Dr. William Randazzo
AutoRad, LLC is a startup company, established in 2016, operating in the burgeoning medical artificial intelligence space. The field of radiology has been undergoing substantial change over the last decade: scan interpretation is becoming more centralized and the number of scans ordered continues to increase. This has increased the time pressure a radiologist faces to quickly and accurately diagnose images. In particular, radiologists have no guidance as to which scans should be read most urgently when they sit in front of a queue of unordered scans. We propose software which automatically “triages” unread head computed tomography (CT) scans. Our principal approach is to use a convolutional neural net (CNN)-based artificial intelligence (AI) method to train an interpreter using thousands of images we have acquired from a large teleradiology company based in the United States. This interpreter will then be able to triage head CT scans into urgent and nonurgent categories. The output will be used to rearrange pending, unread head CT scans in front of the radiologist so that he or she reads the most urgent scans first. This technology is exciting for high-volume radiology centers that are inundated with radiographs because it will facilitate decreased "time to read" for the most urgent scans. Later, AutoRad will shift into providing more granular predictive diagnosis (aka automated diagnosis) after it amasses enough training images to create sufficiently sensitive and specific performance.
Recipient of PennHealthX Ventures Seed Funding
Established LLC in Delaware
Formed partnership with teleradiology company for image acquisition and beta testing
Hired Machine Learning developer