Optimizing radiological workflow with automated diagnostics

Principal Investigator

Name
Mark-Jan Harte

Degrees
MSc

Institution
Aidence

Position Title
Researcher

Email
markjan@aidence.com

About this CDAS Project

Study
NLST (Learn more about this study)

Project ID
NLST-181

Initial CDAS Request Approval
Nov 25, 2015

Title
Optimizing radiological workflow with automated diagnostics

Summary
Radiologists currently must manually read every image; by introducing software tools to support them the quality of readings and thus the diagnostic reports is expected to increase. In order to create such software a large body of annotated images is required to train the algorithm the distinction between normal and pathological cases. Therefore the lung CT image set from the NLST is an important element in creating this software as well as for validating its clinical accuracy.

Aims

1) To create a machine learning algorithm that is able to detect lung nodules and determine their malignancy
2) To achieve a specificity and sensitivity that are better than existing CAD solutions

Collaborators

N/A