Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Symmetric Regressor for MRI-Based Assessment of Striatal Dopamine Transporter Uptake in Parkinson's Disease With Enhanced Uncertainty Estimation

Walid Abdullah Al, Il Dong Yun, Yun Jung Bae

TL;DR

The paper addresses the need for safer, MRI-based assessment of striatal dopamine transporter uptake in Parkinson's disease by proposing a symmetric regressor that predicts both right and left SBRs from paired nigral patches. It leverages lateral symmetry with a shared regressor and a symmetric loss that enforces similarity between hemispheres, and it introduces a symmetric Monte-Carlo dropout scheme to produce reliable uncertainty intervals. On a dataset of 734 nigral patches from 367 patients, the symmetric model achieves improved predictive performance (MAE up to $0.6741$, $R=0.7291$) compared with baselines, and provides more accurate and sharper prediction intervals (e.g., 95\% CP with width $0.251$, vs $0.281$ for standard MC). These results demonstrate enhanced explainability, stronger cross-hemisphere learning, and practical potential for safer MRI-based PD monitoring in clinical settings.

Abstract

Dopamine transporter (DAT) imaging is commonly used for monitoring Parkinson's disease (PD), where striatal DAT uptake amount is computed to assess PD severity. However, DAT imaging has a high cost and the risk of radiance exposure and is not available in general clinics. Recently, MRI patch of the nigral region has been proposed as a safer and easier alternative. This paper proposes a symmetric regressor for predicting the DAT uptake amount from the nigral MRI patch. Acknowledging the symmetry between the right and left nigrae, the proposed regressor incorporates a paired input-output model that simultaneously predicts the DAT uptake amounts for both the right and left striata. Moreover, it employs a symmetric loss that imposes a constraint on the difference between right-to-left predictions, resembling the high correlation in DAT uptake amounts in the two lateral sides. Additionally, we propose a symmetric Monte-Carlo (MC) dropout method for providing a fruitful uncertainty estimate of the DAT uptake prediction, which utilizes the above symmetry. We evaluated the proposed approach on 734 nigral patches, which demonstrated significantly improved performance of the symmetric regressor compared with the standard regressors while giving better explainability and feature representation. The symmetric MC dropout also gave precise uncertainty ranges with a high probability of including the true DAT uptake amounts within the range.

A Symmetric Regressor for MRI-Based Assessment of Striatal Dopamine Transporter Uptake in Parkinson's Disease With Enhanced Uncertainty Estimation

TL;DR

The paper addresses the need for safer, MRI-based assessment of striatal dopamine transporter uptake in Parkinson's disease by proposing a symmetric regressor that predicts both right and left SBRs from paired nigral patches. It leverages lateral symmetry with a shared regressor and a symmetric loss that enforces similarity between hemispheres, and it introduces a symmetric Monte-Carlo dropout scheme to produce reliable uncertainty intervals. On a dataset of 734 nigral patches from 367 patients, the symmetric model achieves improved predictive performance (MAE up to , ) compared with baselines, and provides more accurate and sharper prediction intervals (e.g., 95\% CP with width , vs for standard MC). These results demonstrate enhanced explainability, stronger cross-hemisphere learning, and practical potential for safer MRI-based PD monitoring in clinical settings.

Abstract

Dopamine transporter (DAT) imaging is commonly used for monitoring Parkinson's disease (PD), where striatal DAT uptake amount is computed to assess PD severity. However, DAT imaging has a high cost and the risk of radiance exposure and is not available in general clinics. Recently, MRI patch of the nigral region has been proposed as a safer and easier alternative. This paper proposes a symmetric regressor for predicting the DAT uptake amount from the nigral MRI patch. Acknowledging the symmetry between the right and left nigrae, the proposed regressor incorporates a paired input-output model that simultaneously predicts the DAT uptake amounts for both the right and left striata. Moreover, it employs a symmetric loss that imposes a constraint on the difference between right-to-left predictions, resembling the high correlation in DAT uptake amounts in the two lateral sides. Additionally, we propose a symmetric Monte-Carlo (MC) dropout method for providing a fruitful uncertainty estimate of the DAT uptake prediction, which utilizes the above symmetry. We evaluated the proposed approach on 734 nigral patches, which demonstrated significantly improved performance of the symmetric regressor compared with the standard regressors while giving better explainability and feature representation. The symmetric MC dropout also gave precise uncertainty ranges with a high probability of including the true DAT uptake amounts within the range.
Paper Structure (28 sections, 10 equations, 6 figures, 5 tables, 1 algorithm)

This paper contains 28 sections, 10 equations, 6 figures, 5 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Works
  3. Standard DAT Uptake Assessment
  4. DAT Uptake Prediction using MRI
  5. Deep Learning Based Prediction
  6. Methodology
  7. Data
  8. Deep Regressor
  9. Symmetric Regressor
  10. Identical Input Type
  11. Correlated Output
  12. Proposed Model
  13. Symmetric Uncertainty in Prediction
  14. Monte-Carlo Dropout:
  15. Symmetric MC dropout:
  16. ...and 13 more sections

Figures (6)

  • Figure 1: The proposed symmetric regressor model. Both the right and left SBRs are predicted simultaneously. Besides the regular MSE loss w.r.t. the true SBRs, the symmetric loss i.e., MSE between the right and left predictions is also minimized.
  • Figure 2: R-R plot of our right and left SBRs showing a correlation of $0.93$.
  • Figure 3: Dependency on the clipping margin (left) and weight (right) of the symmetric loss. The clipping margin in the left plot is normalized relative to the maximum SBR ($6.84$) of our dataset. The weight axis in the right plot is in log scale.
  • Figure 4: Sharpness-CP curves for the prediction intervals of different methods.
  • Figure 5: t-SNE visualization of the extracted features by the standard and proposed regressors. Each point is color-graded based on its corresponding true SBR value.
  • ...and 1 more figures