John Pepper

Ph.D.

National Cancer institute, Division of Cancer Prevention

Biologist

NLST (Learn more about this study)

NLST-1194

Feb 14, 2024

Automated Machine Learning System for Analysis and Interpretation of Lung CT images

Especially in the wake of the NLST trial, there is a growing need for cancer prevention research to alleviate the bottleneck of image interpretation by trained specialists. There has recently been promising progress toward this by using automated machine learning systems for image analysis and interpretation. A recent paper reported an image analysis pipeline for classifying lung CT images with accuracy and sensitivity of 99.09% and 98.33%, respectively. (Nitha & Chandra 2023). Our goal is to evaluate, adapt, and replicate such a system using NCI computational resources and North American data . We plan to carry this out using NLST images and clinical data. Success could help open the way to leveraging the successes of the NLST trial into a greatly scaled -up program of cancer screening available to all who need it. We are pursuing this goal through a collaboration between NCI’s Division of Cancer Prevention, and its Computational Genomics & Bioinformatics Branch.


Nitha, V. R. and SSV Chandra (2023). "ExtRanFS: An Automated Lung Cancer Malignancy Detection System Using Extremely Randomized Feature Selector." Diagnostics 13(13).

-- Evaluate, adapt, and replicate a reported system for automated machine learning systems for image analysis and interpretation of lung CT images, using NCI computational resources and North American data.

-- Open the way to leveraging the successes of the NLST trial into a greatly scaled -up program of cancer screening available to all who need it.

Two groups at NCI:

- the Biometry and Data Analytics Branch of the Division of Cancer Biology
and,
- the Computational Genomics & Bioinformatics Branch of the Center for Biomedical Informatics and Information Technology (CBIIT)