Minimum lesion detectability as a measure of PET system performance

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Radiology

Background:

• Developed a phantom and imaging protocol to assess small lesion detection limits on PET systems.
• Utilized seven spheres (3.95-15.43 mm) in varied contrast conditions (15:1 to 1.88:1).

Purpose of the Study:

• To measure the limit of small lesion detection on different PET systems.
• To compare the performance of multiple clinical PET/CT scanners.

Main Methods:

• Imaged spheres on Philips Gemini TF/Vereos, Siemens mCT/HRRT, and GE Discovery 710 PET/CT systems.
• Varied imaging times (1-16 min) and reconstructed images with 1-4 mm voxel sizes.
• Conducted a human observer study to determine the smallest observable sphere.

Main Results:

• All systems detected the smallest sphere (3.95 mm) at 15:1 contrast with 16 min scan time.
• At typical scan times (2-4 min), detected spheres ranged from 4.95-6.23 mm (15:1 contrast) to 12.43-15.43 mm (1.88:1 contrast).
• System performance was consistent with Rose criteria for lesion detectability.

Conclusions:

• Current clinical PET/CT systems exhibit similar small lesion detection abilities.
• Scan time is critical for detecting small lesions with low contrast uptake (around 2:1).
• Recommends longer scan times in regions suspected of small lesions to improve detection.