Description:FIELD OF THE INVENTION
The present invention relates to the field of medical imaging and artificial intelligence (AI). More particularly, it concerns a system and method for automated detection and segmentation of chronic kidney disease (CKD) from MRI images using a hybrid convolutional neural network (CNN) and U-Net architecture. The invention enables precise and explainable identification of kidney abnormalities by integrating detection and segmentation in a single unified framework.
BACKGROUND OF THE INVENTION
With the rising prevalence of Chronic Kidney Disease (CKD), accurate and early detection through medical imaging has become essential for effective treatment and management. However, current diagnostic systems face significant challenges in terms of precision, computational efficiency, and contextual interpretation of MRI images. Conventional image processing and rule-based detection techniques often fail to recognize subtle or early-stage kidney abnormalities due to their limited semantic understanding of complex medical patterns. Additionally, many existing models are either designed for classification or segmentation alone, lacking a unified approach that effectively handles both tasks. These systems often suffer from domain dependency and do not generalize well across diverse MRI datasets or imaging conditions. Furthermore, the complexity and opacity of current deep learning models hinder clinical trust and interpretability. These limitations highlight the pressing need for an integrated, intelligent framework that combines the power of deep learning architectures such as CNN and U-Net for precise, scalable, and explainable CKD detection and segmentation.
US20180372717: A medical diagnostic system is provided to automate analysis of samples to predict a medical condition, such as pregnancy or chronic kidney disease. The system may provide test strip usage automation. The medical diagnostic system may include a sample collection component, collection cup contamination protection mechanism, sample volume control component, test strip reader component, which may be manifested as a lateral flow strip reader, flow reader, sample analytic component, data processing component, data communication component, networked data management component, and device cleaning mechanism. A method to automate analysis of samples to predict a medical condition using the medical diagnostic system is also provided.
US20110183434: The present invention describes the ability to identify chronic kidney disease (CKD) mortality risk in asymptomatic patients. For example, a patient having a normal glomerular filtration rate would be considered likely to have an increased mortality risk for chronic kidney disease upon the detection of an FGF-23 amino acid sequence that is above a normal level, but below CKD Stage 1 levels. Consequently, therapeutic strategies may be implements to prevent morbidity and mortality following chronic kidney disease progression. Such therapeutic strategies can involve phosphate reduction strategies (i.e., for example, reduced dietary intake of phosphorus and/or administration of phosphate binding compound). Further, kits are described providing instruction to determine a specific mortality risk based upon measured FGF-23 levels and estimated glomerular filtration rates.
Chronic Kidney Disease (CKD) is a growing global health challenge requiring accurate and early diagnosis. Conventional MRI-based diagnostic methods depend heavily on manual image interpretation, which is time-consuming, error-prone, and inconsistent across clinicians. Existing automated tools are often limited to either classification or segmentation, without providing a unified framework capable of handling both. Moreover, standalone CNN or UNet models struggle with subtle early-stage abnormalities, domain dependency across datasets, and lack interpretability. Current systems thus fail to provide high-precision, scalable, and explainable solutions.
The present invention addresses these shortcomings by employing a hybrid CNN-UNet architecture that combines robust spatial feature extraction with fine-grained pixel-level segmentation. This dual capability ensures accurate detection and delineation of CKD regions in MRI scans, improves generalization across diverse imaging datasets, and provides real-time diagnostic support with improved trustworthiness for clinical adoption.
SUMMARY OF THE INVENTION
This summary is provided to introduce a selection of concepts, in a simplified format, that are further described in the detailed description of the invention.
This summary is neither intended to identify key or essential inventive concepts of the invention and nor is it intended for determining the scope of the invention.
The invention provides a hybrid AI-based diagnostic framework for the automated detection and segmentation of CKD from MRI images. The system integrates two deep learning components: a convolutional neural network (CNN) for extracting discriminative spatial features and classifying abnormalities, and a U-Net module for precise pixel-level segmentation of affected regions.
MRI images are preprocessed through normalization, contrast enhancement, and noise reduction before being fed into the hybrid network. The CNN module identifies potential abnormal kidney structures, while the U-Net architecture segments the diseased areas with high resolution. The integration of both modules enables simultaneous classification and localization, enhancing diagnostic accuracy compared to standalone methods.
The invention further incorporates a modular design, allowing scalability across different MRI modalities and adaptability to various clinical environments. Its real-time capability ensures efficient diagnostic workflows, while embedded explainability modules provide interpretable outputs to build clinical trust.
By unifying detection and segmentation in a single hybrid architecture, the invention significantly reduces diagnostic time, increases accuracy, and enhances reproducibility across healthcare facilities, providing a reliable AI-driven solution for CKD management.
To further clarify advantages and features of the present invention, a more particular description of the invention will be rendered by reference to specific embodiments thereof, which is illustrated in the appended drawings. It is appreciated that these drawings depict only typical embodiments of the invention and are therefore not to be considered limiting of its scope. The invention will be described and explained with additional specificity and detail with the accompanying drawings.
The present invention proposes an automated diagnostic system that enhances the accuracy and efficiency of Chronic Kidney Disease (CKD) detection through advanced deep learning techniques, particularly using a hybrid CNN-UNet architecture. This method surpasses traditional rule-based and manual MRI analysis systems by integrating both detection and segmentation tasks within a unified framework. The CNN component is responsible for extracting rich spatial features from MRI images, enabling precise localization of abnormal kidney structures. Simultaneously, the UNet module provides detailed pixel-level segmentation to map affected regions with high resolution.
BRIEF DESCRIPTION OF THE DRAWINGS
The illustrated embodiments of the subject matter will be understood by reference to the drawings, wherein like parts are designated by like numerals throughout. The following description is intended only by way of example, and simply illustrates certain selected embodiments of devices, systems, and methods that are consistent with the subject matter as claimed herein, wherein:
FIGURE 1: SYSTEM ARCHITECTURE
The figures depict embodiments of the present subject matter for the purposes of illustration only. A person skilled in the art will easily recognize from the following description that alternative embodiments of the structures and methods illustrated herein may be employed without departing from the principles of the disclosure described herein.
DETAILED DESCRIPTION OF THE INVENTION
The detailed description of various exemplary embodiments of the disclosure is described herein with reference to the accompanying drawings. It should be noted that the embodiments are described herein in such details as to clearly communicate the disclosure. However, the amount of details provided herein is not intended to limit the anticipated variations of embodiments; on the contrary, the intention is to cover all modifications, equivalents, and alternatives falling within the scope of the present disclosure as defined by the appended claims.
It is also to be understood that various arrangements may be devised that, although not explicitly described or shown herein, embody the principles of the present disclosure. Moreover, all statements herein reciting principles, aspects, and embodiments of the present disclosure, as well as specific examples, are intended to encompass equivalents thereof.
The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments. As used herein, the singular forms “a",” “an” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms “comprises,” “comprising,” “includes” and/or “including,” when used herein, specify the presence of stated features, integers, steps, operations, elements and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and/or groups thereof.
It should also be noted that in some alternative implementations, the functions/acts noted may occur out of the order noted in the figures. For example, two figures shown in succession may, in fact, be executed concurrently or may sometimes be executed in the reverse order, depending upon the functionality/acts involved.
In addition, the descriptions of "first", "second", “third”, and the like in the present invention are used for the purpose of description only, and are not to be construed as indicating or implying their relative importance or implicitly indicating the number of technical features indicated. Thus, features defining "first" and "second" may include at least one of the features, either explicitly or implicitly.
Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which example embodiments belong. It will be further understood that terms, e.g., those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
The present invention proposes an automated diagnostic system that enhances the accuracy and efficiency of Chronic Kidney Disease (CKD) detection through advanced deep learning techniques, particularly using a hybrid CNN-UNet architecture. This method surpasses traditional rule-based and manual MRI analysis systems by integrating both detection and segmentation tasks within a unified framework. The CNN component is responsible for extracting rich spatial features from MRI images, enabling precise localization of abnormal kidney structures. Simultaneously, the UNet module provides detailed pixel-level segmentation to map affected regions with high resolution.
The hybrid structure allows the system to generalize across different MRI datasets, overcoming limitations related to domain dependency. The model is trained on annotated datasets of kidney MRIs and uses advanced preprocessing techniques such as normalization, noise filtering, and contrast enhancement to improve input quality.
Moreover, the system supports real-time processing and is scalable for diverse clinical environments. Its modularity ensures adaptability across different MRI modalities, providing robust, explainable, and automated CKD diagnostic support for medical professionals.
The proposed system achieves superior performance over existing CKD detection methods by integrating a hybrid CNN-UNet architecture that simultaneously performs both detection and segmentation of kidney abnormalities from MRI images, enabling domain-independent, high-precision, and fully automated medical image analysis.
The present invention discloses a hybrid artificial intelligence-based framework designed for the automated detection and segmentation of Chronic Kidney Disease (CKD) using MRI images. The system integrates convolutional neural networks (CNNs) with a U-Net architecture to deliver both classification and segmentation within a unified diagnostic process. By combining these two architectures, the invention leverages the strengths of CNNs for extracting spatial features and the capabilities of U-Net for fine-grained pixel-level mapping.
The system begins with the intake of MRI images from different modalities such as T1- and T2-weighted scans. These images undergo preprocessing steps that include normalization, denoising, and contrast enhancement. Such preprocessing ensures consistency in the input data and improves the ability of the model to identify subtle abnormalities in the kidney regions. The preprocessed data is first analyzed by the CNN module, which extracts discriminative spatial features and highlights regions suspected of containing CKD-related abnormalities.
The features generated by the CNN are then passed into the U-Net module. U-Net employs an encoder–decoder structure, where the encoder condenses image information into feature maps and the decoder reconstructs these maps into segmented images. Skip connections between the encoder and decoder stages allow the system to preserve fine image details, ensuring precise segmentation boundaries for kidney tissues. This enables the model to not only detect the presence of disease but also provide a detailed outline of the affected regions.
The hybrid framework is trained using annotated MRI datasets containing both normal and diseased kidney images. Supervised learning allows the system to associate image features with ground-truth labels, improving accuracy across both detection and segmentation tasks. Data augmentation methods, such as image rotation, scaling, and flipping, are employed to improve robustness and ensure the model generalizes across diverse MRI datasets. This capability addresses one of the major shortcomings of existing systems, namely their dependency on dataset-specific features.
Once the segmentation is complete, the system applies post-processing operations. Morphological filters are used to refine segmentation boundaries, reduce noise, and improve clarity of the detected regions. The final outputs include classification labels indicating the likelihood of CKD presence and segmented images that highlight the diseased areas. These outputs can be delivered through a clinical interface for radiologist review.
A notable feature of the invention is its capacity for real-time inference, which makes it suitable for direct integration into clinical workflows. Hospitals can integrate the system with their existing PACS (Picture Archiving and Communication System) to automate the analysis of kidney MRI scans, thereby reducing diagnostic delays. Furthermore, the modular design of the invention allows scalability, enabling the same architecture to be adapted for other medical imaging applications beyond CKD, such as liver, lung, or cardiac disease analysis.

The invention also incorporates explainable AI mechanisms to improve clinical trust. The system highlights areas of the MRI scans that influenced its decision-making, providing interpretability to healthcare professionals. This transparency ensures that the model does not function as a “black box” but rather as a tool that enhances human decision-making with justifiable results.
By integrating CNN-based detection with U-Net-based segmentation, the invention creates a powerful hybrid solution that achieves both classification accuracy and segmentation precision. It provides an end-to-end automated workflow capable of detecting, localizing, and explaining kidney abnormalities in MRI images. The system is domain-independent, highly accurate, scalable, and interpretable, making it a significant advancement in medical imaging diagnostics.
Best Method of Working
The best method of working involves training the hybrid CNN-UNet model on a large annotated dataset of kidney MRI scans. Preprocessing techniques standardize input images, ensuring consistent feature extraction. The CNN module detects abnormalities by learning spatial representations of diseased kidneys. Segmentation is then performed by the UNet module, which delineates diseased regions with high resolution. The integration of classification and segmentation ensures both detection and localization in a single process. The system should be deployed with GPU acceleration for real-time inference, and can be integrated with hospital imaging systems for automated workflow support. This embodiment represents the most efficient and clinically practical implementation of the invention.
 
, Claims:1. A system for automated detection and segmentation of Chronic Kidney Disease (CKD) from MRI images, comprising:
an input module for receiving MRI image data;
a preprocessing unit configured for normalization, denoising, and contrast enhancement;
a convolutional neural network (CNN) for extracting spatial features and detecting abnormal kidney structures;
a UNet module for performing pixel-level segmentation of kidney regions;
a hybrid integration unit combining outputs of the CNN and UNet modules;
a post-processing unit for refining segmentation outputs; and
an output interface for providing classified results and segmented images.
2. The system as claimed in claim 1, wherein the CNN is configured to generate spatial feature maps capturing abnormal kidney structures.
3. The system as claimed in claim 1, wherein the UNet comprises an encoder–decoder architecture with skip connections for preserving fine-grained image details.
4. The system as claimed in claim 1, wherein the preprocessing unit applies normalization, noise reduction, and contrast enhancement to improve MRI input quality.
5. The system as claimed in claim 1, wherein the hybrid integration unit enables simultaneous detection and segmentation of CKD in MRI scans.
6. The system as claimed in claim 1, wherein the post-processing unit applies morphological filtering for refining segmented kidney boundaries.
7. The system as claimed in claim 1, wherein the system supports real-time inference and integration with hospital PACS systems.
8. The system as claimed in claim 1, wherein explainability modules highlight regions influencing classification decisions for clinical interpretability.
9. The system as claimed in claim 1, wherein the system generalizes across multiple MRI datasets using data augmentation techniques.
10. A method for automated detection and segmentation of CKD from MRI images, comprising the steps of:
receiving MRI image data;
preprocessing the MRI data with normalization, noise filtering, and contrast enhancement;
extracting spatial features using a CNN module;
segmenting kidney regions using a UNet architecture;
integrating CNN-based detection with UNet-based segmentation in a hybrid framework;
refining segmented outputs using post-processing filters; and
providing classification and segmentation results through an output interface.