MRI-guided voxel-based automatic semi-quantification of dopamine transporter imaging

Purpose: Functional imaging with ¹²³I-FP-CIT SPECT suffers from poor spatial resolution resulting in partial-volume effect, which affects the subsequent semi-quantification. Definition of regions of interest for semi-quantification is further subject to user's experience and inter-observer variability. The aim of this work has been to develop an automatic method for definition of volumes of interest and partial-volume correction using patient-specific MRI and providing complete contrast recovery in striatal region.

Method: The method consists of spatial pre-processing (image segmentation and multi-modality registration), partial-volume correction (performed by region-based voxel-wise technique), and calculation of uptake indices in striatal structures. Anthropomorphic striatal phantom was used to optimize the method and to assess linearity, accuracy, and reproducibility. The method was tested on 58 patient datasets and compared with clinical assessment and BasGan software.

Results: The method works automatically. The output is highly linear regarding changing striatal uptake. Complete contrast recovery is achieved using 6.5 mm FWHM. Accuracy is better than 0.15 in terms of RMSE between measured and true uptake indices. Reproducibility is better than 5% for normal uptake ratio. The method outperformed clinical assessment in all measures. With patient data, it provided results closer to BasGan (RMSE 0.9) than to clinical assessment (RMSE 1.9) and fairly correlated with both.

Conclusion: The proposed method provides complete recovery of striatal contrast under given acquisition and reconstruction conditions. It reduces intra- and inter-observer variability, accurately defines volumes of interest, and effectively suppresses partial-volume effect. It can be reproduced using publicly available software.