The clinical application of nigrosome 1 detection on high-resolution susceptibility-weighted imaging in the evaluation of suspected Parkinsonism: The real-world performance and pitfalls

Abstract

Purpose To evaluate the real-world diagnostic performance of high-resolution susceptibility-weighted imaging (HR-SWI) and investigate whether the reader's predictions can be used to find cases where HR-SWI finding and final clinical diagnosis matched. Methods This retrospective study enrolled patients with suspected Parkinsonism (n = 48) or volunteers with other intracranial pathologies (n = 31) who underwent brain magnetic resonance imaging (MRI) including HR-SWI, which was used to evaluate nigrosome 1 (NG1). All patients with suspected Parkinsonism underwent N-3-fluoropropyl-2-carbomethoxy-3-4-iodophenyl nortropane (FP-CIT) positron emission tomography and a clinical diagnosis was made by a neurologist. The HR-SWI data were qualitatively analyzed by two independent reviewers. A consensus reading was performed and a diagnostic confidence score was assigned. According to final clinical diagnosis, diagnostic sensitivity, specificity, and accuracy were calculated. Receiver operating characteristic (ROC) curve analysis was used to examine whether the diagnostic confidence score could be used to identify HR-SWI finding-final clinical diagnosis matched cases. Results Of the 48 patients with suspected Parkinsonism, 31 were diagnosed with idiopathic Parkinson's disease, and three with multiple system atrophy. The remaining 14 patients were included in the disease control group. Of the 31 volunteers, 10 subjects were excluded due to possibility of nigrostriatal degeneration and finally 21 subjects were enrolled as controls with non-Parkinsonism pathology (non-PD control). After consensus reading, 25 subjects were classified as true positive and 28 as true negative, according to HR-SWI findings. The calculated diagnostic sensitivity, specificity, and accuracy were 73.5%, 80.0%, and 76.8%, respectively. With using diagnostic concordance score, the area under the ROC curve for the detection of concordance case was 0.83 (95% CI: 0.72-0.91, p < 0.05). Conclusion The diagnostic performance of NG1 detection using HR-SWI with 3T MRI was within acceptable range. Using the reader's diagnostic confidence could be helpful to find cases which HR-SWI finding and final clinical conclusion match. So HR-SWI may be of added value in the evaluation of suspected Parkinsonism.