Comparison of Image Quality of Breast-Specific Positron Emission Tomography: Insights from Phantom and Clinical Studies in a Japanese Multicenter Trial ​

Objective Breast-specific positron emission tomography (breast PET), including positron emission mammography (PEM) and dedicated breast PET (dbPET), provides high-resolution functional imaging for detecting small breast cancers. However, direct cross-system comparisons and acquisition protocol optimizations remain underexplored. This study aimed to directly compare the imaging performance of the opposed-type PEM, first-generation photomultiplier tube (PMT)-based dbPET (dbPET1), and second-generation silicon photomultiplier (SiPM)-based dbPET (dbPET2) using clinical imaging protocols, and determine the requisite acquisition conditions for achieving comparable depiction of breast lesions across systems. Methods A cylindrical phantom with four spheres (diameter: 3–10 mm) was prepared with sphere-to-background ratios (SBRs) of 2:1, 4:1, and 8:1, based on clinical images. The phantom was scanned for 10 min in list mode with the spheres at the center and periphery of each detector and reconstructed at 1–10 min. Visual and quantitative evaluations were performed using the coefficient of variation of the background (CV _BG ), detection index (DI), and contrast recovery coefficient (CRC). Representative clinical images of three lesion types, viz. mass-like uptake near the nipple, mass-like uptake close to the chest wall, and non-mass uptake, were also assessed using visual evaluation and the tumor-to-background ratio (TBR). Results Phantom images with SBRs of 2:1 and 4:1 did not sufficiently visualize the small spheres; therefore, an 8:1 ratio was chosen for the analysis. dbPET was capable of visualizing smaller spheres compared with PEM. At the periphery, image quality was reduced for all systems, while all systems were able to identify spheres ≥ 7.5 mm in diameter at a contrast ratio of 1:8 under clinical imaging protocols. The DI decreased with shorter acquisition time, while the CRC remained relatively stable. The CV _BG increased, especially in dbPET2. Clinical evaluation confirmed that clarified the minimum acquisition times required to ensure adequate diagnostic image quality for different breast PET systems (≥ 5 min for dbPET, ≥ 7 min for PEM). dbPET provided superior detectability, whereas PEM had advantages near the chest wall. TBR analysis supported the consistency between the results of evaluation of the phantom and patients. Conclusions This study demonstrated that all breast-specific PET systems can achieve image quality capable of identifying sub-centimeter lesions within clinically feasible scan times (5 min for dbPET, 7 min for PEM). These findings provide the foundation for harmonizing protocols across systems and optimizing their clinical application in breast cancer diagnosis.