user computing device (104) for visualization, storage, or further processing.
[0045] In an embodiment, the application server (101) may access or update adapter training data, feature mask libraries, and metadata logs stored in the database server (102) to support dynamic retrieval and generation operations. The application server (101) may maintain configuration files, version control of adapters, and inference parameters to ensure reproducibility of synthetic image generation.
[0046] In an embodiment, the application server (101) may be implemented as a standalone processing module or a distributed microservice that orchestrates the generation pipeline, from input reception to synthetic image generation and annotation. The server may further support task scheduling, load balancing, and secure communication protocols for interaction with the other system components.
[0047] In an embodiment, the communication network (103) may correspond to a communication medium through which the application server (101), the database server (102), and the user computing device (104) may communicate with each other. Such communication may be performed in accordance with various wired and wireless communication protocols. Examples of such wired and wireless communication protocols include, but are not limited to, Transmission Control Protocol and Internet Protocol (TCP/IP), User Datagram Protocol (UDP), Hypertext Transfer Protocol (HTTP), Wireless Application Protocol (WAP), File Transfer Protocol (FTP), ZigBee, EDGE, infrared IR), IEEE 802.11, 802.16, 2G, 3G, 4G, 5G, 6G, 7G cellular communication protocols, and/or Bluetooth (BT) communication protocols. The communication network (103) may either be a dedicated network or a shared network. Further, the communication network (103) may include a variety of network devices, including routers, bridges, servers, computing devices, storage devices, and the like. The communication network (103) may include, but is not limited to, the Internet, intranet, a cloud network, a Wireless Fidelity (Wi-Fi) network, a Wireless Local Area Network (WLAN), a Local Area Network (LAN), a cable network, the wireless network, a telephone network (e.g., Analog, Digital, POTS, PSTN, ISDN, xDSL), a telephone line (POTS), a Metropolitan Area Network (MAN), an electronic positioning network, an X.25 network, an optical network (e.g., PON), a satellite network (e.g., VSAT), a packet-switched network, a circuit-switched network, a public network, a private network, and/or other wired or wireless communications network configured to carry data.
[0048] In an embodiment, the user computing device (104) may comprise one or more processors and one or more memory. The one or more memory may include computer-readable instructions that may be executable by one or more processors to perform predetermined operations of the system (100). In an embodiment, the user computing device (104) may present a web user interface to display operations performed by the application server (101). The user computing device (104) may be configured to receive user inputs comprising one or more input parameters, such as a feature mask, one or more feature characteristics, and a subtlety control value. These inputs may be provided through a graphical user interface (GUI), a web-based platform, or a dedicated application interface.
[0049] In an embodiment, the user computing device (104) may be further configured to visualize radiographic images augmented with one or more synthetic features generated by the application server (101). The user computing device (104) may also display metadata annotations associated with the synthetic features, such as feature type, location, texture, or subtlety level. In an exemplary embodiment, the user computing device (104) may include tools to edit, approve, or reject the generated synthetic features, or to export them for diagnostic training, validation, or augmentation purposes. Examples of the user computing device (104) may include, but are not limited to, a desktop computer, a laptop, a tablet, a mobile device, or any other computing platform configured to interface with the system (100) through the communication network (103).
[0050] The system (100) may be implemented using hardware, software, or a combination of both, and may include one or more computer programs, web-based interfaces, or machine learning pipelines deployed on-premises or in cloud-based infrastructure. In an embodiment, the system (100) may comprise various modular components including the application server (101), the database server (102), the user computing device (104), and the communication network (103), which interact in a coordinated manner to support controlled generation of synthetic features in radiographic images.
[0051] In an embodiment, the system (100) may utilize a generative diffusion model and one or more adapter modules to synthesize radiographic features such as synthetic lung nodules, based on received input parameters including a feature mask, one or more feature characteristics, and a subtlety control value. The system (100) may be further configured to annotate the generated synthetic features on the radiographic image with corresponding metadata, and to facilitate user review or diagnostic application via the user computing device (104). In an embodiment, the system (100) may serve as the central processing and orchestration layer, dynamically retrieving adapter modules, invoking generative synthesis, managing input/output flows, and executing subtlety control logic. The system (100) is designed to concurrently handle multiple requests and manage adapter selection based on feature type and perceptibility level, enabling controlled and realistic augmentation of medical images. In an exemplary embodiment, the system (100) may be applied for generating synthetic radiographic datasets for medical training, model validation, or subtle lesion detection benchmarking.
[0052] Now referring to Figure 2, illustrates a block diagram showing an overview of various components of the application server (101) configured for controlled generation of synthetic features in an image, in accordance with at least one embodiment of the present subject matter. Figure 2 is explained in conjunction with elements from Figure 1.
[0053] In an embodiment, the application server (101) includes a processor (201), a memory (202), a transceiver (203), an input/output unit (204), a user interface unit (205), a parameter handling unit (206), a module management unit (207), a generation unit (208), and an image processing unit (209). The processor (201) may be communicatively coupled with the memory (202), the transceiver (203), the input/output unit (204), the user interface unit (205), the parameter handling unit (206), the module management unit (207), the generation unit (208), and the image processing unit (209). In an embodiment, the transceiver (203) may be configured to communicate with the communication network (103) of the system (100), thereby enabling interaction with the database server (102) and the user computing device (104). The memory (202) may store input parameters, adapter configurations, and model execution logs. Each of the specialized units within the application server (101) may be configured to execute one or more functions that correspond to method steps as claimed in the present disclosure, including receiving parameters, retrieving adapter modules, generating synthetic features, and annotating the resulting radiographic image.
[0054] In an embodiment, the present subject matter discloses a system (100) for controlled generation of synthetic features in an image. The system (100) comprises an application server (101), a database server (102), a user computing device (104), and a communication network (103) configured to enable interaction among these components. The system (100) is designed to receive one or more input parameters, retrieve one or more adapters based on the received input, and generate synthetic features within an image using a generative diffusion model. The generated features are integrated within the image based on control instructions derived from the input parameters.
[0055] In an embodiment, the processor (201) comprises suitable logic, circuitry, and/or code that may be configured to execute instructions stored in the memory (202) for performing one or more operations related to the generation and insertion of synthetic features in an image. The processor (201) may coordinate execution across the various functional units and ensure synchronization of tasks such as input parameter processing, adapter retrieval, diffusion-based generation, and final image annotation. Examples of the processor (201) include, but not limited to, a standard microprocessor, microcontroller, central processing unit (CPU), an X86-based processor, a Reduced Instruction Set Computing (RISC) processor, an Application- Specific Integrated Circuit (ASIC) processor, and a Complex Instruction Set Computing (CISC) processor, distributed or cloud processing unit, state machines, logic circuitries, and/or any devices that manipulate signals based on operational instructions and/or other processing logic that accommodates the requirements of the present invention.
[0056] The memory (202) comprises suitable logic, circuitry, interfaces, and/or code that may be configured to store the set of instructions, which are executed by the processor (201). Preferably, the memory (202) is configured to store one or more programs, routines, or scripts that are executed in coordination with the processor (201). Additionally, the memory (202) may include any computer-readable medium or computer program product known in the art including, for example, volatile memory, such as static random-access memory (SRAM) and dynamic random-access memory (DRAM), and/or non-volatile memory, such as read-only memory (ROM), erasable programmable ROM, a Hard Disk Drive (HDD), flash memories, Secure Digital (SD) card, Solid State Disks (SSD), optical disks, magnetic tapes, memory cards, virtual memory and distributed cloud storage. The memory (202) may be removable, non-removable, or a combination of the same. Further, the memory (202) may include routines, programs, objects, components, data structures, etc., which perform particular tasks or implement particular abstract data types. The memory (202) may include programs or coded instructions that supplement the applications and functions of the system (100). In one embodiment, the memory (202), amongst other things, serves as a repository for storing data processed, received, and generated by one or more of the programs or the coded instructions. In yet another embodiment, the memory (202) may be managed under a federated structure that enables the adaptability and responsiveness of the application server (101).
[0057] The transceiver (203) comprises suitable logic, circuitry, interfaces, and/or code that may be configured to receive, process or transmit information, data or signals, which are stored by the memory (202) and executed by the processor (201). The transceiver (203) is preferably configured to receive, process or transmit one or more programs, routines, or scripts that are executed in coordination with the processor (201). The transceiver (203) is preferably communicatively coupled to the communication network (103) of the system (100) for communicating all the information, data, signals, programs, routines or scripts through the communication network (103).
[0058] The transceiver (203) may implement one or more known technologies to support wired or wireless communication with the communication network (103). In an embodiment, the transceiver (203) may include but is not limited to, an antenna, a radio frequency (RF) transceiver, one or more amplifiers, a tuner, one or more oscillators, a digital signal processor, a Universal Serial Bus (USB) device, a coder-decoder (CODEC) chipset, a subscriber identity module (SIM) card, and/or a local buffer. Also, the transceiver (203) may communicate via wireless communication with networks, such as the Internet, an Intranet and/or a wireless network, such as a cellular telephone network, a wireless local area network (LAN) and/or a metropolitan area network (MAN). Accordingly, the wireless communication may use any of a plurality of communication standards, protocols and technologies, such as: Global System for Mobile Communications (GSM), Enhanced Data GSM Environment (EDGE), wideband code division multiple access (W-CDMA), code division multiple access (CDMA), time division multiple access (TDMA), Bluetooth, Wireless Fidelity (Wi-Fi) (e.g., IEEE 802.11a, IEEE 802.11b, IEEE 802.11g and/or IEEE 802.11n), voice over Internet Protocol (VoIP), Wi-MAX, a protocol for email, instant messaging, and/or Short Message Service (SMS).
[0059] The transceiver (203) may implement one or more known technologies to support wired or wireless communication with the communication network (103). In an embodiment, the transceiver (203) may include but is not limited to, an antenna, a radio frequency (RF) transceiver, one or more amplifiers, a tuner, one or more oscillators, a digital signal processor, a Universal Serial Bus (USB) device, a coder-decoder (CODEC) chipset, a subscriber identity module (SIM) card, and/or a local buffer. Also, the transceiver (203) may communicate via wireless communication with networks, such as the Internet, an Intranet and/or a wireless network, such as a cellular telephone network, a wireless local area network (LAN) and/or a metropolitan area network (MAN). Accordingly, the wireless communication may use any of a plurality of communication standards, protocols and technologies, such as: Global System for Mobile Communications (GSM), Enhanced Data GSM Environment (EDGE), wideband code division multiple access (W-CDMA), code division multiple access (CDMA), time division multiple access (TDMA), Bluetooth, Wireless Fidelity (Wi-Fi) (e.g., IEEE 802.11a, IEEE 802.11b, IEEE 802.11g and/or IEEE 802.11n), voice over Internet Protocol (VoIP), Wi-MAX, a protocol for email, instant messaging, and/or Short Message Service (SMS).
[0060] The input/output (I/O) unit (204) comprises suitable logic, circuitry, interfaces, and/or code that may be configured to receive or present information. The input/output unit (204) comprises various input and output devices that are configured to communicate with the processor (201). Examples of the input devices include but are not limited to, a keyboard, a mouse, a joystick, a touch screen, a microphone, a camera, and/or a docking station. Examples of the output devices include, but are not limited to, a display screen and/or a speaker. The I/O unit (204) may include a variety of software and hardware interfaces, for example, a web interface, a graphical user interface, and the like. The I/O unit (204) may allow the system (100) to interact with the user directly or through the user computing devices (104). Further, the I/O unit (204) may enable the system (100) to communicate with other computing devices, such as web servers and external data servers (not shown). The I/O unit (204) can facilitate multiple communications within a wide variety of networks and protocol types, including wired networks, for example, LAN, cable, etc., and wireless networks, such as WLAN, cellular, or satellite. The I/O unit (204) may include one or more ports for connecting a number of devices to one another or to another server. In one embodiment, the I/O unit (204) allows the application server (101) to be logically coupled to other user computing devices (104), some of which may be built in. Illustrative components include tablets, mobile phones, wireless devices, etc.
[0061] In an embodiment, the input/output unit (204) facilitates the reception of input parameters from external sources and the delivery of output images or logs. It may support interfaces such as APIs, file upload/download services, and web-based dashboards. The I/O unit (204) may also handle communication between internal units of the application server (101) for efficient task execution.
[0062] In an embodiment, the user interface unit (205) is configured to enable interactive communication with the user computing device (104). The unit may allow users to upload input images, specify feature masks, define generation parameters, and visualize the final annotated images. The unit may also display confidence metrics, generation logs, and preview overlays for the inserted synthetic features.
[0063] In an embodiment, the synthetic features correspond to one or more synthetic lung nodules. In such an embodiment, the feature mask may be referred to as a nodule mask, which may comprise a pixel-wise representation typically binary that specifies the location and shape of a nodule within a radiographic image. The nodule mask is used to define the placement region for the controlled synthetic nodule within the image. The system (100) may be configured to receive the feature mask and generate the one or more synthetic nodules in accordance with the feature characteristics and the subtlety control value as defined in the one or more input parameters. The generated nodules are subsequently embedded into images such as chest radiographs, including chest CT or chest X-ray images, ensuring anatomical and structural realism consistent with the received configuration.
[0064] In an embodiment, the parameter handling unit (206) may be configured to receive and process one or more input parameters for generating synthetic features in a medical image. The input parameters may correspond to nodule characteristics such as texture, shape, size, location, and boundary attributes. In an exemplary embodiment, the texture may include one of ground-glass, solid, part-solid, or calcified types. In an embodiment, the shape may include spherical, ovoid, or elongated forms, and the boundary attributes may include diffused, spiculated, or sharply defined edges. The parameter handling unit (206) may validate the received input values, perform consistency checks, and transform them into a standard internal representation compatible with downstream modules. In an embodiment, the parameter handling unit (206) may further include predefined templates or rules for guiding the generation of synthetic nodules with varying degrees of subtlety and realism based on diagnostic requirements or training use cases.
[0065] In an embodiment, the module management unit (207) may be configured to retrieve one or more adapters from the memory (202) based on at least one of the feature characteristics, the subtlety control value, or a combination thereof. The one or more adapters may include, but are not limited to, a Low-Rank Adaptation (LoRA) module, an adapter module, a prompt-tuning module, a hypernetwork module, a prefix-tuning module, or any combination thereof. These adapters may be utilized to condition or fine-tune a generative diffusion model, thereby enabling the synthesis of synthetic features with desired structural and visual attributes.
[0066] In an embodiment, the module management unit (207) may maintain a registry of model adapters or plug-in components, each corresponding to specific anatomical scenarios, feature configurations, or imaging contexts. For instance, distinct adapters may be associated with anatomical regions exhibiting pulmonary emphysema, pleural proximity, or high-opacity zones. The module management unit (207) dynamically selects and loads the appropriate adapter(s) based on the anatomical context, the nodule placement location, and one or more input conditions including clinical constraints.
[0067] In another embodiment, the one or more adapters retrieved by the module management unit (207) may comprise at least one of feature adapters and subtlety adapters, also referred to as a subtlety slider. Each of the one or more feature adapters may be trained on a corresponding feature characteristic from the one or more feature characteristics, such as texture, shape, calcification, or border definition. Each of the one or more subtlety adapters may be trained on at least one of the one or more feature characteristics, one or more subtlety estimation metrics, or a combination thereof. These adapters are designed to modulate the latent space of the generative diffusion model, thereby controlling both the structural attributes and visual perceptibility of the synthetic features. The feature adapters inject domain-specific characteristics into the generation process, while the subtlety adapters enable fine-tuned control over the visibility of the synthetic nodules. This modular configuration allows the generation framework to adapt across varying anatomical scenarios and diagnostic use cases with precision.
[0068] In an exemplary embodiment, the one or more plug-and-play modules, such as feature adapters and subtlety adapters, may be combined at inference time to generate a composite synthetic feature. This feature may incorporate a combination of multiple characteristics and subtlety levels within a single synthetic region, enabling simulation of complex visual pathologies or multi-feature training examples. The module management unit (207) may coordinate the composition of such adapters to ensure that their latent contributions align with the anatomical constraints and user-defined parameters.
[0069] In an embodiment, the generation unit (208) may be configured to generate one or more controlled synthetic features, such as nodules, at a designated location in an image. The image may include, but is not limited to, a medical image such as a computed tomography (CT) scan, a chest X-ray (CXR), or any diagnostic imaging modality. The generation unit (208) may be configured to coordinate the execution of a generative diffusion model, such as a denoising diffusion probabilistic model (DDPM) or a latent diffusion model, based on one or more parameters processed by the parameter handling unit (206) and one or more adapters retrieved by the module management unit (207).
[0070] In an embodiment, the synthetic feature generation process executed by the generation unit (208) includes a forward diffusion step, wherein noise is incrementally added to the image region to simulate a corrupted sample. This is followed by a reverse denoising process guided by a trained neural network within a diffusion model architecture. The neural network progressively removes the noise, reconstructing a synthetic image region with desired characteristics and subtlety based on the conditioning inputs, such as feature masks and adapter modules.
[0071] In an embodiment, the generated synthetic image regions may be annotated with metadata that describes the one or more feature characteristics (e.g., calcification, texture, border definition) and the subtlety control value used during generation. This metadata facilitates traceability and enables automated dataset labeling for downstream machine learning applications such as classification, segmentation, or diagnostic support systems. Such annotations are particularly valuable in benchmarking model performance on varying levels of visual difficulty.
[0072] In another embodiment, the system (100) may be configured to generate a plurality of synthetic image regions, each corresponding to a different combination of feature mask, feature characteristics, and subtlety control values. Each generated region is annotated accordingly, thereby enabling the creation of enriched and diversified training datasets. Such capability supports balanced representation of rare or hard-to-detect patterns, ultimately improving the robustness and generalizability of clinical AI models trained on the augmented datasets.
[0073] In an embodiment, the generative diffusion model receives as input a sample image, one or more feature masks, one or more adapter weights derived from one or more adapters, or a combination thereof. The feature mask defines the location, size, and shape for placing the synthetic feature, and the adapter weights modulate the diffusion process based on desired characteristics and subtlety.
[0074] In various embodiments, the system and method disclosed herein are applicable to a variety of image modalities including, but not limited to, computed tomography (CT), magnetic resonance imaging (MRI), chest X-ray (CXR), and other diagnostic imaging techniques. References to “image” throughout this disclosure shall be interpreted to include such modalities unless otherwise explicitly stated.
[0075] In an embodiment, the generative diffusion model may be fine-tuned using adapter-based learning techniques, such as Low-Rank Adaptation (LoRA). The adapters are lightweight modules containing task-specific weights that are combined with the base model at inference time, thereby allowing efficient customization without altering the base model. In an embodiment, the LoRA adapters may offer advantages including parameter efficiency, modularity, and deploy ability. Specifically, they enable support for domain-specific styles or user-specific preferences, while preserving the integrity of the underlying diffusion model. The use of such adapters allows the generation unit (208) to dynamically generate synthetic features tailored to different clinical contexts, feature types, or training objectives, with minimal computational overhead.
[0076] In an embodiment, the one or more adapters retrieved by the module management unit (207) may comprise Low-Rank Adaptation (LoRA) modules that fine-tune the generative diffusion model without modifying the original pre-trained model weights. These adapters are trained separately and integrated at inference time to produce synthetic features with customized visual and structural characteristics. Each LoRA adapter produces a compact set of adapter weights, typically a few megabytes in size, instead of a complete fine-tuned model, thereby enabling a lightweight and efficient architecture. LoRA-based tuning provides significant parameter efficiency by training only a small fraction (e.g., 0.1%–5%) of the model’s parameters, resulting in reduced training cost, memory requirements, and storage footprint. Moreover, LoRA adapters offer high modularity, allowing different domain-specific or feature-specific adapters to be swapped without re-training the base diffusion model. For example, separate LoRA adapters may be used to generate synthetic features corresponding to different anatomical regions or subtlety levels. LoRA-based customization further enables personalizing the generative output based on diagnostic intent, image context, or user preferences, while preserving the integrity of the base model. The trained LoRA adapters may be deployed as discrete weight files (e.g., .pt, .ckpt, .safetensors) and combined with the base generative model at inference time to produce task-specific or stylistically customized synthetic outputs.
[0077] In an exemplary embodiment, the core functionality of the generative diffusion model involves transforming random noise into realistic synthetic patches within the image. This is achieved through a forward diffusion step in which noise is progressively added to the image or patch, followed by a reverse denoising process guided by a neural network that incrementally reconstructs the feature. During training, the generative diffusion model may be conditioned on nodule masks to learn the spatial and structural distribution of pathological regions. The trained model, at inference time, is conditioned on the received feature mask and the adapter weights to produce a controlled synthetic output with the desired size, shape, texture, boundary attributes, and subtlety level, as defined by the input parameters.
[0078] In an exemplary embodiment, the generation unit (208) may also include a subtlety control mechanism that adjusts the perceptibility of the synthetic feature based on one or more contextual factors such as anatomical depth, proximity to dense tissues (e.g., bone), or a user-defined realism index. The generation unit (208) may apply the synthetic feature to the original image at the specified location with anatomical consistency and desired visual subtlety. The final output includes a modified image comprising the embedded synthetic feature that conforms to the characteristics and perceptibility defined by the user.
[0079] In an embodiment, the one or more subtlety estimation metrics are computed using a technique that takes as input a nodule image and its corresponding mask to produce a scalar value representing the perceptual subtlety of the nodule. These metrics are configured to capture intensity variations and neighborhood pattern features within the region of interest.
[0080] In an exemplary embodiment, such subtlety estimation may be performed using one or more machine learning techniques capable of quantifying distributional or structural discrepancies between real and synthetic nodules. In another embodiment, the machine learning techniques may include, but are not limited to, Optimal Transport (OT), Maximum Mean Discrepancy (MMD), Kullback–Leibler Divergence (KL Divergence), Jensen–Shannon Divergence (JSD), Earth Mover’s Distance (EMD), Energy Distance, Chamfer Distance, Cramer Distance, f-Divergences, Sinkhorn Distance, Entropic OT, or Wasserstein Distance. The resulting subtlety metric serves as the basis for generating a subtlety control value. In an embodiment, the subtlety control value may correspond to one of the computed subtlety estimation metrics and is subsequently used to regulate the perceptibility of the synthetic feature during the generation process, enabling fine-grained control over visibility and realism of the output features.
[0081] In an exemplary embodiment, OT subtlety routine produces a single number that indicates how much the appearance of an X-ray or CT image changes after a synthetic nodule is generated. The number reflects the visual impact of the inserted nodule on its local surroundings. Alternatively, the number is also termed as a "noticeability score". A small number indicates the added nodule is hard to spot and a large number indicates the nodule is easy to see.
[0082] In another embodiment, the system (100) may be configured to annotate the image after insertion of the controlled synthetic features. The annotation may include metadata corresponding to the feature characteristics and subtlety control value, which may be used for downstream processing, training, or evaluation purposes.
[0083] In an embodiment, the image on which the one or more controlled synthetic features are inserted corresponds to a radiographic image. The radiographic image may include, but is not limited to, plain radiographs, fluoroscopy images, contrast-enhanced radiographs, computed radiography (CR) images, digital radiography (DR) images, mammograms, and cone-beam computed tomography (CBCT) images. In one embodiment, the radiographic image may be a chest X-ray, however, the system (100) is not limited to chest imaging and may process and generate synthetic features on radiographic images corresponding to various anatomical regions, including but not limited to brain X-rays, abdominal X-rays, pelvic radiographs, spinal X-rays, skeletal system images, dental radiographs, craniofacial X-rays, and extremity images (e.g., hand, foot, or limb radiographs). The flexibility to operate on these various radiographic modalities enables the system (100) to simulate realistic and diverse clinical scenarios, thereby supporting a wide range of diagnostic, training, benchmarking, and data augmentation use cases. The synthetic features generated on such radiographic images may be configured in terms of anatomical placement, size, shape, texture, and subtlety, as defined in the received input parameters.
[0084] Now referring to figure 3, a flowchart describing a method (300) for controlled generation of synthetic features in an image , in accordance with at least one embodiment of the present subject matter. The flowchart is described in conjunction with Figure 1 and Figure 2. The method (300) starts at step (301) and proceeds to step (304).
[0085] At step (301), the method (300) comprises receiving one or more input parameters, wherein the input parameters comprise at least a feature mask, one or more feature characteristics, and a subtlety control value.
[0086] At step (302), the method (300) includes retrieving one or more adapters based on the received input parameters.
[0087] At step (303), the method (300) includes generating the controlled synthetic features by providing the feature mask and the one or more adapters to a generative diffusion model.
[0088] At step (304), the method (300) includes providing the controlled synthetic features on the image.
[0089] Let us delve into a detailed example of the present disclosure.
[0090] Working Example 1:
[0091] Consider a hospital research team developing a computer-aided diagnosis (CAD) system to assist radiologists in identifying lung nodules from chest X-rays. To improve the performance of their AI model, the team faces a critical issue: the lack of diverse, high-quality training images, particularly those with rare or subtle nodule presentations.
[0092] To address this challenge, the team adopts the disclosed method for controlled generation of synthetic features within radiographic images. The process begins by defining a set of parameters that control the appearance and placement of synthetic nodules. These parameters include a feature mask (defining size, shape, and location within the image), a set of nodule characteristics (such as texture type, edge sharpness, or calcification), and a subtlety value (indicating the desired visibility or perceptibility of the nodule).
[0093] Based on these parameters, the system retrieves specialized tuning components referred to as adapters, that adjust the behavior of the image-generation model. For example, one adapter may be trained to generate nodules with sharp, irregular borders, while another controls the subtlety level to simulate low-contrast nodules often missed in real scans.
[0094] These adapters are then applied to a generative diffusion model, which creates the synthetic nodule. The model operates by first introducing noise to a sample image patch (simulating a forward diffusion step) and then gradually removing this noise (in a reverse denoising process), guided by a trained neural network and conditioned on the input parameters and adapter configurations.
[0095] As a result, the model generates a synthetic nodule that matches the desired location, shape, characteristics, and subtlety level. The generated nodule is seamlessly blended into the original X-ray image, ensuring anatomical realism and diagnostic utility.
[0096] The final image is annotated with metadata that includes the parameter values used for generation, allowing traceability and enabling the image to be used as labeled training data. This process is repeated across multiple parameter configurations to create a diverse dataset.
[0097] Using this approach, the hospital team generates thousands of synthetic radiographs with controlled variations. These are then used to train a deep learning model, which is later deployed in clinical settings. During testing, the model demonstrates improved sensitivity to subtle and irregular nodules—an outcome attributed to the enhanced diversity and control offered by the synthetic image generation process.
[0098] This example illustrates how the disclosed invention enables controlled and efficient creation of synthetic image features, directly addressing limitations in medical AI training and contributing to improved diagnostic performance.
[0099] Working Example 2:
[0100] Consider a scenario in which a healthcare AI company, MedSynthAI, is developing a diagnostic support tool for early lung cancer detection in rural telemedicine setups. Due to limited access to high-quality annotated radiographic data, especially those depicting rare or subtle lung nodules, MedSynthAI incorporates the disclosed invention to synthetically enrich its training database.
[0101] The system receives as input a batch of digital chest radiographs (CXR images) from a public data repository. A preprocessing pipeline standardizes the images in terms of resolution, grayscale intensity, and lung segmentation. These preprocessed images are forwarded to the application server (101), where the image processing pipeline is initiated.
[0102] The parameter handling unit (206) receives user-specified input parameters for synthetic nodule generation. These include a feature mask defining the insertion region (e.g., lower right lobe), feature characteristics (e.g., ovoid shape, spiculated edge, calcified texture), and a subtlety control value (e.g., 0.3 on a scale of 0 to 1). The unit validates and transforms these parameters into a standardized internal representation.
[0103] The module management unit (207) then retrieves appropriate plug-and-play adapter modules from memory. These include a LoRA-based feature adapter tuned to generate spiculated edges, and a subtlety adapter trained to regulate perceptual visibility based on prior data. The modules are composed dynamically based on the current task.
[0104] The generation unit (208) invokes a pre-trained generative diffusion model, supplying it with the input radiograph patch, the selected adapter weights, and the defined feature mask. The model performs a forward diffusion process where noise is added to the patch, followed by a reverse denoising process guided by the neural network and modulated by the adapters.
[0105] In this case, the subtlety adapter reduces the visual prominence of the nodule such that it mimics borderline-detectable nodules often missed by generalist algorithms. The resulting synthetic patch contains a nodule with controlled size, texture, boundary, and visibility.
[0106] The synthetic patch is merged back into the full radiograph at the coordinates defined by the mask. An image processing unit (209) ensures pixel blending, histogram matching, and spatial smoothing to maintain anatomical realism.
[0107] Finally, the system generates metadata for the synthesized image, including a unique image ID, applied feature parameters, adapter configurations, and subtlety metric value. This metadata is appended in a machine-readable format, such as JSON or XML, suitable for integration into AI training pipelines.
[0108] The system repeats this process across hundreds of variations, automatically creating a balanced dataset of synthetic CXRs containing diverse nodule types. These annotated images are used to train a deep convolutional neural network (CNN) that can now detect subtle nodules with higher sensitivity during tele-radiology diagnosis.
[0109] The described scenario demonstrates the application of the invention in a real-world training setup, offering configurable nodule synthesis that adapts to clinical complexity and improves AI readiness in low-data settings.
[0110] A person skilled in the art will understand that the scope of the present disclosure is not limited to scenarios based on the aforementioned factors and techniques. The example provided is illustrative and does not limit the breadth of the invention.
[0111] Now referring to Figure 4, illustrates a block diagram (400) of an exemplary computer system (401) for implementing embodiments consistent with the present disclosure. Variations of computer system (401) may be used as the method for controlled generation of synthetic features in an image. The computer system (401) may comprise a central processing unit (“CPU” or “processor”) (402). The processor (402) may comprise at least one data processor for executing program components for executing user or system-generated requests. A user may include a person, a person using a device such as those included in this disclosure, or such a device itself. Additionally, the processor (402) may include specialized processing units such as integrated system (bus) controllers, memory management control units, floating point units, graphics processing units, digital signal processing units, or the like. In various implementations the processor (402) may include a microprocessor, such as AMD Athlon, Duron or Opteron, ARM’s application, embedded or secure processors, IBM PowerPC, Intel’s Core, Itanium, Xeon, Celeron or other line of processors, for example. Accordingly, the processor (402) may be implemented using mainframe, distributed processor, multi-core, parallel, grid, or other architectures. Some embodiments may utilize embedded technologies like application-specific integrated circuits (ASICs), digital signal processors (DSPs), or Field Programmable Gate Arrays (FPGAs), for example.
[0112] Processor (402) may be disposed in communication with one or more input/output (I/O) devices via I/O interface (403). Accordingly, the I/O interface (403) may employ communication protocols/methods such as, without limitation, audio, analog, digital, monoaural, RCA, stereo, IEEE-1394, serial bus, universal serial bus (USB), infrared, PS/2, BNC, coaxial, component, composite, digital visual interface (DVI), high-definition multimedia interface (HDMI), RF antennas, S-Video, VGA, IEEE 802.n /b/g/n/x, Bluetooth, cellular (e.g., code-division multiple access (CDMA), high-speed packet access (HSPA+), global system for mobile communications (GSM), long-term evolution (LTE), WiMAX, or the like, for example.
[0113] Using the I/O interface (403), the computer system (401) may communicate with one or more I/O devices. For example, the input device (404) may be an antenna, keyboard, mouse, joystick, (infrared) remote control, camera, card reader, fax machine, dongle, biometric reader, microphone, touch screen, touchpad, trackball, sensor (e.g., accelerometer, light sensor, GPS, gyroscope, proximity sensor, or the like), stylus, scanner, storage device, transceiver, video device/source, or visors, for example. Likewise, an output device (405) may be a user’s smartphone, tablet, cell phone, laptop, printer, fax machine, video display (e.g., cathode ray tube (CRT), liquid crystal display (LCD), light- emitting diode (LED), plasma, or the like), or audio speaker, for example. In some embodiments, a transceiver (406) may be disposed in connection with the processor (402). The transceiver (406) may facilitate various types of wireless transmission or reception. For example, the transceiver (406) may include an antenna operatively connected to a transceiver chip (example devices include the Texas Instruments® WiLink WL1283, Broadcom® BCM4750IUB8, Infineon Technologies® X-Gold 618-PMB9800, or the like), providing IEEE 802.11a/b/g/n, Bluetooth, FM, global positioning system (GPS), and/or 2G/3G/5G/6G HSDPA/HSUPA communications, for example.
[0114] In some embodiments, the processor (402) may be disposed in communication with a communication network (408) via a network interface (407). The network interface (407) is adapted to communicate with the communication network (408). The network interface, coupled to the processor may be configured to facilitate communication between the system and one or more external devices or networks. The network interface (407) may employ connection protocols including, without limitation, direct connect, Ethernet (e.g., twisted pair 10/100/1000 Base T), transmission control protocol/internet protocol (TCP/IP), token ring, or IEEE 802.11a/b/g/n/x, for example. The communication network (408) may include, without limitation, a direct interconnection, local area network (LAN), wide area network (WAN), wireless network (e.g., using Wireless Application Protocol), or the Internet, for example. Using the network interface (407) and the communication network (408), the computer system (401) may communicate with devices such as shown as a laptop (409) or a mobile/cellular phone (410). Other exemplary devices may include, without limitation, personal computer(s), server(s), fax machines, printers, scanners, various mobile devices such as cellular telephones, smartphones (e.g., Apple iPhone, Blackberry, Android-based phones, etc.), tablet computers, eBook readers (Amazon Kindle, Nook, etc.), laptop computers, notebooks, gaming consoles (Microsoft Xbox, Nintendo DS, Sony PlayStation, etc.), or the like. In some embodiments, the computer system (401) may itself embody one or more of these devices.
[0115] In some embodiments, the processor (402) may be disposed in communication with one or more memory devices (e.g., RAM 413, ROM 414, etc.) via a storage interface (412). The storage interface (412) may connect to memory devices including, without limitation, memory drives, removable disc drives, etc., employing connection protocols such as serial advanced technology attachment (SATA), integrated drive electronics (IDE), IEEE-1394, universal serial bus (USB), fiber channel, small computer systems interface (SCSI), etc. The memory drives may further include a drum, magnetic disc drive, magneto-optical drive, optical drive, redundant array of independent discs (RAID), solid-state memory devices, or solid-state drives, for example.
[0116] The memory devices may store a collection of program or database components, including, without limitation, an operating system (416), user interface application (417), web browser (418), mail client/server (419), user/application data (420) (e.g., any data variables or data records discussed in this disclosure) for example. The operating system (416) may facilitate resource management and operation of the computer system (401). Examples of operating systems include, without limitation, Apple Macintosh OS X, UNIX, Unix-like system distributions (e.g., Berkeley Software Distribution (BSD), FreeBSD, NetBSD, OpenBSD, etc.), Linux distributions (e.g., Red Hat, Ubuntu, Kubuntu, etc.), IBM OS/2, Microsoft Windows (XP, Vista/7/8, etc.), Apple iOS, Google Android, Blackberry OS, or the like.
[0117] The user interface (417) is for facilitating the display, execution, interaction, manipulation, or operation of program components through textual or graphical facilities. For example, user interfaces may provide computer interaction interface elements on a display system operatively connected to the computer system (401), such as cursors, icons, check boxes, menus, scrollers, windows, or widgets, for example. Graphical user interfaces (GUIs) may be employed, including, without limitation, Apple Macintosh operating systems’ Aqua, IBM OS/2, Microsoft Windows (e.g., Aero, Metro, etc.), Unix X-Windows, or web interface libraries (e.g., ActiveX, Java, JavaScript, AJAX, HTML, Adobe Flash, etc.), for example.
[0118] In some embodiments, the computer system (401) may implement a web browser (418) stored program component. The web browser (418) may be a hypertext viewing application, such as Microsoft Internet Explorer, Google Chrome, Mozilla Firefox, Apple Safari, or Microsoft Edge, for example. Secure web browsing may be provided using HTTPS (secure hypertext transport protocol), secure sockets layer (SSL), Transport Layer Security (TLS), or the like. Web browsers may utilize facilities such as AJAX, DHTML, Adobe Flash, JavaScript, Java, or application programming interfaces (APIs), for example. In some embodiments the computer system (401) may implement a mail client/server (419) stored program component. The mail server (419) may be an Internet mail server such as Microsoft Exchange, or the like. The mail server may utilize facilities such as ASP, ActiveX, ANSI C++/C#, Microsoft .NET, CGI scripts, Java, JavaScript, PERL, PHP, Python, or WebObjects, for example. The mail server (419) may utilize communication protocols such as internet message access protocol (IMAP), messaging application programming interface (MAPI), Microsoft Exchange, post office protocol (POP), simple mail transfer protocol (SMTP), or the like. In some embodiments, the computer system (401) may implement a mail client (420) stored program component. The mail client (420) may be a mail viewing application, such as Apple Mail, Microsoft Entourage, Microsoft Outlook, or Mozilla Thunderbird.
[0119] In some embodiments, the computer system (401) may store user/application data (421), such as the data, variables, records, or the like as described in this disclosure. Such databases may be implemented as fault-tolerant, relational, scalable, secure databases such as Oracle or Sybase, for example. Alternatively, such databases may be implemented using standardized data structures, such as an array, hash, linked list, struct, structured text file (e.g., XML), table, or as object-oriented databases (e.g., using ObjectStore, Poet, Zope, etc.). Such databases may be consolidated or distributed, sometimes among the various computer systems discussed above in this disclosure. It is to be understood that the structure and operation of the any computer or database component may be combined, consolidated, or distributed in any working combination.
[0120] Furthermore, one or more computer-readable storage media may be utilized in implementing embodiments consistent with the present invention. A computer-readable storage medium refers to any type of physical memory on which information or data readable by a processor may be stored. Thus, a computer-readable storage medium may store instructions for execution by one or more processors, including instructions for causing the processor(s) to perform steps or stages consistent with the embodiments described herein. The term “computer- readable medium” should be understood to include tangible items and exclude carrier waves and transient signals, i.e., non-transitory. Examples include Random Access Memory (RAM), Read- Only Memory (ROM), volatile memory, non-volatile memory, hard drives, Compact Disc (CD) ROMs, Digital Video Disc (DVDs), flash drives, disks, and any other known physical storage media.
[0121] Various embodiments of the present disclosure provide a non-transitory computer-readable medium, and/or a non-transitory machine-readable storage medium, having stored thereon instructions comprising machine-executable code. When executed by one or more processors, the instructions cause the machine to perform a method for controlled generation of synthetic features in an image, such as a radiographic image. In an embodiment, the machine-executable instructions cause the processor to perform the steps including: receiving one or more input parameters, the input parameters comprising a feature mask, one or more feature characteristics, and a subtlety control value. The processor then retrieves one or more adapters from memory, based on the received parameters. These adapters are selected from a set of parameter-efficient fine-tuning modules such as Low-Rank Adaptation (LoRA), hypernetwork modules, or prompt-tuning modules, corresponding to specific nodule characteristics and subtlety levels. Subsequently, the processor provides the feature mask and the adapter weights to a generative diffusion model, which is configured to synthesize controlled synthetic features within the image. The generative diffusion model progressively denoises random noise guided by the feature mask and adapter weights, thereby generating image regions exhibiting the desired structural attributes and perceptibility. The processor integrates the generated synthetic features into the radiographic image such as a chest X-ray at the location defined by the feature mask. The resulting synthetic image may be annotated with metadata encoding the feature characteristics and subtlety control value used for generation. This metadata enables effective use of the image for training and validation of machine learning models. In some embodiments, the machine-readable instructions may also include steps for dynamically composing multiple adapters during inference to synthesize composite features with blended characteristics and subtlety levels, thereby enriching the diversity of generated datasets. Such functionality supports AI development pipelines in domains such as lung cancer detection, classification of radiographic findings, or medical image-based risk stratification.
[0122] Various embodiments of the disclosure encompass numerous advantages, including the system and the method for controlled generation of synthetic features in an image. The disclosed system and the method have several technical advantages, but not limited to the following:
[0123] Improved data diversity: Enables the creation of large, balanced datasets containing a wide spectrum of nodule types, including rare and subtle nodules, which are often underrepresented in clinical datasets.
[0124] Fine-grained control: Allows targeted generation of synthetic nodules by controlling attributes such as calcification, shape, size, texture, homogeneity, and border irregularity.
[0125] Subtlety modulation: Incorporates a novel subtlety control mechanism to simulate nodules ranging from highly visible to extremely subtle, enhancing the robustness of detection models.
[0126] Plug-and-play adaptability: Utilizes modular design (e.g., LoRA-based plug-and-play components) to dynamically adapt to different clinical use cases or training needs.
[0127] Improved AI performance: Facilitates training of more accurate and generalizable AI models for lung cancer detection by augmenting real-world datasets with synthetic examples.
[0128] Reduced annotation burden: Generates pre-annotated synthetic data, thereby reducing the need for manual radiologist annotations.
[0129] Enables benchmarking and explainability: Supports rigorous evaluation and interpretation of AI models by generating ground-truth-known synthetic cases for testing explainability.
[0130] In summary, the technical problem addressed by the disclosed invention relates to the limited availability and variability of annotated medical imaging data, particularly chest radiographs (CXR), for training AI-based lung disease detection systems. Current approaches rely on manual annotation or handcrafted augmentation, which are time-consuming, labor intensive, suffer from intra-observer variability, prone to inconsistency, and unable to replicate subtle pathological features. Existing synthetic generation techniques using GANs or traditional diffusion models lack precise control over key attributes such as calcification, border definition, or subtlety critical for clinically realistic data generation. Moreover, radiographic images inherently suffer from low contrast and structural overlap, making it challenging to detect early-stage pulmonary nodules, especially those with subtle visual signatures.
[0131] The disclosed invention solves these challenges by providing a method and system for controlled generation of synthetic nodules using a generative diffusion model conditioned on input parameters including a feature mask, feature characteristics, and a subtlety control value. The invention introduces adapter-based conditioning, such as Low-Rank Adaptation (LoRA), and employs a perceptual subtlety estimation technique using Optimal Transport metrics to guide the visibility of the generated nodules. This framework enables precise, parameter-efficient customization of synthetic features and enhances the realism of training data for clinical AI models. As a result, the system improves the reliability of early-stage disease detection while reducing dependence on manually curated datasets.
[0132] The claimed invention comprises a concrete implementation involving tangible hardware and software modules that work in unison to achieve technical outcomes. These include a processor configured to execute a trained diffusion model, memory to store adapter modules and masks, and a module management system for dynamically selecting and composing plug-and-play conditioning modules. The system supports scalable, automated generation of annotated synthetic regions, enabling the enrichment of datasets used for developing, validating, and personalizing diagnostic algorithms. The invention thus offers a non-trivial technical advancement in medical image synthesis, contributing to higher AI model accuracy and robustness across a wide range of radiographic applications.
[0133] In light of the above-mentioned advantages and the technical advancements provided by the disclosed method and system, the claimed steps as discussed above are not routine, conventional, or well understood in the art, as the claimed steps enable the following solutions to the existing problems in conventional technologies. Further, the claimed steps clearly bring an improvement in the functioning of the device itself as the claimed steps provide a technical solution to a technical problem.
[0134] The present disclosure may be realized in hardware, or a combination of hardware and software. The present disclosure may be realized in a centralized fashion, in at least one computer system, or in a distributed fashion, where different elements may be spread across several interconnected computer systems. A computer system or other apparatus adapted for carrying out the methods described herein may be suited. A combination of hardware and software may be a general-purpose computer system with a computer program that, when loaded and executed, may control the computer system such that the computer system carries out the methods described herein. The present disclosure may be realized in hardware that includes a portion of an integrated circuit that also performs other functions.
[0135] A person with ordinary skills in the art will appreciate that the systems, modules, and sub-modules have been illustrated and explained to serve as examples and should not be considered limiting in any manner. It will be further appreciated that the variants of the above disclosed system elements, modules, and other features and functions, or alternatives thereof, may be combined to create other different systems or applications.
[0136] Those skilled in the art will appreciate that any of the aforementioned steps and/or system modules may be suitably replaced, reordered, or removed, and additional steps and/or system modules may be inserted, depending on the needs of a particular application. In addition, the systems of the aforementioned embodiments may be implemented using a wide variety of suitable processes and system modules, and are not limited to any particular computer hardware, software, middleware, firmware, microcode, and the like. The claims can encompass embodiments for hardware and software, or a combination thereof.
[0137] While the present disclosure has been described with reference to certain embodiments, it will be understood by those skilled in the art that various changes may be made, and equivalents may be substituted without departing from the scope of the present disclosure. In addition, many modifications may be made to adapt a particular situation or material to the teachings of the present disclosure without departing from its scope. Therefore, it is intended that the present disclosure is not limited to the particular embodiment disclosed, but that the present disclosure will include all embodiments falling within the scope of the appended claims.
, C , Claims:WE CLAIM:
1. A method (300) for controlled generation of synthetic features in an image, wherein the method (300) comprises:
receiving (301), via a processor (201), one or more input parameters, wherein the one or more parameters comprise a feature mask, one or more feature characteristics, a subtlety control value;
retrieving (302), via the processor (201), one or more adapters based on the one or more input parameters;
generating (303), via the processor (201), the controlled synthetic features by providing the feature mask and the one or more adapters to a generative diffusion model; and
providing (304), via the processor (201), the controlled synthetic features on the image.
2. The method (300) as claimed in claim 1, wherein the feature mask indicates at least one of location, size, shape, or a combination thereof, of the synthetic features;
wherein the one or more feature characteristics comprise at least one of calcification, border definition, texture, or a combination thereof;
wherein the subtlety control value indicates a perceptibility level of the synthetic features.
3. The method (300) as claimed in claim 1, wherein the synthetic features correspond to one or more synthetic lung nodules;
wherein the feature mask corresponds to a nodule mask corresponding to each of the one or more synthetic lung nodules.
4. The method (300) as claimed in claim 1, wherein the one or more adapters are retrieved based on at least one of the one or more feature characteristics, the subtlety control value, or a combination thereof;
wherein the one or more adapters corresponds to at least one of a Low-Rank Adaptation (LoRA) modules, an adapter module, a prompt-tuning module, a hypernetwork module, a prefix-tuning module, or a combination thereof.
5. The method (300) as claimed in claim 2, wherein the controlled synthetic features are placed on the location in the image with size and shape defined in the feature mask;
wherein the controlled synthetic feature comprises the one or more characteristics and subtlety based on the received one or more input parameters.
6. The method (300) as claimed in claim 1, wherein the one or more adapters comprise at least one of one or more feature adapters, one or more subtlety adapters or a combination thereof;
wherein each of the one or more feature adapters are trained on a corresponding feature characteristic from the one or more feature characteristics;
wherein each of the one or more subtlety adapters are trained on at least one of the one or more feature characteristics, one or more subtlety estimation metrics, or a combination thereof.
7. The method (300) as claimed in claim 6, wherein the generative diffusion model is trained for image generation based on receiving at least one of one or more sample images, one or more feature masks, one or more adapter weights, or a combination thereof, as input.
8. The method (300) as claimed in claim 6, wherein the one or more subtlety estimation metrics are computed based on intensity variations and neighborhood pattern features by utilizing one or more machine learning techniques;
wherein the one or more machine learning techniques comprise at least one of Optimal Transport (OT), Maximum Mean Discrepancy (MMD), Kullback–Leibler Divergence (KL Divergence), Jensen–Shannon Divergence (JSD), Earth Mover’s Distance (EMD) technique, Energy Distance, Chamfer distance, Cramer Distance, f-Divergences, Sinkhorn Distance, Entropic OT, or Wasserstein Distance;
wherein the subtlety control value corresponds to one of the one or more subtlety estimation metrics.
9. The method (300) as claimed in claim 1, wherein the image corresponds to radiographic image; wherein the radiograph image comprises chest radiograph image.
10. The method (300) as claimed in claim 1, wherein providing the controlled synthetic features on the image corresponds to annotating the image with metadata corresponding to the one or more feature characteristics and the subtlety control value based on the received one or more input parameters.
11. A system (100) for controlled generation of synthetic features in an image, the system (100) comprises:
a processor (201);
a memory (202) communicatively coupled with the processor (201), wherein the memory (202) is configured to store one or more executable instructions that, when executed by the processor (201), cause the processor (201) to:
receive (301) one or more input parameters, wherein the one or more parameters comprise a feature mask, one or more feature characteristics, a subtlety control value;
retrieve (302) one or more adapters based on the one or more input parameters;
generate (303) the controlled synthetic features by providing the feature mask and the one or more adapters to a generative diffusion model; and
provide (304) the controlled synthetic features on the image.
12. A non-transitory computer-readable storage medium having stored thereon, a set of computer-executable instructions that, when executed by a processor (201), cause the processor (201) to perform steps comprising:
receiving (301) one or more input parameters, wherein the one or more parameters comprise a feature mask, one or more feature characteristics, a subtlety control value;
retrieving (302) one or more adapters based on the one or more input parameters;
generating (303) the controlled synthetic features by providing the feature mask and the one or more adapters to a generative diffusion model ; and
providing (304) the controlled synthetic features on the image.
Dated this 04th Day of August 2025

ABHIJEET GIDDE
IN/PA-4407
AGENT FOR THE APPLICANT