MUMBAI, India, May 29 -- Intellectual Property India has published a patent application (202641064442 A) filed by Pokkunuri Pardha Saradhi; Chilakala Lokanath Reddy; M V D Prasad; and Koneru Lakshmaiah Education Foundation, Vaddeswaram, Andhra Pradesh, on May 21, for 'an intelligent system and method for lung cancer detection and classification using multi-scale deep convolutional neural network with adaptive feature fusion on medical imaging data.'
Inventor(s) include Chilakala Lokanath Reddy; M V D Prasad; and P. Pardhasaradhi.
The application for the patent was published on May 29, under issue no. 22/2026.
According to the abstract released by the Intellectual Property India: "The present invention discloses a novel intelligent seven-layer system and method for early-stage lung cancer detection from CT scans and chest radiographs. The system comprises: (1) a multi-pathway DICOM/FHIR data ingestion layer with PACS integration; (2) an Adaptive Multi-Modal Normalisation pre-processing engine with NLM denoising, Lung Window normalisation, elastic augmentation, and 1mm voxel resampling; (3) a Modified U-Net with Depthwise Separable Convolutions and Dual Attention Gates (DSC=0.934); (4) a Multi-Scale Deep CNN with Adaptive Feature Fusion processing images at three parallel spatial resolutions; (5) an Ensemble Classification Module with Class-Balanced Focal Training achieving 96.4% sensitivity, 94.8% specificity, AUC=0.978, and 1.12 FP/scan; (6) a Grad-CAM++ Explainability Module reporting Lung-RADS 1-4X categories and quantitative nodule descriptors; and (7) an HL7 FHIR R4 Patient Reporting System with SHA-256 tamper-evident audit trail. The system is deployable on commodity hardware (8GB RAM, GTX 1060) and complies with India's Medical Devices Rules 2017, PDPA 2022, AERB 2004, and NDHM standards."
Disclaimer: Curated by HT Syndication.
