MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202641127093 A) filed by Karpagam Academy Of Higher Education; Karpagam Institute Of Technology; Dr. P. Rutravigneshwaran; and Pavithra S on October 26, 2026, for Vgg-19 Transfer Learning Based Convolutional Neural Network For Analysing And Diagnosing Mri Brain Tumor Images.
Inventors include Dr. P. Rutravigneshwaran; and Pavithra S.
The application for the patent was published on October 30, 2026, under issue no. 44/2026.
Abstract: Brain tumors constitute one of the most severe neurological disorders, often leading to irreversible cognitive impairment, physical disability, and high mortality rates if not diagnosed at an early stage. Accurate detection and classification of brain tumors play a crucial role in clinical decision-making, treatment planning, and improving patient survival outcomes. Magnetic Resonance Imaging (MRI) is the preferred imaging modality for brain tumor diagnosis due to its superior soft-tissue contrast and non-invasive nature. However, manual interpretation of MRI scans is time-consuming, subject to inter-observer variability, and highly dependent on the expertise of radiologists, particularly when tumor boundaries are complex or poorly defined. To overcome these challenges, this work presents a VGG-19 transfer learning–based convolutional neural network framework for automated analysis, detection, and classification of brain tumors from MRI images. The proposed system integrates advanced image preprocessing, deep feature extraction, and supervised classification techniques to enhance diagnostic accuracy and reliability. Initially, MRI brain images undergo preprocessing operations such as noise reduction, intensity normalization, resizing, and contrast enhancement to improve image quality and ensure uniformity across diverse datasets. The core of the proposed framework is a pre-trained VGG-19 deep convolutional neural network, originally trained on large-scale image datasets and subsequently fine-tuned for brain tumor analysis using transfer learning. By leveraging the hierarchical feature learning capability of VGG-19, the model effectively captures both low-level features, including edges and textures, and high-level semantic features associated with tumor shape, size, and intensity variations. This approach significantly reduces the need for extensive labeled medical data while improving model generalization and training efficiency. Following feature extraction, customized fully connected layers and optimized activation functions are employed to classify MRI images into clinically relevant categories, such as normal brain tissue and multiple brain tumor types or grades. The system supports multi-class classification, enabling comprehensive diagnostic assessment rather than limiting analysis to binary tumor detection. Supervised learning and loss optimization techniques are applied to enhance classification performance and robustness. Experimental evaluation demonstrates that the proposed VGG-19 transfer learning–based framework achieves high accuracy, sensitivity, and specificity in brain tumor classification, outperforming traditional machine learning and conventional deep learning models trained from scratch. The automated workflow minimizes human intervention, reduces diagnostic time, and addresses inter-observer variability, making the system suitable for real-world clinical applications. Overall, the proposed method provides an efficient, scalable, and reliable computer-aided diagnostic solution for MRI-based brain tumor analysis. By combining transfer learning with a deep convolutional architecture, the system enhances diagnostic precision, supports clinical decision-making, and contributes to improved patient care and survival outcomes in neuro-oncology. Brain tumors constitute one of the most severe neurological disorders, often leading to irreversible cognitive impairment, physical disability, and high mortality rates if not diagnosed at an early stage. Accurate detection and classification of brain tumors play a crucial role in clinical decision-making, treatment planning, and improving patient survival outcomes. Magnetic Resonance Imaging (MRI) is the preferred imaging modality for brain tumor diagnosis due to its superior soft-tissue contrast and non-invasive nature. However, manual interpretation of MRI scans is time-consuming, subject to inter- observer variability, and highly dependent on the expertise of radiologists, particularly when tumor boundaries are complex or poorly defined. To overcome these challenges, this work presents a VGG-19 transfer learning–based convolutional neural network framework for automated analysis, detection, and classification of brain tumors from MRI images. The proposed system integrates advanced image preprocessing, deep feature extraction, and supervised classification techniques to enhance diagnostic accuracy and reliability. Initially, MRI brain images undergo preprocessing operations such as noise reduction, intensity normalization, resizing, and contrast enhancement to improve image quality and ensure uniformity across diverse datasets. The core of the proposed framework is a pre-trained VGG-19 deep convolutional neural network, originally trained on large-scale image datasets and subsequently fine-tuned for brain tumor analysis using transfer learning. By leveraging the hierarchical feature learning capability of VGG-19, the model effectively captures both low-level features, including edges and textures, and high- level semantic features associated with tumor shape, size, and intensity variations. This approach significantly reduces the need for extensive labeled medical data while improving model generalization and training efficiency. Following feature extraction, customized fully connected layers and optimized activation functions are employed to classify MRI images into clinically relevant categories, such as normal brain tissue and multiple brain tumor types or grades. The system supports multi-class classification, enabling comprehensive diagnostic assessment rather than limiting analysis to binary tumor detection. Supervised learning and loss optimization techniques are applied to enhance classification performance and robustness. Experimental evaluation demonstrates that the proposed VGG-19 transfer learning–based framework achieves high accuracy, sensitivity, and specificity in brain tumor classification, outperforming traditional machine learning and conventional deep learning models trained from scratch. The automated workflow minimizes human intervention, reduces diagnostic time, and addresses inter- observer variability, making the system suitable for real-world clinical applications. Overall, the proposed method provides an efficient, scalable, and reliable computer-aided diagnostic solution for MRI-based brain tumor analysis. By combining transfer learning with a deep convolutional architecture, the system enhances diagnostic precision, supports clinical decision- making, and contributes to improved patient care and survival outcomes in neuro-oncology.
Disclaimer: Curated by HT Syndication.
