DRCNN-Lesion Proxy: a hybrid CNN with lesion-inspired feature simulation for diabetic retinopathy severity classification Priyadharshini Sekar, Kanaga S. Suba Raja, Ramaswamy Krishnaraj Scientific Reports, 2025 Diabetic Retinopathy (DR) remains a leading cause of vision loss globally, necessitating accurate and scalable diagnostic solutions. Existing Deep Learning (DL) models often underutilize lesion-specific cues that are critical for early DR grading, while detection based models require costly lesion annotations. To address these limitations, we propose DRCNN-Lesion Proxy, a hybrid architecture that integrates a ResNet34 based CNN backbone for extracting global image level features with a Lesion Proxy Module, which simulates lesion-inspired cues without explicit lesion bounding box annotations. These heterogeneous features are fused through a late fusion classification head to enable robust multiclass DR severity prediction. The model was trained on a composite dataset and rigorously evaluated across six publicly available benchmarks namely EyePACS, Messidor-2, APTOS 2019, DDR, DIARETDB1, and IDRiD. Experimental findings show that the proposed framework consistently outperforms baseline CNNs and recent hybrid methods, achieving up to 98.37% accuracy, 97.28% F1-score, and 98.14% AUC. Statistical significance testing confirmed that these improvements were not due to chance. Furthermore, Grad-CAM visualizations highlighted clinically relevant retinal regions, and a pilot validation with three ophthalmologists on 20 cases reported mean scores above 3.5 out of 5, confirming that the explanations were perceived as clinically meaningful and useful for grading. The proposed framework provides an annotation light solution with strong generalizability, diagnostic precision, and clinically validated interpretability, advancing the state of the art in automated DR screening and offering a practical pathway for real world deployment.
An Efficient and Secure WBAN Based on Optimal Privacy Preservation Scheme With Deep Learning and Blockchain Technology Balasubramanian Chandra, Subramanian Kanaga Suba Raja, Suresh Sudha Transactions on Emerging Telecommunications Technologies, 2025 Securing the trustworthiness, privacy, and legitimacy of shared medical data in Wireless Body Area Network (WBAN) is a primary concern. Hence, a blockchain technology‐based secure medical data storage scheme is developed in this paper. This developed model includes four primary phases. Before initializing, the WBAN data are collected. In the first phase, the user authentication is verified. For this purpose, the user's iris images are aggregated. These iris images are subjected to the Residual Attention Network (RAN). From the RAN, the user is authorized, and then security keys are given to the authorized user. Only after verifying the authentication of the user, the healthcare data is allowed to be stored in the blockchain. In the second phase, data sanitization takes place. The obtained WBAN medical data are sanitized using a data sanitization process with the optimal keys obtained from the Fusion of Golden Eagle and Eurasian Oystercatcher Optimization Algorithm (FGE‐EOOA). Here, the data are encrypted by employing the Rivest‐Shamir‐Adleman (RSA) approach, and then encrypted medical data are stored in the blockchain. This ensures multi‐step data security, which allows secure storage of WBAN healthcare data in the blockchain. While retrieving the stored data, the user authentication is verified on the user side, as well as in the same RAN model. This is the third phase of the developed model. When the user is proven to be an authorized one, the stored data in the blockchain corresponding to that particular user is retrieved. Using the data restoration process, which is the fourth phase of the developed model, the actual medical data is retrieved. If the user is unauthorized, then no access is provided to them. This ensures a multi‐level of security for storing and retrieving data from the blockchain. The security offered by this model is evaluated and validated by contrasting and comparing it with other conventional data transfer methods.
Caries-segnet: Multi-scale cascaded hybrid spatial channel attention encoder-decoder for semantic segmentation of dental caries Jayaraman Priya, Subramanian Kanaga Suba Raja Biomedizinische Technik, 2025 Objectives Dental caries is a prevalent oral health issue around the world that leads to tooth aches, root canal infections, and even tooth extractions. Existing dental caries diagnosis models may misdiagnose the disorder and take more time to segment the caries. This research work aims to provide an in-depth analysis of spatial and channel attention mechanism techniques used for semantic segmentation in an encoder-decoder network. For effective performance, the research implements novel techniques to segment the dental caries accurately. Methods Deep Fully Connected Residual Block (DFCR) is designed to provide relevant features without the loss of significant information. A novel Hybrid Spatial Channel Attention (HSCA) module is developed for combining significant features with the help of multi-scale spatial features and cross-dimensional channel features. Results The proposed methodology performs better than other cutting-edge algorithms by achieving 96.63 % accuracy, 95.77 % dice score, 96.28 % Intersection over Union (IOU) score for the caries dataset, and 96.93 % accuracy, 95.21 % dice value, and 96.1 % IOU for the Tufts dental dataset. Conclusions The developed model facilitates in detection of cavities precisely at an earlier stage with the help of dental images. The semantic segmentation of dental caries provides accurate diagnosis by assisting medical professionals.
A comprehensive approach enhancing home automation security with artificial intelligence firewalls: Design and evaluation V. Asha, Kanaga Suba Raja S. Navigating Computing Challenges for A Sustainable World, 2025 This study introduced an innovative AI-driven solution to enhance the security of home automation systems, addressing the growing vulnerabilities associated with IoT devices. The proposed system combines sophisticated machine-learning techniques with a blockchain-based data-management platform to analyze network-traffic patterns and device behaviors, ensuring data integrity while proactively identifying and combating various cyber threats. Experimental findings demonstrate the system's ability to recognize and mitigate attacks such as denial-of-service and unauthorized access attempts. During a 24-hour evaluation, the AI-powered firewall processed 10,000 requests, successfully blocking 8% of the malicious traffic and identifying suspicious activities for further examination. Notably, it outperformed traditional systems that lacked anomaly-based algorithms by effectively detecting a wide range of significant threats. Future research will focus on enhancing anomaly detection algorithms through user feedback, highlighting the potential of AI-enhanced firewalls to provide robust protection against cyber threats while preserving data confidentiality in emerging 6G networks and smart home ecosystems.
Advanced potato leaf disease classification using YOLOv9: A deep learning approach S. Kanaga Suba Raja, S. Dharun Kumar, R. Sakthisri Navigating Computing Challenges for A Sustainable World, 2025 Although potato is an important global crop, it suffers from a range of diseases that can cause substantial yield losses. Despite being a significant crop worldwide, potatoes are vulnerable to various diseases that can lead to significant decreases in production. The timely and accurate detection of diseases is crucial for managing crops effectively. This paper outlines a fresh approach for real-time potato disease classification using the YOLOv9 deep learning model. The dataset used to train the model contained both healthy and diseased potato plants, enabling it to identify the main diseases found in potato plants: late blight, early blight, and black leg. The process involved data preprocessing, model training, and evaluation to enhance the model's performance. The findings demonstrate that the YOLOv9 model attains high accuracy and quick inference time, making it suitable for agricultural applications. By utilizing advanced deep learning and substantial computational resources, this study offers a valuable tool to assist farmers and professionals.
Enhancing VPN security: A comparative study of LSTM and GRU models V. Jeyajeev, S. Kanaga Suba Raja, S. P. Thirumukhil Navigating Computing Challenges for A Sustainable World, 2025 This chapter presents a comprehensive assessment of VPN (Virtual Private Network) security using different deep learning algorithms. The purpose of the study is to predict whether a given VPN IP address is safe to use by detecting IP address leaks, DNS leaks, encryption strength, and anomalies in VPN behavior. We use machine deep learning algorithms like Long Short Memory (LSTM), Gated Recurrent Units (GRU), to address these critical aspects. The performance of these algorithms is comprehensively compared using evaluation metrics such as Accuracy percentage, Mean Squared Error (MSE), R-Squared(R^2) etc. The results provide valuable information for choosing the most effective technology to ensure the security and privacy of VPN connections.
Smart Cities Secured: Utilizing AI Firewalls for Sustainable Urban Environments V Asha, S Kanaga Suba Raja International Journal on Smart Sensing and Intelligent Systems, 2025 The rapid transformation of urban landscapes into intelligent cities, propelled by the Internet of Things (IoT), artificial intelligence (AI), and machine learning (ML), presents substantial opportunities for enhancing urban sustainability, efficiency, and quality of life. However, the interconnected nature of smart urban infrastructures renders them susceptible to significant cybersecurity risks, potentially jeopardizing essential services, and citizen privacy. This study examines the implementation of AI firewalls as a robust solution to strengthen the cybersecurity capabilities of smart cities. Utilizing advanced ML techniques for threat detection, anomaly analysis, and automated responses, AI firewalls can safeguard critical urban infrastructure. A comprehensive review of current literature and emerging AI technologies demonstrates that AI firewalls effectively mitigate cyber threats, protect sensitive data, and ensure the continuity of vital services, thereby fostering public trust and promoting sustainability. The proposed methodology outlines a systematic approach for designing, developing, and implementing AI firewalls, emphasizing scalability, adaptability, and energy efficiency in smart city environments. It also addresses challenges such as cost, privacy concerns, and interoperability through solutions including federated learning, edge computing, and explainable AI (XAI). This research highlights the crucial role of AI firewalls in creating secure, resilient, and sustainable smart cities, facilitating proactive urban governance. Future investigations will focus on enhancing real-time threat response, improving cross-platform integration, and refining privacy-preserving techniques, ensuring AI firewalls remain integral to the evolving landscape of secure urban systems. This study underscores the importance of AI firewalls as a key component of smart city cybersecurity, aligning technological advancements with urban resilience and sustainability objectives.
Prognosis of Stroke using Machine Learning Algorithms Kanaga Suba Raja. S, B. Chandra, K. Kausalya, Ciddarth RM, Gokul Ranjith V Proceedings 7th International Conference on Computing Methodologies and Communication Iccmc 2023, 2023
Deep Key Generation: End to End Encryption for Generic Communication Kanaga Suba Raja S, K. Kausalya, B. Chandra, Manoj V, Mahesh Saravanan G, Harish. A Proceedings of the 2nd IEEE International Conference on Advances in Computing Communication and Applied Informatics Accai 2023, 2023
Object Recognition in Soccer Sports Videos Usha Kiruthika S, K. Kausalya, Kanaga Suba Raja S Proceedings 2nd International Conference on Smart Technologies Communication and Robotics 2022 Stcr 2022, 2022
An optimal algorithm to improve resource utilization in cloud data centre Elakkya Manoharan, Deepika.U, Arunkumar.KEaswari Engineering College, RamapuramRamapuram Chennai, India, S.Kanaga suba Raja, Mrs.M. Hema, Easwari Engineering College , Chennai, India International Journal of Engineering and Advanced Technology, 2019
Detection and classification of paddy crop disease using deep learning techniques SRM Institute of Science, Technology, Kattankulathur Campus, Chennai. Tamil Nadu, India, Usha Kiruthika*, Kanagasuba Raja S, Department of IT, SRM Eswari Engineering College, Chenaai, Tamil Nadu, India., Jaichandran R*, Department of CSE, Aarupadai Veedu Institute of Technology, Vinayaka Mission’s Research Foundation (Deemed to be University), Rajiv Gandhi Salai, Old Mamallapuram Road, Paiyanoor 603104, Kanchipuram (DT), Tamil Nadu, India, Priyadharshini C, Department of Information Technology, Easwari Engineering College, Ramapuram Campus, Chennai. International Journal of Recent Technology and Engineering, 2019
Sentiment analysis of movies on social media using R studio Department of CSE, Aarupadai Veedu Institute of Technology, Vinayaka Mission’s Research Foundation (Deemed to be University), Rajiv Gandhi Salai, Old Mamallapuram Road, Paiyanoor 603104, Kanchipuram (DT), Tamil Nadu,, Jaichandran R*, C. Bagath Basha, Department of Computer Science, Engineering, Vinayaka Mission’s Research Foundation, Salem, Tamil Nadu, India., Shunmuganathan K.L, principal, Aarupadai Veedu Institute of Technology, Vinayaka Mission’s Research Foundation (Deemed to be University), Rajiv Gandhi Salai, Old Mamallapuram Road, Paiyanoor 603104, Kanchipuram (DT), Tamil Nadu, India,, Rajaprakash, Department of CSE, Aarupadai Veedu Institute of Technology, Vinayaka Mission’s Research Foundation (Deemed to be University), Rajiv Gandhi Salai, Old Mamallapuram Road, Paiyanoor 603104, Kanchipuram (DT), Tamil Nadu, India,, Kanagasuba Raja S, Department of IT, SRM Eswari Engineering College, Chenaai, Tamil Nadu, India. International Journal of Engineering and Advanced Technology, 2019
Face recognition using Haar - Cascade classifier for criminal identification International Journal of Recent Technology and Engineering, 2019
Demand based crop recommender system for farmers S. Kanaga Suba Raja, R. Rishi, E. Sundaresan, V. Srijit Proceedings 2017 IEEE Technological Innovations in ICT for Agriculture and Rural Development Tiar 2017, 2017
Embedding data in audio signals using HSA-EMD algorithm R. Senthamizh Selvi, R. Kishore, G. R. Suresh, S. Kanaga Suba Raja Iconstem 2017 Proceedings 3rd IEEE International Conference on Science Technology Engineering and Management, 2017
IOT based automation of fish farming Journal of Advanced Research in Dynamical and Control Systems, 2017