Impact of CT Exposure Characteristic on Irreversible Compressibility and its Effect on Classification into Benign and Malignant Categories
Principal Investigator
Name
David Clunie
Degrees
MBBS
Institution
PixelMed
Position Title
Independent Consultant
Email
About this CDAS Project
Study
NLST
(Learn more about this study)
Project ID
NLST-123
Initial CDAS Request Approval
Mar 2, 2015
Title
Impact of CT Exposure Characteristic on Irreversible Compressibility and its Effect on Classification into Benign and Malignant Categories
Summary
Practical deployment of large scale lung cancer screening, particularly in remote areas and developing countries, may depend on the use of irreversible compression prior to interpretation. Such compression is known to have an impact on human and machine performance depending on type and degree. Compression effectiveness is impacted by image noise which is in turn dictated by exposure parameters.
This project is to examine the NLST image data set and accompanying outcomes data to identify significant factors related to compressibility and screening performance.
This project is to examine the NLST image data set and accompanying outcomes data to identify significant factors related to compressibility and screening performance.
Aims
Characterize and stratify image sets based on CT exposure characteristics. Perform lossy compression with various schemes (including JPEG, JPEG 2000 and near-lossless JPEG-LS) and bit rates. Derive metrics of compression impact and evaluate with respect to impact on human and machine classification performance and correspondence with outcomes (biopsy, recurrence, mortality).
Collaborators
None