MUMBAI, India, Feb. 6 -- Intellectual Property India has published a patent application (202641006673 A) filed by Usha Rani Macigi, Tirupati, Andhra Pradesh, on Jan. 23, for 'enhanced pneumonia diagnosis using mask r-cnn and resnet50-based deep learning on chest x-rays.'
Inventor(s) include Usha Rani Macigi; and G. Girija Rani.
The application for the patent was published on Feb. 6, under issue no. 06/2026.
According to the abstract released by the Intellectual Property India: "This study proposes a robust and efficient two-stage deep learning framework aimed at the accurate classification of Chest X-ray images into NORMAL and PNEUMONIA categories. The methodology integrates Mask R-CNN for precise segmentation of lung regions, followed by a fine-tuned RestNet50 model for binary classification. By effectively isolating lung areas, Mask R-CNN enhances the relevance of extracted features, thereby improving diagnostic performance. On the original imbalanced dataset, the framework achieved a classification accuracy of 90.71%. Upon the incorporation of data augmentation techniques to address class imbalance, the performance significantly improved, reaching an accuracy of 95.26%. These results underscore the critical role of data balancing in enhanced model generalization and reliability. This research highlights the synergistic benefit of combining instance segmentation with deep classification networks for superior pneumonia detection in chest radiographs. The integration of segmentation enables the model to focus on diagnostically relevant regions, thus boosting the efficacy of feature ex traction an d classification. Future research will investigate the inclusion of clinical metadata (e.g.: age, symptoms, comorbidities) to further personalize and refine diagnostics accuracy. This advancement holds potential for facilitating the deployment of AI-assisted diagnostic tools in the real-world clinical settings, ultimately contributing to timely and reliable medical decision making."
Disclaimer: Curated by HT Syndication.
