MUMBAI, India, Aug. 24 -- Intellectual Property India has published a patent application (202641097942 A) filed by Sri Eshwar College Of Engineering on August 12, 2026, for Multimodal Explainable Ai System For Chest X-Ray Diagnosis.
Inventors include Dr. V. Karunakaran; and Vishnu Sagar V.
The application for the patent was published on August 21, 2026, under issue no. 34/2026.
Abstract: The proposed multimodal explainable artificial intelligence (AI) diagnostic system addresses the limitations of traditional single-modal medical imaging models by integrating chest X-ray imaging with patient clinical history and laboratory test data to achieve enhanced diagnostic accuracy and interpretability. The core innovation lies in its ability to perform multi-condition prediction with cross-modal reasoning, enabling early detection of complex and co-existing lung diseases such as tuberculosis, pneumonia, lung malignancy, and chronic obstructive pulmonary disease. The system utilizes a convolutional neural network (CNN)-based image processing module to extract high-level spatial features from chest X-ray images, while a transformer-based clinical data engine processes structured laboratory values and unstructured patient history. A multimodal fusion module employing attention-based mechanisms integrates these data streams to generate a unified diagnostic representation. This architecture enables the system to correlate imaging abnormalities with clinical indicators, thereby improving prediction reliability. A key feature of the system is its explainable AI (XAI) framework, which provides transparent and clinician-friendly outputs. These include Grad-CAM-based heatmaps for highlighting critical regions in X-ray images, SHAP-based attribution for identifying influential laboratory parameters, and structured textual reasoning that explains the basis of each prediction. This ensures that the system functions as an interpretable decision- support tool rather than a black-box model. Additionally, the system incorporates a clinician feedback loop, allowing medical professionals to validate, correct, or refine predictions, thereby enabling continuous learning and improvement of the model over time. The architecture is designed to support deployment in real-world healthcare environments, including resource-constrained settings, by optimizing computational efficiency and scalability. Overall, the proposed system enhances diagnostic accuracy, transparency, and clinical usability, providing a reliable and cost- effective solution for AI-assisted lung disease diagnosis.
Disclaimer: Curated by HT Syndication.
