Exploiting macro- and micro-structural brain changes for improved Parkinson’s disease classification from MRI data
Abstract Parkinson’s disease (PD) is the second most common neurodegenerative disease. Accurate PD diagnosis is crucial for effective treatment and prognosis but can be challenging, especially at early disease stages. This study aimed to develop and evaluate an explainable deep learning model for PD...
Main Authors: | , , , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Nature Portfolio
2024-02-01
|
Series: | npj Parkinson's Disease |
Online Access: | https://doi.org/10.1038/s41531-024-00647-9 |
_version_ | 1797275425456521216 |
---|---|
author | Milton Camacho Matthias Wilms Hannes Almgren Kimberly Amador Richard Camicioli Zahinoor Ismail Oury Monchi Nils D. Forkert For the Alzheimer’s Disease Neuroimaging Initiative |
author_facet | Milton Camacho Matthias Wilms Hannes Almgren Kimberly Amador Richard Camicioli Zahinoor Ismail Oury Monchi Nils D. Forkert For the Alzheimer’s Disease Neuroimaging Initiative |
author_sort | Milton Camacho |
collection | DOAJ |
description | Abstract Parkinson’s disease (PD) is the second most common neurodegenerative disease. Accurate PD diagnosis is crucial for effective treatment and prognosis but can be challenging, especially at early disease stages. This study aimed to develop and evaluate an explainable deep learning model for PD classification from multimodal neuroimaging data. The model was trained using one of the largest collections of T1-weighted and diffusion-tensor magnetic resonance imaging (MRI) datasets. A total of 1264 datasets from eight different studies were collected, including 611 PD patients and 653 healthy controls (HC). These datasets were pre-processed and non-linearly registered to the MNI PD25 atlas. Six imaging maps describing the macro- and micro-structural integrity of brain tissues complemented with age and sex parameters were used to train a convolutional neural network (CNN) to classify PD/HC subjects. Explainability of the model’s decision-making was achieved using SmoothGrad saliency maps, highlighting important brain regions. The CNN was trained using a 75%/10%/15% train/validation/test split stratified by diagnosis, sex, age, and study, achieving a ROC-AUC of 0.89, accuracy of 80.8%, specificity of 82.4%, and sensitivity of 79.1% on the test set. Saliency maps revealed that diffusion tensor imaging data, especially fractional anisotropy, was more important for the classification than T1-weighted data, highlighting subcortical regions such as the brainstem, thalamus, amygdala, hippocampus, and cortical areas. The proposed model, trained on a large multimodal MRI database, can classify PD patients and HC subjects with high accuracy and clinically reasonable explanations, suggesting that micro-structural brain changes play an essential role in the disease course. |
first_indexed | 2024-03-07T15:13:11Z |
format | Article |
id | doaj.art-8da5db4b61c5422fab5a52810c745b6f |
institution | Directory Open Access Journal |
issn | 2373-8057 |
language | English |
last_indexed | 2024-03-07T15:13:11Z |
publishDate | 2024-02-01 |
publisher | Nature Portfolio |
record_format | Article |
series | npj Parkinson's Disease |
spelling | doaj.art-8da5db4b61c5422fab5a52810c745b6f2024-03-05T18:03:21ZengNature Portfolionpj Parkinson's Disease2373-80572024-02-0110111210.1038/s41531-024-00647-9Exploiting macro- and micro-structural brain changes for improved Parkinson’s disease classification from MRI dataMilton Camacho0Matthias Wilms1Hannes Almgren2Kimberly Amador3Richard Camicioli4Zahinoor Ismail5Oury Monchi6Nils D. Forkert7For the Alzheimer’s Disease Neuroimaging InitiativeBiomedical Engineering Graduate Program, University of CalgaryDepartment of Radiology, University of CalgaryHotchkiss Brain Institute, University of CalgaryBiomedical Engineering Graduate Program, University of CalgaryNeuroscience and Mental Health Institute and Department of Medicine (Neurology), University of AlbertaHotchkiss Brain Institute, University of CalgaryHotchkiss Brain Institute, University of CalgaryDepartment of Radiology, University of CalgaryAbstract Parkinson’s disease (PD) is the second most common neurodegenerative disease. Accurate PD diagnosis is crucial for effective treatment and prognosis but can be challenging, especially at early disease stages. This study aimed to develop and evaluate an explainable deep learning model for PD classification from multimodal neuroimaging data. The model was trained using one of the largest collections of T1-weighted and diffusion-tensor magnetic resonance imaging (MRI) datasets. A total of 1264 datasets from eight different studies were collected, including 611 PD patients and 653 healthy controls (HC). These datasets were pre-processed and non-linearly registered to the MNI PD25 atlas. Six imaging maps describing the macro- and micro-structural integrity of brain tissues complemented with age and sex parameters were used to train a convolutional neural network (CNN) to classify PD/HC subjects. Explainability of the model’s decision-making was achieved using SmoothGrad saliency maps, highlighting important brain regions. The CNN was trained using a 75%/10%/15% train/validation/test split stratified by diagnosis, sex, age, and study, achieving a ROC-AUC of 0.89, accuracy of 80.8%, specificity of 82.4%, and sensitivity of 79.1% on the test set. Saliency maps revealed that diffusion tensor imaging data, especially fractional anisotropy, was more important for the classification than T1-weighted data, highlighting subcortical regions such as the brainstem, thalamus, amygdala, hippocampus, and cortical areas. The proposed model, trained on a large multimodal MRI database, can classify PD patients and HC subjects with high accuracy and clinically reasonable explanations, suggesting that micro-structural brain changes play an essential role in the disease course.https://doi.org/10.1038/s41531-024-00647-9 |
spellingShingle | Milton Camacho Matthias Wilms Hannes Almgren Kimberly Amador Richard Camicioli Zahinoor Ismail Oury Monchi Nils D. Forkert For the Alzheimer’s Disease Neuroimaging Initiative Exploiting macro- and micro-structural brain changes for improved Parkinson’s disease classification from MRI data npj Parkinson's Disease |
title | Exploiting macro- and micro-structural brain changes for improved Parkinson’s disease classification from MRI data |
title_full | Exploiting macro- and micro-structural brain changes for improved Parkinson’s disease classification from MRI data |
title_fullStr | Exploiting macro- and micro-structural brain changes for improved Parkinson’s disease classification from MRI data |
title_full_unstemmed | Exploiting macro- and micro-structural brain changes for improved Parkinson’s disease classification from MRI data |
title_short | Exploiting macro- and micro-structural brain changes for improved Parkinson’s disease classification from MRI data |
title_sort | exploiting macro and micro structural brain changes for improved parkinson s disease classification from mri data |
url | https://doi.org/10.1038/s41531-024-00647-9 |
work_keys_str_mv | AT miltoncamacho exploitingmacroandmicrostructuralbrainchangesforimprovedparkinsonsdiseaseclassificationfrommridata AT matthiaswilms exploitingmacroandmicrostructuralbrainchangesforimprovedparkinsonsdiseaseclassificationfrommridata AT hannesalmgren exploitingmacroandmicrostructuralbrainchangesforimprovedparkinsonsdiseaseclassificationfrommridata AT kimberlyamador exploitingmacroandmicrostructuralbrainchangesforimprovedparkinsonsdiseaseclassificationfrommridata AT richardcamicioli exploitingmacroandmicrostructuralbrainchangesforimprovedparkinsonsdiseaseclassificationfrommridata AT zahinoorismail exploitingmacroandmicrostructuralbrainchangesforimprovedparkinsonsdiseaseclassificationfrommridata AT ourymonchi exploitingmacroandmicrostructuralbrainchangesforimprovedparkinsonsdiseaseclassificationfrommridata AT nilsdforkert exploitingmacroandmicrostructuralbrainchangesforimprovedparkinsonsdiseaseclassificationfrommridata AT forthealzheimersdiseaseneuroimaginginitiative exploitingmacroandmicrostructuralbrainchangesforimprovedparkinsonsdiseaseclassificationfrommridata |