Validation of pre-trained deep learning algorithms on the PLCO Lung dataset

Principal Investigator

Name
Ammar Jagirdar

Degrees
M.Tech

Institution
Qure.ai

Position Title
Product Manager

Email
ammar.jagirdar@qure.ai

About this CDAS Project

Study
PLCO (Learn more about this study)

Project ID
PLCO-484

Initial CDAS Request Approval
May 31, 2019

Title
Validation of pre-trained deep learning algorithms on the PLCO Lung dataset

Summary
Most deep learning algorithms tend to perform well when tested on data similar to the training set. The PLCO dataset is unique and previously unseen and serves as a valuable archive to test our pre trained models against radiologist opinions reported. Our project would aim to test the performance of deep learning algorithms trained to identify different findings.

Aims

- Testing for generalisation of multiple findings on the Chest X-ray
- Evaluating multiple findings on the Chest X-ray by measuring AUC, sensitivity and specificity

Collaborators

None