Abstract:
Synthetic CT images from modified Dixon (mDixon) MR data are proposed. Attenuation correction (AC) is performed on abdomen and pelvis PET data using synthetic CT. AC is needed in PET/MR systems to be quantitatively accurate and meet qualification standards for many multi-center trials.
MR does not contain photon attenuation information. Existing MR-based synthetic CT generation methods either use advanced MR sequences with long acquisition time and limited clinical availability or match MR images from a newly scanned subject to images in a library of MR-CT pairs, which has difficulty accounting for the diversity of human anatomy, especially in pathological patients.
A five-phase interlinked method using mDixon MR acquisition and advanced machine learning for synthetic CT generation addresses these issues. TFC and TFC-ALC are used. Our work is fourfold: 1) TFC-ALC generates better synthetic CT than Dixon-based scanning on the difficult abdomen. 2) Transfer learning divides MR voxels into four groups: fat, bone, air, and soft tissue.
ALC can learn insightful classifiers using as few but informative labeled examples to distinguish bone, air, and soft tissue. The TFC-ALC method overcomes CT and MR image co-registration imperfections and uncertainty by combining them. 3) TFC-ALC has better synthetic CT generation and parameter robustness than other methods, making it clinically feasible. The mean absolute pre…
Note: Please discuss with our team before submitting this abstract to the college. This Abstract or Synopsis varies based on student project requirements.
Did you like this final year project?
To download this project Code with thesis report and project training... Click Here
