Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 31 No. 2s (2024) 253 https://internationalpubls.com Advancements in Non-Linear Content-Based Image Retrieval (CBIR) Systems for Image Analysis 1Dr Y Padma, 2Koppula Srinivas Rao, 3Dr. K. Swathi, 4Rambabu Pemula, 5*Ms. Faiza Iram, 6Kayam Sai kumar 1Assistant Professor, Department of Information Technology, P V P Siddhartha Institute of Technology, Vijayawada , padmayenuga@pvpsiddhartha.ac.in 2Professor in Department of Computer Science and Engineering, MLR Institute of Technology, Hyderabad, Email: ksreenu2k@gmail.com 3Associate Professor, Dept. of CSE, Koneru Lakshmaiah Education Foundation, Vaddeswaram, A. P. dr.kswathi@kluniversity.in 4Associate Professor, Department of Artificial Intelligence, Vidya Jyothi Institute of Technology, Hyderabad, India, rpemula@gmail.com 5Assistant Professor, School of Computer Science and Artificial Intelligence, SR University, Warangal, Telangana, India, Email : faiza.iram@sru.edu.in 6School of engineering, department of CSE, Malla Reddy University, Maisammaguda, Dulapally, Hyderabad, Telangana 500043, saikumarkayam4@ieee.org Article History: Received: 30-03-2024 Revised: 15-05-2024 Accepted: 25-05-2024 Abstract: Content-Based Image Retrieval (CBIR) systems have witnessed significant advancements in recent years, particularly in the domain of medical image analysis. This paper provides an introductory overview of these advancements in CBIR systems tailored for medical image analysis. The evolution of CBIR techniques and their applications in the medical field are discussed, highlighting their importance in tasks such as diagnosis, treatment planning, and research. Various modalities of medical imaging, including X-rays, CT scans, MRI, and PET, are addressed in the context of CBIR systems. Lung cancer incidence rates vary across regions, with the highest rates observed in North America, Europe, Eastern Asia, and South America, and the lowest rates in certain geographic areas. Moreover, the integration of machine learning and deep learning approaches in CBIR systems is explored, emphasizing their role in enhancing image retrieval accuracy and efficiency. Additionally, challenges and future directions in the development and deployment of CBIR systems for medical image analysis are discussed, providing insights into potential avenues for further research and innovation in this rapidly evolving field. Keywords: Machine Learning, Deep Learning, Neural Networks, Pathology, Histopathology. 1. Introduction In recent years, Content-Based Image Retrieval (CBIR) systems have emerged as crucial tools in the realm of medical image analysis, offering innovative solutions for retrieving and analyzing medical images based on their content characteristics. These systems leverage advanced algorithms and techniques to enable efficient and effective retrieval of medical images, facilitating tasks such as diagnosis, treatment planning, and research. This introduction provides an overview of the advancements in CBIR systems specifically tailored for medical image analysis. Carcinoma is the Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 31 No. 2s (2024) 254 https://internationalpubls.com leading cause of cancer-related deaths globally, affecting both men and women. According to WHO reports, there were an estimated 9.6 million deaths due to lung cancer in 2018, with approximately 2.09 million new cases reported. As of January 1, 2016, there were over 15.5 million Americans living with a history of cancer, many of whom were diagnosed years ago and are currently disease-free. Lung cancer incidence rates vary across regions, with the highest rates observed in North America, Europe, Eastern Asia, and South America, and the lowest rates in certain geographic areas. Despite lower smoking prevalence, lung cancer rates among Chinese women exceed those in European countries due to indoor pollution. The evolution of CBIR technology has been driven by the increasing demand for automated image analysis tools in the medical field. Traditional methods of image retrieval often rely on text- based queries or manual annotations, which may not fully capture the rich content of medical images. CBIR systems, on the other hand, analyze the visual content of images, including features such as texture, shape, and intensity, to retrieve images that are visually similar to a given query image. One of the key areas of advancement in CBIR systems for medical image analysis is the integration of machine learning and deep learning techniques. These approaches enable CBIR systems to learn complex patterns and relationships within medical images, leading to improved accuracy and robustness in image retrieval tasks. Additionally, the availability of large-scale medical image datasets has facilitated the training of deep learning models, further enhancing the performance of CBIR systems. Another important aspect of CBIR systems in medical image analysis is their ability to handle various modalities of medical imaging, including X-rays, CT scans, MRI, PET, and more. Each modality presents unique challenges and characteristics, and CBIR systems must be able to adapt to these differences to provide accurate and relevant image retrieval results. Despite the significant advancements in CBIR systems for medical image analysis, several challenges remain. These include issues related to image variability, noise, and interpretation, as well as the need for standardized evaluation metrics and benchmarks. Addressing these challenges will be crucial for further improving the performance and usability of CBIR systems in the medical domain. In the current landscape of clinical imaging technology, advancements across various disciplines within the medical field have significantly increased their utilization and formed a robust network for research and diagnosis. These advancements have led to the emergence of numerous sophisticated techniques for automatically extracting valuable information from radiology images, consequently unveiling new insights. The accuracy of any diagnostic method utilizing medical images is pivotal, with expert diagnosis being essential. A modified procedural model for medical image information retrieval systems, focusing particularly on the classification of mammogram and lung nodule images, has been proposed. The development of algorithms for automated, computer-assisted content retrieval of medical images poses a challenging research area aimed at achieving precise diagnoses. A schematic diagram illustrating this approach is presented in Figure 1. This research offers a novel methodology for enhancing classification accuracy and grading identification in digital mammogram and lung nodule analyses. Such improvements hold promise for advancing future research endeavors. This section provides an overview of the subject domain, challenges, motivations, implications, and significant contributions of the work. Overall, the introduction of advancements in CBIR systems for medical image analysis Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 31 No. 2s (2024) 255 https://internationalpubls.com contributes to the ongoing efforts to enhance diagnostic accuracy, automate image retrieval processes, and foster innovation in the field of medical imaging. • Enhanced Classification Accuracy: The proposed methodology introduces modifications to existing procedural models for medical image information retrieval systems, particularly focusing on the classification of mammogram and lung nodule images. Through the development of sophisticated algorithms, the work aims to improve the accuracy of classification, aiding in more precise diagnoses. • Automated Content Retrieval: By developing algorithms for automated, computer-assisted content retrieval of medical images, the research addresses a critical challenge in the field. This automation streamlines the process of information extraction from radiology images, facilitating faster and more efficient diagnosis. • Novel Approach for Diagnosis: The unique approach presented in this work offers a fresh perspective on medical image analysis. By leveraging advancements in CBIR systems, the research provides new insights into the classification and identification of digital mammogram and lung nodule images, contributing to the advancement of diagnostic methodologies. • Scope for Further Research: The introduction of these advancements not only improves the current state of medical image analysis but also opens up avenues for future research. The work identifies challenges and opportunities in the field, laying the groundwork for continued exploration and innovation in CBIR systems for medical imaging. 2. Literature Survey De Oliveira et al., [1] accomplished a 2-dimensional head phase examination (PCA) to stunned the modern-day PCA as it is professed toward be much less hard also increasingly more direct to usage on behalf of the portrayal of bosom thickness floor. The recovery method becomes finished utilizing a help vector system (SVM) which tackled fluctuated studying, characterization, and expectation troubles. Then once more, Sharma et al. [2] proposed some other method to decide the closeness between histological pictures thru the diagram theoretic depiction and coordinating in convalescing histological photographs from larger databases. As against the above techniques that straightforwardly measure the similitude as far as image information by myself, Classifier-based totally closeness measures don't legitimately gauge likeness regarding picture statistics while contrasted with the approach referenced earlier than. It utilizes the grouping of a query image depending on a stable arrangement of foreordained names to survey comparability [18,36]. Popova and Neshov [3] endeavored to increase the recovery viability also exactness of picture examine in enormous records. Their works prescribed two arrangements of various element blends that perform well for therapeutic picture classifications, IRMA. The main three highlights utilized in the examination were Color Layout, Edge Histogram, and DCT Coefficients, which were altogether joined on behalf of a greater picture re-unimportant positioning than others that depended on separable highlights. The outcomes restored a 14.49% development of the recovery MAP (50.68% for consolidated highlights Set). Later on, the exploration is fixed to research fresh highlights also apply Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 31 No. 2s (2024) 256 https://internationalpubls.com fitting loads on behalf of the specific highlights utilized in the blend. In dealing with an uproarious picture in Relevance Feedback CBIR frameworks, Da Silva et al., [4] directed three examinations utilizing mammograms taken from the University of South Carolina and the University of Sao Paulo. This strategy of highlight determination upgraded closeness search exactness and brought down significantly information dimensionality, which thusly improved access strategies effectiveness and the CBIR framework. For future work, it is prescribed that another nearby search into GA and the collaboration flanked by channel based techniques and the GA wrapper-based strategy in CBIR frameworks can be acquainted with expand the productivity of the suggested technique. Other than those, all literary data on patients' clinical history container remain consolidated into the inquiry component. Additionally, Wei et al. [5] anticipated a lot of mammogram descriptors as indicated by the BI-RADS definition. The proposed getting to know method enhanced the characterization precision as high as 72% in addition to seventy four%. The calcification Query-By-Example (QBE) become accounted for to be somewhat better than the mass QBE is an increasing number of feasible in casting off calcification sores whilst contrasted with mass accidents. This investigation settled that in the time spent issue extraction, the calcification highlights can face up to ability outcomes of errors, commotions, and misses in spot identity. Mass sores, in spotlight examinations with the ones of calcification, are visible as increasingly sensitive to division imprecision and extraction. This examination plans to explore lengthy haul studying in importance input for an ongoing framework to accumulate customers' inquiry information. Singh and Hemachandran [6] proposed a CBIR framework utilizing Color minutes and Gabor surface highlights. To extricate the shading data from the photograph, the photograph is remoted into three non-covering quantities. The preliminary 3 snapshots of the shading dispersions are extricated from each fragment. HSV shading area is utilized for the shading spotlight extraction. Surface highlights are extricated making use of Gabor floor descriptor in addition to joined through the shading spotlight vector. They utilized the Canberra separation to figure the separation among the inquiry image also the database pix. For check determination, they utilized the Coral photograph exhibition and the outcomes are contrasted also the beyond CBIR strategies. To quantify the exhibition of the projected strategies, exactness and overview have been applied on the presentation measurements. The creators likewise carried out examinations using just shading or surface highlights yet the outcomes showed that solitary shading or floor can not talk to the photo well. Yue et al., [7] utilized the co-occasion lattice for disposing of the shading and floor highlights. They concept approximately the exhibition of community shading histogram, global shading histogram & floor highlights in place of a CBIR framework. On behalf of this exam, they deliberate a CBIR technique utilizing melded shading and floor highlights. Sakhar and Nasre [8] applied the nearby shading histogram, international shading histogram and fluffy shading histogram for the shading spotlight removal. To extricate surface, they applied Tammura highlights also the wavelet trade. The euclidean separation become utilized to manner the separation among the query picture and the database photos. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 31 No. 2s (2024) 257 https://internationalpubls.com Arai et al. [9] carried out 2D and 3-D local binary patterns (LBP) for the order of knobs. The association precision expending 3D LBP is 78% while exactness is 43% utilizing 2D LBP Han et al. [10] tested the process of 3 understood floor highlights (Haralick, Gabor, and LBP) in the grouping of pneumonic knobs. They installation the significance of Haralick consists of while contrasted with the Gabor and LBP highlights. The knobs have been looked after as generous and harmful relying on the composite role of chance. Haralick's highlights processed in 3-d were utilized for knob order. The most multiplied arrangement exactness was obtained via considering the knobs with the composite function of risk "1" and "2" as kindhearted and "four" and "five" as dangerous. This method is predicated upon radiologists for a department of knobs and thus cannot be utilized by and by way of Suzuki et al. [11] applied a huge preparing ANN for an association of benevolent and dangerous knobs. Three dim level-based totally highlights, two area-based totally highlights, a morphological issue, and medical statistics had been applied to talk to the knobs. They moreover assessed the classifier execution making use of the round-robin technique for making ready and testing. McNitt-Gray et al. [12] used to form also surface highlights of a delegate knob reduce for arrangement. Knobs are sectioned utilizing a semi-automatic forming approach. A few quantitative measures had been extricated identified with length, shape, lessening, appropriation of constriction, and floor. A step- smart discriminant investigation was finished. 3. Proposed Approach 3.1 Over view The origins of image retrieval can be traced back to the late 1970s. Early techniques [13] primarily focused on textual annotations associated with images rather than their graphical content. These systems found applications in various domains including crime prevention, intellectual property management, journalism, advertising (including video), and internet searching. • Content-Based Image Retrieval (CBIR) Over the past two decades, numerous research endeavors have been initiated to create fast and cost- effective image retrieval systems. Smeulders et al. and Rui et al. have conducted comprehensive reviews on Content-Based Image Retrieval (CBIR) system techniques, focusing on various anatomical organs. Veltkamp et al. [14] have provided a detailed survey of existing CBIR systems [15]. These systems are designed to search through image databases using graphical content such as color, texture, and shape elements [16]. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 31 No. 2s (2024) 258 https://internationalpubls.com Figure 1. General CBIR system for Medical Images • CBIR applications in medical domain Content-Based Image Retrieval (CBIR) has emerged as a crucial research area in radiology, aiming to streamline diagnostic decision-making processes through medical image interpretation [17]. A significant research challenge lies in developing algorithms for automated or computer-assisted classification and retrieval of medical images based on their structural content, particularly as regions of interest (ROIs) often exhibit irregularities, overlap, partial obstruction, or high degrees of generalization [18]. While content-based image retrieval has been frequently proposed for use in medical image management [19], only a few content-based retrieval systems have been specifically developed for medical imagery. Ultimately, the expert selects the most similar images, from their perspective, from the set of retrieved images and accesses the associated data [20]. As medical images are digitally represented in various formats depending on their modality and the imaging device used, image retrieval systems need to be developed for their specific image types [21]. To adapt CBIR systems for medical images, algorithms need to be developed that can accurately identify ROIs [22]. • Challenges Encountered in Primary Retrieval and Classification The presence of small pulmonary nodules often coincides with vascular intersections, leading to false positives in automated lung nodule detection [23]. Nodules smaller than the CT slice thickness are filtered out using the partial volume effect. Pulmonary nodules exhibit varying internal and external connections, posing significant challenges for developing a generalized segmentation method and improving segmentation accuracy [24]. Another challenge lies in the representation of pulmonary nodules, as exploiting a diverse set of features poses a significant challenge in designing a retrieval or classification framework as represented in figure 2 and figure 3 [25]. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 31 No. 2s (2024) 259 https://internationalpubls.com Figure 2. Block diagram of CBIR System for pulmonary nodules Figure 3. Steps for classification of pulmonary nodules Steps Involved in Pulmonary Nodule Detection and Classification The primary steps for the development of a retrieval framework or a classification scenario involve the segmentation of nodules and extraction of features from each segmented nodule [26]. Segmentation of Pulmonary Nodules has been explained. The majority of previous works have focused on the segmentation of solid nodules [27]. Some alternative approaches have been proposed for the segmentation of non-solid nodules [28], with only one approach addressing the segmentation of solid, part-solid, and non-solid nodules [29]. Henschke et al. highlight that part-solid and non-solid nodules are at a higher risk of malignancy compared to solid nodules, emphasizing the importance of distinguishing between part-solid and non- solid nodules in the segmentation process [30]. Algorithm Input: Image database containing medical images with associated metadata. Query image for which similar images are to be retrieved. Preprocessing: Extract relevant features from each image in the database. Store the features along with their corresponding image identifiers in a structured format. Query Processing: Extract features from the query image. Calculate similarity between the query image features and features of images in the database. Rank the database images based on their similarity to the query image. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 31 No. 2s (2024) 260 https://internationalpubls.com Output: Retrieve the top-ranked images from the database as the search results. Display Results: Present the retrieved images to the user in a user-friendly format, such as thumbnails or a gallery view. Provide options for the user to explore detailed information about each retrieved image, Results MATLAB_R2014a was used to develop a flow model for detecting and removing lung nodules. Cancer Imaging (CI) online data was utilized, comprising CT scan (CTS) images with dimensions of 512.0x512.0 pixels. A total of 90 images were used for training the system, with 65 containing nodules and 25 without. During the testing phase, 150 images were validated, all adhering to the DICOM format commonly used in radiology. Optimal thresholding (th) was employed to convert grayscale images into binary images (BI) with a threshold value of 0.42. The outcomes from CI data were categorized into the following groups to obtain precise and organized results (RLTS): True Positive (TP) for images correctly identifying a disease when present, True Negative (TN) for images correctly indicating no disease when absent, False Positive (FP) for images incorrectly detecting a disease when none exists, and False Negative (FN) for images failing to detect a disease that does exist. These classified results were used to determine the system's efficiency, yielding the following results: 𝐴𝑐𝑐𝑢𝑟𝑎𝑐𝑦 = 𝑇𝑃 + 𝑇𝑁 𝑇𝑃 + 𝐹𝑃 + 𝑇𝑁 + 𝐹𝑁 𝑆𝑒𝑛𝑠𝑖𝑡𝑖𝑣𝑖𝑡𝑦 = 𝑇𝑃 𝑇𝑃 + 𝐹𝑁 The accuracy of the outcomes represents the proportion of correct results in the overall outcome set. Sensitivity reflects the importance of positive results when the patient has the disease. Specificity indicates the reliability of negative results when the individual has the illness. In the final set of 150 images analyzed, the system yielded 102 true positive (TP) outcomes, 21 false positive (FP) outcomes, 24 true negative (TN) outcomes, and 3 false negative (FN) outcomes. These parameters of accuracy, sensitivity, and specificity were then calculated. Following the removal of geometric features, mathematical constraints using GLCM were computed. The results of nodule detection are illustrated in Figures 5.10 and 5.11. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 31 No. 2s (2024) 261 https://internationalpubls.com Figure 4 Statistical features after segmented Lung nodule malignant Figure 5 Statistical features after segmented Lung nodule benign The potential outcomes in discrete categorization include true positive (TP), false positive (FP), true negative (TN), and false negative (FN). FN occurs when a sample is incorrectly identified as positive when it's actually negative, leading to misclassification. True positive and true negative refer to correct identifications of positive and negative samples, respectively. Using the confusion matrix, accuracy, precision, true positive rate (TPR), and false positive rate (FPR) are calculated using equations 4.1 and 4.2. 𝑇𝑃𝑅 = 𝑇𝑃 𝑇𝑃 + 𝐹𝑁 Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 31 No. 2s (2024) 262 https://internationalpubls.com 𝐹𝑃𝑅 = 𝐹𝑃 𝑇𝑁 + 𝐹𝑃 Figure 6: Distribution of image samples for benign and malignant cases. The effectiveness of the proposed retrieval system is evaluated for each query using average precision [29]. Equation (1) can be utilized to calculate the area under the precision-recall curve for each query, where precision represents the ratio of relevant images to all retrieved images. 𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 = 𝑟𝑒𝑙𝑒𝑣𝑎𝑛𝑡 𝑖𝑚𝑎𝑔𝑒𝑠 − 𝑟𝑒𝑡𝑒𝑟𝑖𝑣𝑒𝑑 𝑖𝑚𝑎𝑔𝑒𝑠 𝑟𝑒𝑡𝑒𝑟𝑖𝑣𝑒𝑑 𝑖𝑚𝑎𝑔𝑒𝑠 In this context, the query images were shown on samples like animals and buses. Figures 4,5 and figure 6 display the retrieved images from an animal database, with the corresponding accuracy of each image relative to the searched image provided below [30]. Figure 7: Retrieved animal images. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 31 No. 2s (2024) 263 https://internationalpubls.com Figure 8: Retrieved bus images. Conclusion In this survey, we have summarized recent developments of Content-Based Image Retrieval from technological and practical applications in the last decade. First, we review the developments of image representation (or feature extraction) and database search for CBIR. We then present the typical practical applications of CBIR on fashion image retrieval, person re-identification, e-commerce product retrieval, remote sensing image retrieval and trademark label image retrieval, respectively. By leveraging content-based features such as texture, shape, and intensity, these systems have facilitated tasks such as diagnosis, treatment planning, and research, ultimately leading to improved patient care and outcomes. We've developed a two-step CBIR method for lung nodule classification, using the LIDC-IDRI CT dataset. With this dataset, we can retrieve nodule probability and query nodule characteristics as described by radiologists, such as calcification and sphericity. The lung volume is segmented using optimal thresholding and intensity level adjustment. Geometric features are extracted for nodule and vessel classification, with an LDA classifier used to analyze geometric feature distribution. Statistical features using GLCM are also calculated for nodules. Our proposed algorithm Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 31 No. 2s (2024) 264 https://internationalpubls.com employs median filtering to reduce noise and enhance segmentation accuracy, achieving 87% precision and 53.33% specificity. References [1] De Oliveira, J. E., Machado, A. M., Chavez, G. C., Lopes, A. P. B., Deserno, T. M., & Araújo, A. D. A. (2010). MammoSys: A content-based image retrieval system using breast density patterns. Computer methods and programs in biomedicine, 99(3), 289-297. [2] Sharma, H., Alekseychuk, A., Leskovsky, P., Hellwich, O., Anand, R. S., Zerbe, N., & Hufnagl, P. (2012). Determining similarity in histological images using graph-theoretic description and matching methods for content- based image retrieval in medical diagnostics. Diagnostic pathology, 7(1), 134. [3] Popova, A. A., & Neshov, N. N. (2013). Combining Features Evaluation Approach in Content-Based Image Search for Medical Applications [4] Da Silva, S. F., Ribeiro, M. X., Neto, J. D. E. B., Traina-Jr, C., & Traina, A. J. (2011). Improving the ranking quality of medical image retrieval using a genetic feature selection method. Decision Support Systems, 51(4), 810-820 11. [5] Wei, C. H., Li, Y., & Huang, P. J. (2011). Mammogram retrieval through machine learning within BI-RADS standards. Journal of biomedical informatics, 44(4), 607-614. [6] Singh, S. M., & Hemachandran, K. (2012). Content based image retrieval based on the integration of color histogram, color moment and Gabor texture. International Journal of Computer Applications, 59 [7] Huang, W., Zhang, P., & Wan, M. (2013). A novel similarity learning method via relative comparison for content- based medical image retrieval. Journal of digital imaging, 26(5), 850- 865. [8] Sakhare, S. V., & Nasre, V. G. (2011). Design of feature extraction in content based image retrieval (CBIR) using color and texture. International Journal of Computer Science & Informatics, 1(II). [9] Ma, Y. et al. (2010). “A novel shape feature to classify microcalcifications”. In: 17th IEEE International Conference on Image Processing (ICIP). IEEE, pp. 2265–2268 [10] Rui, Y., Huang, T. S., Ortega, M., & Mehrotra, S. (1998). Relevance feedback: a power tool for interactive content- based image retrieval. IEEE Transactions on circuits and systems for video technology, 8(5), 644-655. 9. [11] de Ves, E., Benavent, X., Coma, I., & Ayala, G. (2016). A novel dynamic multi-model relevance feedback procedure for content-based image retrieval. Neurocomputing, 208, 99- 107. [12] Strange, H. et al. (2014). “Modelling mammographic microcalcification clusters using persistent mereotopology”. In: Pattern Recognition Letters 47, pp. 157–163 [13] T. Kubota, A. K. Jerebko, M. Dewan, M. Salganicoff, and A. Krishnan, “Segmentation of pulmonary nodules of various densities with morphological approaches and convexity models,” Medical Image Analysis, vol. 15, no. 1, pp. 133–154, 2011. [14] J. Dehmeshki, H. Amin, M. Valdivieso, and X. Ye, “Segmentation of pulmonary nodules in thoracic CT scans: A region growing approach,” IEEE Transactions on Medical Imaging, vol. 27, no. 4, pp. 467–480, 2008. [15] S. Kligerman and C. White, “Imaging characteristics of lung cancer,” Seminars in Roentgenology, vol. 46, no. 3, pp. 194–207, 2011. [16] D. S. Raicu, E. Varutbangkul, J. G. Cisneros, J. D. Furst, D. S. Channin, and S. G. Armato III, “Semantics and image content integration for pulmonary nodule interpretation in thoracic computed tomography,” in Proceedings of SPIE Medical Imaging 2007, San Diego, USA, February 2007, pp. 65 120S–65 120S–12. [17] W. H. Horsthemke, D. S. Raicu, and J. D. Furst, “Characterizing pulmonary nodule shape using a boundary-region approach,” in Proceedings of SPIE Medical Imaging 2009, vol. 7260, San Diego, United States, February 2009, pp. 72 602Y–72 602Y–9. [18] J. Xu, S. Napel, H. Greenspan, C. F. Beaulieu, N. Agrawal, and D. Rubin, “Quantifying the margin sharpness of lesions on radiological images for content-based image retrieval,” Medical Physics, vol. 39, no. 9, pp. 5405–5418, September 2012. [19] M. R. Teague, “Image analysis via the general theory of moments,” Journal of the Optical Society of America, vol. 70, no. 8, pp. 920–930, 1980. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 31 No. 2s (2024) 265 https://internationalpubls.com [20] R. M. Rangayyan, N. M. El-Faramawy, J. L. Desautels, and O. A. Alim, “Measures of acutance and shape for classification of breast tumors,” IEEE Transactions on Medical Imaging, vol. 16, no. 6, pp. 799–810, 1997. [21] Revathi, R., Kumar, K. S., Hamid, H. A., & Ahammad, H. S. (2024, March). A real time homogeneous wireless sensor network functioning using semi-tier clustering protocol for energy efficiency routing. In AIP Conference Proceedings (Vol. 2919, No. 1). AIP Publishing. [22] Yamsani, N., Sarada, K., Ahmed, M. A., & Saikumar, K. (2024, March). Estimate and prevention of malicious URL using logistic regression ML techniques. In AIP Conference Proceedings (Vol. 2919, No. 1). AIP Publishing. [23] Kumar, G. N. K., Reddy, M. S., Malleswari, D. N., Rao, K. M., & Saikumar, K. (2024). A Real-Time Hadoop Bigdata Maintenance Model using A Software-Defined and U-Net Deep Learning Mode. International Journal of Intelligent Systems and Applications in Engineering, 12(7s), 364-376. [24] Saikumar, K., & Rajesh, V. (2024). A machine intelligence technique for predicting cardiovascular disease (CVD) using Radiology Dataset. International Journal of System Assurance Engineering and Management, 15(1), 135- 151. [25] Kumar, K. S., Kalangi, R. R., Gurrala, L., Hanoon, N., Suman, M., & Saikumar, K. (2023, December). Revolutionizing Image Recommendations: A Novel Approach with Social Context and CNN. In 2023 3rd International Conference on Smart Generation Computing, Communication and Networking (SMART GENCON) (pp. 1-6). IEEE. [26] Kalangi, R. R., Rao, K. S., Prakash, D. B., Iram, F., Suman, M., & Saikumar, K. (2023, December). Phishing Attack Detection using Collaborative Learning Approach at Dynamic Cloud Platform. In 2023 Global Conference on Information Technologies and Communications (GCITC) (pp. 1-6). IEEE. [27] Ramana, S. S. V., Mydhili, T., Siddardha, P., Charyulu, G. M., & Saikumar, K. (2023). Computational Intelligence Techniques for Behavioral Research on the Analysis of Investment Decisions in the Commercial Realty Market. In Computational Intelligence for Modern Business Systems: Emerging Applications and Strategies (pp. 47-62). Singapore: Springer Nature Singapore. [28] Srinivasarao, G., Rajesh, V., Saikumar, K., Baza, M., Srivastava, G., & Alsabaan, M. (2023). Cloud-Based LeNet- 5 CNN for MRI Brain Tumor Diagnosis and Recognition. Traitement du Signal, 40(4). [29] Saikumar, K., Iram, F., Sampath Dakshina Murthy, A., Alzeyadi, A. K., & Al-Ameedee, S. A. (2023). Enhancing the Protection of Information in Digital Voting Using Application of Block Chain Technology. In Micro- Electronics and Telecommunication Engineering: Proceedings of 6th ICMETE 2022 (pp. 357-365). Singapore: Springer Nature Singapore. [30] Vadlamudi, M. N., Saikumar, K., Kota, S., Al-Hameed, M., Zearah, S. A., & Kanuganti, D. (2023, April). Exploratory Review of Deep Learning Techniques for ECG Signals Analysis. In 2023 IEEE International Conference on Contemporary Computing and Communications (InC4) (Vol. 1, pp. 1-5). IEEE.