Automatic detection of abnormalities on frontal Chest X-Ray

Principal Investigator

Name
Eldad Elnekave

Degrees
MD

Institution
Zebra Medical Vision

Position Title
Chief Medical Officer

Email
eldad@zebra-med.com

About this CDAS Project

Study
PLCO (Learn more about this study)

Project ID
PLCO-232

Initial CDAS Request Approval
Sep 30, 2016

Title
Automatic detection of abnormalities on frontal Chest X-Ray

Summary
This project aims to generate a completely automatic algorithm which identifies abnormal opacities and lucencies on frontal chest radiographs. The opacities may represent a spectrum of clinical diagnoses including but not limited to pneumonia, granuloma, mass, atalectasis, pneumothorax, pneumomediastium and pulmonary edema. We aim to have at least 10,000 samples of each relevant pathology and approximately 40,000 normal CXR examples. The images will be used in an iterative machine learning process based upon convolutional neural network technique.

Aims

To automatically detect acute and chronic pathology on frontal chest radiographs with high sensitivity and specificity.

Collaborators

The project is at entirely internal to Zebra-medical vision.