HDgtl Pattern analytics under development enables pattern recognition on Ultrasound Obstetric scan images / Breast Screen - Thermal Images leveraging a cutting-edge open source platform developed by UC Santa Barbara. Bisque (Bio-Image Semantic Query User Environment) – stores, visualizes, organizes, and analyzes images of any format (X-Ray, MRI, Ultrasound, Angiogram, Thermal Images etc.) for up to 5-dimension space. Bisque satisfies all compliance rules for Data Storage and Processing and is HIPAA compliant as well. Image analytics is carried out by using libraries of python, viz. MATLAB, PIL functions. Images are then mapped to treatment prediction using the Clustering Algorithms of Naïve Bayes. The predictions are then submitted to the Surgeon / Radiologist / Physician for their professional opinion. Suggested diagnosis is sent to the cloud storage that tags the information masked, using the hashing algorithm and updates the origin with the treatment approach to be advised for the patient. Our Self Learning model incorporates these inputs and updates the system.
Analytics techniques like classification, feature extraction, etc. are applied on the image to have a basic understanding of the image that is received from the front-end app to the analytic tool. Once, the machines recognize the patterns, the CNN technique matches it with the training database and annotates the image. It then compares the annotations to interpret the treatment required for the image. The HDgtl Medical Advisory Board along with our Domain Specialists would provide guidance on marker thresholds in regions of interest to be identified and reported.