Description:FORM 2
THE PATENTS ACT, 1970
(39 of 1970)
&
THE PATENT RULES, 2003
Complete Specification
(See section10 and rule13)

1. Title of the Invention: LEAST-CORRELATED WEIGHT FACTOR–DRIVEN PIXEL EDGE SEGMENTATION FOR MRI-BASED LUNG TUMOR IDENTIFICATION
2.Applicants: -
SR University India Ananthasagar, Hasanparthy (PO), Warangal-506371, Telangana, India.
Inventors:-
Name Nationality Address
Aliya Thaseen Indian School of Computer Science & Artificial Intelligence, SR University, Ananthasagar, Hasanparthy (PO), Warangal-506371, Telangana, India.
Dr. Durgesh Nandan Indian School of Computer Science & Artificial Intelligence, SR University, Ananthasagar, Hasanparthy (PO), Warangal-506371, Telangana, India.
Dr. Sheshikala Martha Indian School of Computer Science & Artificial Intelligence, SR University, Ananthasagar, Hasanparthy (PO), Warangal-506371, Telangana, India.
3. Preamble to the description:
The following specification particularly describes the invention and the manner in which it is to be performed.

4. DESCRIPTION
FIELD OF THE INVENTION
The field of current invention is related to the technical sphere of medical images processing and machine learning and more generally to a segmentation and feature extraction approach in lung tumor detection. It is interested in the application of Magnetic Resonance Imaging (MRI)-based edge segmentation methods in performing early and accurate diagnosis on cancer. The invention of Linked Pixel Edge Segmentation with Least Correlated Weight Factor or LPES-LCWF model suggests better tumor detection. It falls under the purveillance of computing- assisted strategies of sickness diagnosis in medical technologies. BACKGROUND OF THE INVENTION
The lung cancer has remained in the list of prevalent cancer death in all sections of the world that has recorded a significant rate of losing patients to cancer in one year. Judging by the trends in the rest of the global cancer statistics; millions of new cases have been reported annually and cancer of the lung is one of the most prevalent causes of death. One of the reasons why its total number of deaths is large is its insensitivity to early diagnosis. In order to diagnose nodules, radiologists use various medical imaging examinations that comprise X-rays, CT scan, and MRI. However, segmentation and processing of medical images in a manual manner is highly labour intensive, time consuming and prone to human error. The latter is such an issue that it is not possible to distinguish between benign and malignant nodules by standard methods, in particular when the tumor margins are not straight, but rather cross and/or do not exist (at all).
It is a case that generally leads to the false positive, a treatment course that is also delayed, thus an additional burden to the medical systems. Through the implementation of the computer-aided diagnostic (CAD) systems, computerized medical imaging has developed as a growth revolution since it introduces the automated process of segmentation and identification of them. The deep learning models and image processing algorithms properties promise a lot by increasing the precision of segmentation. However, as these advancements take place, the existing processes remain susceptible to errors, such as, over-segmentation, computational inefficiency and lack of extrapolation to different patients. In addition, dissimilades in medical resonance imaging (MRI)-images, image noise and artifiers also make these models unreliable. In particular, it is significant that segmentation accuracy is high, which directly influences the recognition of relevant features that a tumor is called based on them.
Accepted segmentation hinting, such as thresholding, region-based and privatizing to detect not gracious of the edges, or partitionamps redundant features correlation. This brings ineffectiveness of classification and feature extraction. A pulling desire, therefore, exists towards a solid solution that is not just capable of multiplying the sums of the segmentation accuracy but also the volume of the computational efficiency as well as a decreased false positive. To deal with the limitations given above the current invention does include a Linked Pixel Edge Segregation with Least Correlated Weight Factor (LPES-LCWF) approach. In this system, feature roll and emphasis on feature roll is one of the general features being weighted to spin with correlation.
Since the invention can optimize the ratio of least-correlated features, it is sure to eliminate redundancy, to maximize the vectors of features and in fact the invention is on schedule to detect tumor boundaries to the dot. Unlike classical models, the concept LPES-LCWF attends to machine learning in the recognition of the patterns, and this aspect gives the computer system powers to develop and even become eventually. This, in its turn, renders it appropriate to a clinical real-life implementation, and it can be duly replicated after the letter in other imaging protocols, and even in patient demographics. Hence, invention is viable competent and most precise verse in the life time scenario of spotting of lung tumor by MRI scan and thus allows radiologists to have an improved and creative direction of their diagnosis, the extensional awareness to diagnose and additionally info on decision and care taking.
SUMMARY OF THE INVENTION
The given invention is therefore providing a novel approach of identifying tumors in the lungs on the one hand based on foundations of MRI based Linked Pixel Edge Segmentation with Least Correlated Weight Factor (LPES-LCWF). In contrast to the classical methods of creating segmentation models, the invention is based on the contribution of the use of pixel-linking model to focus on continuous tracing of the edge and accurate delineation of the tumor. It addresses the irregular and ill-defined tumor boundaries that are typically hard to deal with in the existing systems through edge connectivity, which makes it very clear. One of the major aspects of this invention is its least correlated weight factor mechanism that gives higher weight to features with minimum degree of correlation, in the quest to excel over feature redundancy and selection.
This provides a very exact feature set which will increase the accuracy of the classification of tumors. The invention is founded on integrating the segmentation, edge detecting and correlation-based feature weighting in a machine learning model under study of which enable an adaptive learning and better recognition of patterns with time. It possesses an calculation efficiency and highly achieves the ability to consider numerous varied qualities and protocols of MRI images to provide robustness among various demographics of the patients. A comparative analysis with the traditional and the deep learning-based models proves that LPES-LCWF can be brought to the front due to its better accuracy in segmentation; it needs a shorter time to extract features and has less false positive percentage. The invention is providing major advancements in the clinical practices, by allowing radiologists to detect the tumor earlier, provide better measurement of tumor areas and provide better treatment plan. Its potential to detect both benign and malignant tumours, at high accuracies, is one such factor that plays into the decreased diagnostic uncertainty and better patient outcome. Thus, LPES-LCWF model is a novel approach to the field of medical images, and it can get a strong, effective and intelligent sample to identify and classify lung tumor.
OBJECTIVE OF INVENTION
• To design effective and computationally efficient MRI image segmentation model, to early neurological tumor detection by the Linked Pixel Edge Segmentation technique.
• To maximize the feature extraction and the estimation by slowest weight factor assigning in the classification to minimize redundancies and enhancement of diagnostic accuracy.
BRIEF DESCRIPTION OF THE DRAWINGS
Fig.1: Depicts Flow diagram for the Proposed Invention.

Fig.2: Depicts Synergy in Lung Tumor Detection.

Fig.3: Depicts Cycle of LPES-LCWF Model.

BRIEF DESCRIPTION OF THE INVENTION
The invention described here is of the medical image analysis and computer aided diagnosis (CAD) system, relating to the field of lung tumors detection application through an MRI image. Lung cancer is the most common and among the most deadly cancer in the world and claims millions of lives every year. Although the sphere of medicine has been enhanced, inefficiencies of timely disease detection remains the major aspect contributing to the decrease in the mortality rate. In searching abnormalities, radiologists typically adopt the search technology namely X-rays, Computer Tomography (CT) and Magnetic Resonance Imaging (MRI).
But, manual procedures of analysing MRI images are subject to human interpretation and can be extremely time consuming and subject to variation in factors like expertise. The more typical segmentation and categorization methods though useful, lack accuracy, efficiency and adaptability. Such strategies are inclined not to delimitate tumor margins clearly, in particular situations involving irregular tumefactions, noisoy images or overlapping anatomic structures. Besides, the additional qualities gained in images are also extraneous, leading to an increase in false positives and the doubt in diagnosis. To address them, a different segmentation and feature extraction model has been developed that is referred to as the current invention as Linked Pixel Edge Segmentation with Least Correlated Weight Factor (LPES-LCWF).]
LPES-LCWF model combines the segmentation on pixel level, the ordeal of edge detecting and edge weighing basing on the concurrence within a model. Compared to existing deep learning frameworks (which use rich data), the proposed invention demonstrates immense efficiency on working with MRI scans that have a reduced data representation, thus making it both pragmatic and large scale to use in hospitals. This system attempts to improve the level of accuracy of segmentation, uniqueness, computationally efficiently, and further aspires to provide improved decision support to the radiologist. The invention thus takes a big step as far as the automation of tumor diagnosis in the lung is concerned.
The principle of operation of the invention of LPES-LCWF, works through the staged structure; first by inducing image of MRI of the lungs with magnetization and then stepping through registration of pixel extraction, node detection, feature classification and weighting. All the steps are prior thought, so as to transformation the diagnosis level impeccably and reducing the inefficient space of computing.

This starts with a process of MRI Image Acquisition where images of the lungs are captured on the patients to be analyzed. In comparison to CT scans, MRI image data, MRI does not involve uses harmful radiation in order to give the detailed soft tissue contrast therefore making MRI a best use of image of repeated screening. These MRI images are fed into the system and the system implements several preprocessing methods to enhance image noise reduction along with image preparation in terms of making the image ready to undergo segmentation.
During the segmentation step the invention is based on an exclusive Linked Pixel Edge Segmentation (LPES) method. In this method, the adjacent pixels are interrelated with similarity values, e.g., intensity, texture etc. This can be used to maintain continuity of tumor boundaries when needed (considering similarly tumors are irregular, or poorly-defined, etc.) instead of processing each pixel separately. This procedure removes errors of segmentation that occurs in the process of over-segmentation and under-segmentation that occurs in the traditional processes. After segmentation followed by model, first relevant each segment of the images are extracted. The features one generates like Intensity value, The texture patterns and shape descriptors on the pixels extracted. At this point in the system it holds onto them regions of interest (ROI) thereby keeping the 1s un required calculation a minimum.
Stage of Edge Development One part of the invention phase is very critical, known as stage of edge development. The model, by identifying a sudden change in pixels intensity, can detect the edges of the possible tumour regions with specific precision. As compared to the conventional edge detector that may not be able to detect the tiny border, the improved edge continuity offered by LPES by utilization of linked pixel technique, which guarantees the tracing of correct tumor edge. It is particularly important when those tumours are located near other tissues of similar density such as blood vessels or breastwalls.
Once the appropriate features are obtained the system refers to Least Correlated Weight Factor (LCWF) system. Here we compare features with each other to come up with some result in terms of correlation. A feature that's found to have a strong correlation with another is likely to be redundant as far as that factor goes and offers little in the way of specific information to a diagnosis. Fear of many features with many-to-many relation is therefore assigned lesser priority in order to have this feature in this model thus will be given high weight to the respective features. This is used to ensure that only the most important and special unique characters are used in the process of classification such that the possibilities of false positive is reduced and diagnostic process more credible.
Finally, some machine learning classification algorithm is applied for processing the selected and filtered set of features. With the assistance of the classifier, the fine feature vectors of former and viable nodules are utilized for making the classification. Instances are introduced through the model and the model develops and increases with amount of time since many patients go through the model a few times. We discover that this flexibility element is too well adaptable considering toward the varieties of demographics of patients as well as imaging regimens.
The near exclusive outcome will be the fact that, it will have high level of segmentation accuracy at lower level of redundancy of features, faster computational capability and improved classification efficiencies and will, therefore, become an irreplaceable aspect of clinical use.
MECHANISMS WHICH HAVE BEEN ACQUIRED OR BEEN ADDED TO THE INVENTION
The use of this LPES-LCWF invention has numerous special advantages over the traditional methods of detecting lung tumor. These advantages can be described into the technical benefits, clinical and operational benefits.
A. Technical Advantages
For this, the biggest technical innovation is the so-called Linked Pixel Edge Segmentation method. Contrary to threshold specific or clustering-based algorithms where inter-pixel relation is not of significant concern, in LPES, inter-pixel relation is directly addressed in such a way that it can also efficiently process arbitrary tumor boundaries and achieve high levels of accuracy. This is the new development which ensures that the identification of it tumors could being suspended on the border of its flags without the inconvenience which was present of the history of models in the situation of Irregular or spiculated nodes.
B. Clinical Advantages
The invention, AARP promises to diagnose lung cancer as early and good detection, and first it will save lives sick patients with the presence of cancer. This means that early tumors that are hard to spot by the conventional method (as it is subtle) become easy to spot with LPES-LCWF. The invention will assist the oncologist plan targeted therapy and surgery, radiation is helping cure How the cancer involving measurement of the frontiers, depending on a particular approach.
Additionally, the system is vulnerable to huge amount of diversity on the MRI protocols and different kinds of patients, and strong in practical use. It is also decreasing false positive and it is decreasing the number of unnecessary biopsies and consequences of treatments, which is decreasing the level of anxiety of patient and cooling healthcare expenses.
C. Operational Advantages
The computational efficiency is high as far operation efficiency of the invention is concerned. The older model of deep learning is usually resource-intensive both in the amount of data and computing power and cannot be used to smaller sized hospitals. LPES-LCWF on the Office, in its turn, lack of resources but yet a better outcome, so that it may be implemented to an equally diverse range of healthcare institutions.
Its design also encourages the benevolence of combining it to the existing hospital imaging tools and CAD and can also be easily adapted without entailing gigantically large hardware upgrade. This makes the invention not technologically, but also very practically possible regarding its use in the extensive clinical practice.
D. He or she must invent something new rather than compare it with anything that has been implemented by other systems
Compared to UNet, UNet++, CNN-transformer hybrid, or fuzzy attention networks, LPES-LCWF is the only one that uses pixel connectivity and correlation weighting pixel connectivity association thirst fuzzy attention network. The proficient two concentration operates to avail greater accuracy and cluster optimization of characteristics and flexibility in segmentation than the current approaches that have allowed such a model to emerge as a novel and culture changeover development in the medical imaging technology.
HAS INDUSTRIAL & CLINICAL APPLICATIONS
The LPES-LCWF invention application location is also incredibly wide in terms of its application in the area of healthcare, diagnostics and research.
• Finding to Finding Lung Cancer
The most frequent application of it is the primary diagnosis of lung cancer. Such an invention ensures that the radiologists could receive the much needed diagnostic aids by classifying the MRI pictures into right categories and then classify the image into the benign and the malignant. This increases the level of confidence in the clinical decision making and decreases the level of diagnostic ambiguity. This has the direct positive effect on patient outcome as they will be in a position to start their treatment early enough.
• Planning and Re-evaluation of Treatment.
Adequate delineating of the limits of the tumor and feature separation enables the oncologist to find out the exact size of the tumor, tumor shape and tumor progress. This information is then used in surgery, chemotherapy and radiation therapy planning served with great importance. It is also possible that the model could be used to track the tumor response to therapy as time evolves and lend objective and quantitative data of the effectiveness of the treatment response.
• Satisfactory acquisition of Hospital Models
It was to be integrated in the already existing Picture Archiving and communication system (PACS) and other hospital imaginations processes. And it's a second-opinion device; essentially it makes the radiologists do less work and improves the quality of the diagnosis. Its computing power allows it to extend the benefit of high level diagnostic capability to the less endowed hospitals as well.
• Medical Data analysis & Research
In addition to the clinical component, LPES-LCWF may be utilized for research practices within medical communities, either to examine tumor patterns of development, radiomics, and loop predictive patient scenario the survival model. The various sets of MRI data types that it can handle allows its use to be extended to epidemiological studies in the large scale as well as in clinical trials.

, Claims:We Claim:
1. A method of identifying the lung tumor based on an MRI image on the basis of Linked Pixel Edge Segmentation by Least Correlated Weight Factor (LPES-LCWF)
2. This invention has been taking advantage of reality that an edge can be recognized in a pixel-to-pixel basis in order to achieve continual boundary recognition of an lung tumor.
3. The least correlated weight factor process is embraced that provides services with the best weights to extracted features to reduce redundancy.
4. It is a hybridized system comprising of of segmentation, edge-detection and feature-weighting embedded in a single diagnostic system.
5. The model may lead to the higher diagnostic accuracy with lower percentage of the false-positive and higher percentage of the features-selection efficiency.
6. This invention provides the high-level of computer power and functionality of operating using a range of MRI imaging protocols.
7. To improve the usage and clinical decision supports Pathologists and radiologists Needs of biopsies, Early Detection, and treatment of lung cancer.

Dated this 15th September 2025