Cancer detection using Image Analytics improves diagnosis accuracy by 96%

Cancer detection using Image Analytics improves diagnosis accuracy by 96%

A platform to analyse numerous CT scans, Xray images and other sources of clinical data to detect anomalies and help doctors arrive at an accurate diagnosis

About the Client

Kerala Institute of Medical Sciences (KIMS) is a multi-specialty tertiary care hospital group where a competent team of 900+ doctors and specialists and sophisticated technology come together to deliver high-quality medical aid across 18 hospitals. KIMS has emerged as one of the leading centers of pioneering medical work, research and academics in South India with a global outreach.

Business Impact
___

    • Accuracy of 96% for detecting lung nodules
    • Building and implementing deep learning at scale

Challenge
___

    • Limited availability of radiologists
    • A clinical quality assurance process for Lung CT Scan and Chest X-ray
    • Small nodules and nodules close to chest walls for CT Scans
    • Large variety of possible anomalies to be detected from chest X-rays

Solution Delivered
___

    • A deep learning (U-net) based image segmentation and localization solution 
    • Combination of public data and clinic specific data for building and testing the solution
    • Complete product platformthat can be used by radiologists for testing and using the solution
    • Capturing radiologist’s feedback and mining the same for annotating future data