MUMBAI, India, Aug. 12 -- Intellectual Property India has published a patent application (202631050265 A) filed by National Institute Of Technology Rourkela Nit on April 20, 2026, for Embedded Deep Learning-Based System For Automated Analysis Of Microscopic Blood Cell Images.
Inventors include Chatterjee, Saptarshi; Das, Rishi; Kushwaha, Kunal; Parashar, Sparsh; and Sarangi, Hrushikesh.
The application for the patent was published on August 07, 2026, under issue no. 32/2026.
Abstract: Abstract EMBEDDED DEEP LEARNING-BASED SYSTEM AND METHOD FOR AUTOMATED ANALYSIS OF MICROSCOPIC BLOOD CELL IMAGES Provided is an embedded image analysis system (100) for automated analysis of microscopic blood cell images. The system (100) comprises an imaging module (102) configured to acquire one or more microscopic blood cell images, an embedded processing unit (106), and a user interface (118). The embedded processing unit (106) is configured to preprocess the acquired microscopic blood cell images to generate standardized image data, generate feature tokens using a convolutional token embedding model, process the generated feature tokens through one or more convolution transformer blocks, generate channel attention weights to recalibrate corresponding feature channels, and generate one or more classification results. The classification results are displayed through the user interface (118). The embedded processing unit (106) performs the image analysis locally without dependence on external computing infrastructure, enabling rapid and cost-effective screening of Acute Lymphoblastic Leukemia (ALL).
Disclaimer: Curated by HT Syndication.
