MUMBAI, India, May 1 -- Intellectual Property India has published a patent application (202641051252 A) filed by Nitte Meenakshi Institute Of Technology, Nitte; Ms. Pallavi Basavaraju; Dr. Ramachandra; Dr. Viswanatha Venkataravanappa; and Dr. Hunasavadi Siddaramaiah Jagadeesh, Bangalore, Karnataka, on April 22, for 'lung sound based diagnostic system for respiratory disease using hybrid cnn-gru deep learning model with mfcc feature extraction.'
Inventor(s) include Nitte Meenakshi Institute Technology, Nitte; Ms. Pallavi Basavaraju; Dr. Ramachandra; Dr. Viswanatha Venkataravanappa; and Dr. Hunasavadi Siddaramaiah Jagadeesh.
The application for the patent was published on May 1, under issue no. 18/2026.
According to the abstract released by the Intellectual Property India: "The present invention relates to a lung sound based diagnostic system for respiratory disease utilizing a hybrid Convolutional Neural Network (CNN) and Gated Recurrent Unit (GRU) deep learning model trained on Mel-Frequency Cepstral Coefficients (MFCCs) extracted from respiratory sounds. The system addresses limitations of traditional auscultation by providing automated, objective, and accurate classification of pulmonary conditions including Chronic Obstructive Pulmonary Disease (COPD), Pneumonia, Upper Respiratory Tract Infection (URTI), and Bronchiolitis. The system comprises a Data Acquisition Module for lung sound capture and database integration; a Preprocessing Module implementing noise reduction, segmentation, augmentation, and class balancing; an MFCC Feature Extraction Module with optimized five-stage pipeline (frame segmentation with 25ms frames and 50% overlap, FFT, Mel filter bank, log compression, and DCT); a Hybrid CNN-GRU Deep Learning Model combining spatial feature extraction through convolutional layers with temporal sequence learning through gated recurrent units; and a Diagnostic Output Module providing disease classification, confidence scores, and clinical reports. The hybrid CNN-GRU architecture effectively captures both spatial acoustic patterns in the frequency domain and temporal sequential dependencies in respiratory signals, achieving 93.2% classification accuracy with strong performance on COPD and Pneumonia detection. The system is deployable on PYNQ-Z2 FPGA hardware platform with ZYNQ XC7Z020 processor, enabling real-time lung sound analysis for clinical and point-of-care applications. By providing objective, automated respiratory disease diagnosis, the invention supports early detection, reduces dependence on subjective clinical interpretation, and enables accessible screening in resource-limited healthcare settings."
Disclaimer: Curated by HT Syndication.
