Category: Technology
Objective: We made the diagnostic model using a deep learning algorithm based on dopamine transporterWe made the diagnostic model using a deep learning algorithm based on dopamine transporter imaging (DatSCAN) and investigate the distinguishing value of Parkinson’s disease (PD) and atypical Parkinsonism (APS). imaging (DatSCAN) and investigate the distinguishing value of Parkinson’s disease (PD) and atypical Parkinsonism (APS).
Background: Although the clinical criteria of PD and APS are well established, it has been notably challenging to differentiate between PD and APS. DatSCAN is currently used for the diagnosis of PD. However, the diagnostic value for discriminating between PD and APS is limited. Here, we made the diagnostic model using a deep learning algorithm based on DatSCAN data and investigate the distinguishing value of PD from APS.
Method: We analyzed the DatSCAN image of 1094 subjects, including 478 PD, 433 normal, and 73 APS (39 Multiple System Atrophy (MSA), 34 Progressive Supranuclear Palsy (PSP). A multistage approach was designed where the first stage performed the detection of Normal, PD, and APS and the second stage performed the detection of MSA and PSP. For both stages, a 3D Convolutional Neural Network was used which internally shared the Internal Parameters (Weights and Biases) from the first stage to the second stage. After the internal validation, we tested 266 samples in other center for external validation.
Results: In internal validation, the sensitivity is 0.96, the specificity of 0.95 and accuracy is 0.96 The results from external validation showed that sensitivity is 0.85, specificity is 0.96 and accuracy is 0.66.
Conclusion: This is the first report of a deep learning algorithm based on DatSCAN for distinguishing PD from APS. Our diagnostic model can be a useful and helpful tool for early diagnosis of PD and APS.
To cite this abstract in AMA style:
JS. Park, JY. Gwak. Distinguishing Parkinson’s disease from atypical Parkinsonian syndrome using multistage deep learning based on dopamine transporter imaging [abstract]. Mov Disord. 2024; 39 (suppl 1). https://www.mdsabstracts.org/abstract/distinguishing-parkinsons-disease-from-atypical-parkinsonian-syndrome-using-multistage-deep-learning-based-on-dopamine-transporter-imaging/. Accessed June 15, 2025.« Back to 2024 International Congress
MDS Abstracts - https://www.mdsabstracts.org/abstract/distinguishing-parkinsons-disease-from-atypical-parkinsonian-syndrome-using-multistage-deep-learning-based-on-dopamine-transporter-imaging/