Deep learning-based susceptibility map-weighted imaging analysis of dorsal nigral hyperintensity as a surrogate for dopamine transporter positron emission tomography in patients with idiopathic rapid eye movement sleep behavior disorder

Abstract

Idiopathic REM sleep behavior disorder (iRBD) is a prodromal stage of alpha-synucleinopathies, and accurate identification of early dopaminergic dysfunction is clinically important. This study evaluated whether deep learning–based susceptibility map-weighted imaging (SMwI) can predict dopamine transporter (DAT) positron emission tomography (PET) abnormalities in patients with iRBD. We retrospectively included 74 patients who underwent both SMwI and 18F-FP-CIT PET and applied a deep learning model to detect the absence of dorsal nigral hyperintensity. Diagnostic performance was assessed at patient and hemisphere levels. Deep learning–based SMwI showed a sensitivity of 78%, specificity of 66.7%, and accuracy of 73% for predicting DAT PET abnormalities. Bilateral SMwI abnormalities strongly correlated with DAT PET positivity (24/26; 92.3%), whereas DAT PET abnormalities were present in 35.5% of patients with visually unremarkable SMwI. These findings indicate that AI-assisted SMwI provides moderate agreement with DAT PET and may serve as a practical triage or follow-up tool for identifying presynaptic dopaminergic dysfunction in iRBD, particularly where access to DAT PET is limited.