3D printed phantom with 12 000 submillimeter lesions to improve efficiency in CT detectability assessment

Area of Science:

• Medical Imaging
• Radiological Physics
• Image Reconstruction

Background:

• Quantifying CT scanner detectability is challenging due to reconstruction nonlinearities.
• An efficient method sensitive to dose and scanner settings is needed.
• Previous work proposed a search challenge phantom with model observer detection.

Purpose of the Study:

• Fabricate an updated, full-size search challenge phantom.
• Evaluate lesion detectability using a nonprewhitening (NPW) model observer.
• Assess the phantom's sensitivity to dose variations.

Main Methods:

• A 3D-printed phantom with 12,000 submillimeter lesions was created.
• Scans were performed across a range of exposure levels (50-625 mAs).
• Lesion detectability was assessed using an NPW model observer and AUC metrics across different reconstruction algorithms (FBP, SAFIRE).

Main Results:

• Lesion detectability (AUC) increased with exposure levels for all tested reconstructions.
• Filtered-backprojection with Br59 kernel (FBP Br59) showed the best performance.
• The method demonstrated high sensitivity, detecting dose changes within 2% for FBP Br59.

Conclusions:

• The 3D-printed phantom and NPW model observer offer an efficient CT detectability assessment.
• The system is sensitive to subtle reconstruction effects and small dose variations.
• This approach enhances the precision of CT performance evaluation.