Room: AAPM ePoster Library
Purpose: To 3D print and validate a clinically relevant, cost effective, and user-friendly phantom that can test and compare lesion detectability across cardiac nuclear imaging systems and protocols.
Methods: A novel left ventricle phantom was redesigned to be watertight and easy to score. The phantom incorporates six solid “low-perfusion lesions” of various volumes (0.2 to 3.3 cc) organized as contrast-detail objects along the axes of the ventricle. A rectangular container was designed to hold the ventricle at a clinical angle while simulating thoracic attenuation with background radioactivity. The phantom was printed with an entry-level printer using transparent polylactide and solid infill. Multiple cardiac SPECT acquisitions were used to demonstrate the functionality of the phantom. Injected Technetium-99m radioactivity created a ventricle to background concentration of 10:1. Matrix size, time per projection, and acquisition zoom were independently changed from a reference protocol. The reference protocol (scan 1) used a 128x128 matrix, zoom of 1, and 20 seconds per projection. Scan 2 altered the matrix size to 64x64. Scan 3 increased time per projection to 30 seconds and scan 4 increased the zoom to 1.33. Polar maps were reconstructed and used to score the detectability based on the number of visible lesions.
Results: For the reference scan, 2 lesions were visible. Scan 2 (64x64 matrix), scan 3 (30 second projections), and scan 4 (1.33x zoom) discerned 1, 3, and 4 lesions, respectively. Increasing time per projection and zoom led to more distinct lesions.
Conclusion: Improvements from a previous prototype have allowed for the construction of a clinically useful phantom for protocol optimization and potentially for device comparison. Improvements include no lesion overlap on polar maps and a more effective watertight design. The phantom also differentiates subtle changes in protocols, suggesting it to be a valuable tool for instrumentation QC and comparing clinical protocols.