Visual Natural Language Processing of Medical Images for Enhanced Value

Project Goal: Develop an image analysis tool to assist in the qualitative analysis of platelets and platelet activation.

Improvements in qualitative analysis of medical images, platelets in particular, are needed to reduce overhead and remove human bias, error and inefficiency.

Fractal Eyes V2.0 serves as a preliminary, neural-network-based platform that classifies platelets in different stages of activation, while it also extracts feature data for further analysis. Developed in Python, the tool captures the area of the platelet, the perimeter, color density differences, light intensity and approximate pixel length and width as it maps the image. A graphical user interface application loads platelet images, view feature data and access log data for image analysis and feature extraction subsystems.

This system will aid the University of Arizona Center for Accelerated Biomedical Innovation's current work on platelets and associated technologies, programs and goals. It serves as foundational work for expansion to other subjects of medical image analysis.