Post-reconstruction enhancement of [(18)F]FDG PET images with a convolutional neural network

BACKGROUND: The aim of the study was to develop and test an artificial intelligence (AI)-based method to improve the quality of [(18)F]fluorodeoxyglucose (FDG) positron emission tomography (PET) images. METHODS: A convolutional neural network (CNN) was trained by using pairs of excellent (acquisition time of 6 min/bed position) and standard (acquisition time of 1.5 min/bed position) or sub-standard (acquisition time of 1 min/bed position) images from 72 patients. A test group of 25 patients was used to validate the CNN qualitatively and quantitatively with 5 different image sets per patient: 4 min/bed position, 1.5 min/bed position with and without CNN, and 1 min/bed position with and without CNN. RESULTS: Difference in hotspot maximum or peak standardized uptake value between the standard 1.5 min and 1.5 min CNN images fell short of significance. Coefficient of variation, the noise level, was lower in the CNN-enhanced images compared with standard 1 min and 1.5 min images. Physicians ranked the 1.5 min CNN and the 4 min images highest regarding image quality (noise and contrast) and the standard 1 min images lowest. CONCLUSIONS: AI can enhance [(18)F]FDG-PET images to reduce noise and increase contrast compared with standard images whilst keeping SUV(max/peak) stability. There were significant differences in scoring between the 1.5 min and 1.5 min CNN image sets in all comparisons, the latter had higher scores in noise and contrast. Furthermore, difference in SUV(max) and SUV(peak) fell short of significance for that pair. The improved image quality can potentially be used either to provide better images to the nuclear medicine physicians or to reduce acquisition time/administered activity.