Quantification of GABAA receptors in the rat brain with [123I]Iomazenil SPECT from factor analysis-denoised images

Purpose: In vivo imaging of GABAA receptors is essential for the comprehension of psychiatric disorders in which the GABAergic system is implicated. Small animal SPECT provides a modality for in vivo imaging of the GABAergic system in rodents using [123I]Iomazenil, an antagonist of the GABAA receptor. The goal of this work is to describe and evaluate different quantitative reference tissue methods that enable reliable binding potential (BP) estimations in the rat brain to be obtained. Methods: Five male Sprague–Dawley rats were used for [123I]Iomazenil brain SPECT scans. Binding parameters were obtained with a one-tissue compartment model (1TC), a constrained two-tissue compartment model (2TCc), the two-step Simplified Reference Tissue Model (SRTM2), Logan graphical analysis and analysis of delayed-activity images. In addition, we employed factor analysis (FA) to deal with noise in data. Results: BPND obtained with SRTM2, Logan graphical analysis and delayed-activity analysis was highly correlated with BPF values obtained with 2TCc (r = 0.954 and 0.945 respectively, p < 0.0001). Equally significant correlations were found between values obtained with 2TCc and SRTM2 in raw and FA-denoised images (r = 0.961 and 0.909 respectively, p < 0.0001). Scans of at least 100 min are required to obtain stable BPND values from raw images while scans of only 70 min are sufficient from FA-denoised images. These images are also associated with significantly lower standard errors of 2TCc and SRTM2 BP values. Conclusion: Reference tissue methods such as SRTM2 and Logan graphical analysis can provide equally reliable BPND values from rat brain [123I]Iomazenil SPECT. Acquisitions, however, can be much less time-consuming either with analysis of delayed activity obtained from a 20-minute scan 50 min after tracer injection or with FA-denoising of images