Performance of CT-guided spatial normalization for semi-quantification of dopamine transporter SPECT and detection of nigrostriatal degeneration

Abstract

Objective

The present study aims to assess the performance of a CT-guided spatial normalization method (CT-method) for the anatomical region-of-interest (ROI)-based semi-quantification of dopamine transporter (DAT) single photon emission computed tomography (SPECT) images and the detection of nigrostriatal degeneration as compared to an effective SPECT template-based method (MSPECT-method) and visual analysis performed by an expert reader.

Methods

Patients who underwent [¹²³I]FP-CIT SPECT/CT in the Hospices Civils de Lyon between 2008 and 2018 for clinically uncertain parkinsonian syndromes were included. The proposed CT-method aimed to spatially normalize the jointly acquired CT scans and apply the deformation fields to the coregistered SPECT images. It was compared to an effective SPECT template-based method using multiple templates as target for the spatial normalization (MSPECT-method). The distribution of specific binding ratios (SBR) was compared between both methods and the SBR classifications were compared to an expert’s visual classification of the scans, which served as the reference.

Results

Overall, 1156 patients (mean age ± SD = 68.7 ± 11.5; 52.6% male) were included. The CT-method provided a good separation between the normal and reduced SBR, with a higher effect size of the distance between the Gaussians (3.31 vs 3.11) and smaller overlap (6.44% vs 8.96%) compared to the MSPECT-method. Both the CT-method and MSPECT-method demonstrated high classification accuracy (96.7%, 95% CI: 95.7-97.7% vs. 94.6%, 95% CI: 93.3-95.9%), sensitivity (96.0%, CI: 94.3-97.7% vs. 89.7%, CI: 87.1-92.3%), and specificity (97.3%, CI: 96.1-98.6% vs. 98.7%, CI: 97.9-99.6%), respectively.

Conclusions

The proposed CT-guided spatial normalization method for automated semi-quantitative [¹²³I]FP-CIT SPECT analysis is a viable option when CT images are available. It offers objective spatial normalization and provides high accuracy for the detection of nigrostriatal degeneration, closely aligning with an expert’s visual interpretation.