MUMBAI, India, June 30 -- Intellectual Property India has published a patent application (202641069043 A) filed by Kingston Engineering College; Vellore on June 02, 2026, for A Low Cost Edge Ai Device For Medical Image Based Diagnostics In Rural Environments.
Inventors include Dr. M. Vasumathy; and Mrs. M. Menaka.
The application for the patent was published on June 26, 2026, under issue no. 26/2026.
Abstract: The present invention discloses a low-cost edge artificial intelligence (AI) device for medical image-based giagnostics in rural environments, integrated with a Real- Time Explainable Intraoperative Imaging Feedback System (REIIFS) for enhanced interpretability and adaptive imaging. The device performs real-time analysis of medical images, including X-rays and ultrasound scans, using an embedded edge computing architecture without reliance on cloud infrastmcture. It comprises a system- on-chip (SoC) or hardware-accelerated module, preproce~sing unit, and optimized deep learning models capable of classification, detection, and segmentation with high accuracy and low latency. A key technical novelty lies in the integration of a dual-stream processing framework, wherein a raw image stream is transmitted with near- zero latency while a parallel AI stream generates real-time saliency maps and feature attribution overlays. The system further incorporates a hardware-in-the-loop adaptive control mechanism, enabling dynamic adjustment of imaging parameters such as focus, gain, and resolution based on detected regions of interest. Additionally, the invention employs resource-aware model optimization techniques, including quantization, pruning, and pipeline parallelism, ensuring efficient execution on low- power edge hardware. The combined architecture delivers explainable, real-time, and adaptive diagnostics in a portable, energy-efficient, and cost-effective platform, thereby improving clinical decision-making and expanding access to advanced medical imaging in rural and resource-limited settings.
Disclaimer: Curated by HT Syndication.
