What question did this study set out to answer?

Develop and evaluate an interpretable deep learning framework using T1-weighted MRI scans for automated Parkinson's disease diagnosis.

August 15, 2026AlgorithmsOpen Access

Bridging Accuracy and Interpretability: Explainable Deep Learning for Parkinson’s Disease Diagnosis from MRI

View Full Paper

Ask AI

Bookmark

Authors

IIIoana-Teodora IsarUniversitatea Națională de Știință și Tehnologie Politehnica București NPNirvana PopescuUniversitatea Națională de Știință și Tehnologie Politehnica București

Discussion

Member takes

Implication

Machine learning study demonstrates automated detection of Parkinson's disease from brain MRI scans, suggesting explainable AI aids clinical interpretability despite performance variability.

Key Points

Develop and evaluate an interpretable deep learning framework using T1-weighted MRI scans for automated Parkinson's disease diagnosis.
Applied Contrast Limited Adaptive Histogram Equalization (CLAHE) to T1-weighted MRI scans from the NTUA dataset and used SMOTE on extracted deep feature vectors to correct class imbalance.
Trained and evaluated multiple convolutional neural networks across six patient-wise train-test splits, including a focused experiment on axial MRI slices from 50 subjects.
Generated Grad-CAM visual heatmaps to identify anatomical regions influencing model predictions.
Model diagnostic performance varied substantially across subject-wise splits, demonstrating high peak metrics on select configurations but a conservative aggregate baseline.
Grad-CAM visualizations confirmed that deep learning models focused primarily on central brain structures linked to neurodegeneration with minimal attention to non-diagnostic areas.

Cite This Study

Isar et al. (2026) studied this question.

synapsesocial.com/papers/6a801a0e75c2e31742c8666a https://doi.org/10.3390/a19080680

View Full Paper

Ask AI

Bookmark

August 15, 2026AlgorithmsOpen Access

Bridging Accuracy and Interpretability: Explainable Deep Learning for Parkinson’s Disease Diagnosis from MRI

View Full Paper

Ask AI

Bookmark

Authors

IIIoana-Teodora IsarUniversitatea Națională de Știință și Tehnologie Politehnica București NPNirvana PopescuUniversitatea Națională de Știință și Tehnologie Politehnica București

Discussion

Member takes

Implication

Machine learning study demonstrates automated detection of Parkinson's disease from brain MRI scans, suggesting explainable AI aids clinical interpretability despite performance variability.

Key Points

Develop and evaluate an interpretable deep learning framework using T1-weighted MRI scans for automated Parkinson's disease diagnosis.
Applied Contrast Limited Adaptive Histogram Equalization (CLAHE) to T1-weighted MRI scans from the NTUA dataset and used SMOTE on extracted deep feature vectors to correct class imbalance.
Trained and evaluated multiple convolutional neural networks across six patient-wise train-test splits, including a focused experiment on axial MRI slices from 50 subjects.
Generated Grad-CAM visual heatmaps to identify anatomical regions influencing model predictions.
Model diagnostic performance varied substantially across subject-wise splits, demonstrating high peak metrics on select configurations but a conservative aggregate baseline.
Grad-CAM visualizations confirmed that deep learning models focused primarily on central brain structures linked to neurodegeneration with minimal attention to non-diagnostic areas.

Cite This Study

Isar et al. (2026) studied this question.

synapsesocial.com/papers/6a801a0e75c2e31742c8666a https://doi.org/10.3390/a19080680

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

View Full Paper

Ask AI

Bookmark