Intellectual Property India Publishes Patent Application for System And Method For Explainable Deep Transfer Learning-Based Breast Cancer Histopathology Image Classification With Intelligent Hospital Recommendation

MUMBAI, India, July 30 -- Intellectual Property India has published a patent application (202641088033 A) filed by Dvr & Dr. Hs Mic College Of Technology; Dr. Rrajaramesh Merugu; Mr. Mahanthi Kasaragadda; Ms. Shaik. Aasha; and Ms. Bommisetti Triveni on July 18, 2026, for System And Method For Explainable Deep Transfer Learning-Based Breast Cancer Histopathology Image Classification With Intelligent Hospital Recommendation.

Inventors include Dr. Rrajaramesh Merugu; Mr. Mahanthi Kasaragadda; Ms. Shaik. Aasha; and Ms. Bommisetti Triveni.

The application for the patent was published on July 24, 2026, under issue no. 30/2026.

Abstract: Breast cancer is a leading cause of cancer-related death among women around the world, which is why it's essential to get an early and accurate diagnosis to save the lives of many patients and their families. Traditional histopathological investigation mainly depends on the experience of pathologists and is typically a slow process leading to inconsistencies between observers and the delay of clinical decisions. With the advent of deep learning and artificial intelligence, automated diagnosis of breast cancer has achieved a higher degree of reliability. Still, many present systems provide predictions of the outcome without revealing the underlying reasoning or how to apply these predictions practically in a clinical setting. We are describing a new system and method for breast cancer histopathology image classification by deep transfer learning that are also intelligible and provide hospital recommendations. The primary feature extraction of this system is done via DenseNet121, a pre- trained convolutional neural network. Before classification takes place, the images are preprocessed to ensure proper quality, decrease overfitting, and enhance model generalization. The whole learning process is composed of two steps first by using pretrained frozen layers for feature extraction and then, fine-tuning selected upper layers for domain-specific feature learning to achieve optimal results. The invention embeds an interpretability (XAI) component built upon a Gradient-weighted Class Activation Mapping (Grad-CAM) technique. The interpretability component creates visual heat maps which show which parts of a histopathological image have the most impact on the classification decision.

Disclaimer: Curated by HT Syndication.