Effects of image matrix on quantitative metrics in ⁶⁸Ga DOTATATE studies: Changes in SUV and signal-to-noise ratio in modern digital PET detectors

Objectives: Modern digital PET detectors can dramatically improve PET sensitivity and spatial resolution. As novel theranostic agents for prostate cancer, fibroblast activating protein inhibitor, and other targets continue towards clinical development, an opportunity arises for improving patient care through improving small lesion detectability. With increasing digital PET sensitivity and spatial resolution, improving small lesion detectability by improving acquisition parameters such as image matrix size is now feasible. The purpose of this study is to quantify the effect of matrix size in ⁶⁸Ga DOTATATE PET scans on SUV and signal-to-noise ratio (SNR) using a modern digital PET system.

Methods: Ten ⁶⁸Ga DOTATATE patients were imaged using a Discovery MI (General Electric, Waukesha WI) PET/CT system. All scans were performed with 5 minutes per bed position using the QClear reconstruction tool with B = 400. Images were reconstructed at the following matrix sizes: 192 x 192, 256 x 256, and 384 x 384 correlating to 3.6, 2.7 and 1.8 mm. To evaluate the effects of matrix size on lesion SUV and SNR, regions of interest (ROI) were created over known tumors (n=66) using PET edge plus (MIM v7.0) by an experienced nuclear medicine radiologist. These ROIs were then superimposed on all reconstructed image sets, and SUV_max and SUV_mean were recorded for each tumor. To evaluate SNR, ROIs were placed in the blood pool (aorta) and liver to measure background SUV_mean and standard deviation for hepatic and extrahepatic tumors. Care was taken to ensure the ROIs avoided tissue boundaries and areas of heterogeneous uptake. SNR for each tumor was calculated as the difference in SUV_mean in the tumor to SUV_mean in the background, divided by the standard deviation of the background. Additionally, tumors were divided into two groups: small tumors (< 1.0 cm diameter, n = 36) and large tumors (≥ 1.0 cm diameter, n = 30). To determine if differences in uptake metrics or SNR were statistically significant, a series of paired t-tests were performed.

Results: For both small and large lesions, increasing the matrix size led to a small reduction in SUV_max. For large lesions, the average SUV_max was 44.8, 44.0, and 42.8 for the 192, 256, and 384 matrix sizes, respectively. For small lesions, the average SUV_max was and 17.6, 17.2, and 16.9 for the 192, 256, and 384 matrix sizes, respectively. These differences, however, were not found to be statistically significant. For SUV_mean, increasing the matrix size from 192 to either 256 or 384 lead to a very small but statistically significant increase (p<0.05). SUV_mean differences between the 256 and 384 matrix sizes were not significantly different. The SNR, however, showed significant differences between matrix sizes. For large lesions, increasing matrix size led to a statistically significant increases in average SNR values of 124.3, 128.6, and 137.2 for 192, 256, and 384 matrix sizes. For small lesions, increasing matrix size led to statistically significant increases in the SNR values of 82.7, 89.2, and 91.7 for 192, 256, and 384 matrix sizes, respectively.

Conclusions: For clinically acquired ⁶⁸Ga DOTATATE PET, SNR increases as matrix size is increased. SUV_mean showed only small changes, and SUV_max was not significantly affected by matrix size. Changing matrix size suggests a potential to influence lesion detectability, but requires further investigation.