A multi-center distributed learning approach for Parkinson's disease classification using the traveling model paradigm.

Distributed learning is a promising alternative to central learning for machine learning (ML) model training, overcoming data-sharing problems in healthcare. Previous studies exploring federated learning (FL) or the traveling model (TM) setup for medical image-based disease classification often relied on large databases with a limited number of centers or simulated artificial centers, raising doubts about real-world applicability. This study develops and evaluates a convolution neural network (CNN) for Parkinson's disease classification using data acquired by 83 diverse real centers around the world, mostly contributing small training samples. Our approach specifically makes use of the TM setup, which has proven effective in scenarios with limited data availability but has never been used for image-based disease classification. Our findings reveal that TM is effective for training CNN models, even in complex real-world scenarios with variable data distributions. After sufficient training cycles, the TM-trained CNN matches or slightly surpasses the performance of the centrally trained counterpart (AUROC of 83% vs. 80%). Our study highlights, for the first time, the effectiveness of TM in 3D medical image classification, especially in scenarios with limited training samples and heterogeneous distributed data. These insights are relevant for situations where ML models are supposed to be trained using data from small or remote medical centers, and rare diseases with sparse cases. The simplicity of this approach enables a broad application to many deep learning tasks, enhancing its clinical utility across various contexts and medical facilities.

Exploiting macro- and micro-structural brain changes for improved Parkinson's disease classification from MRI data.

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.

Identifying Biases in a Multicenter MRI Database for Parkinson's Disease Classification: Is the Disease Classifier a Secret Site Classifier?

Sharing multicenter imaging datasets can be advantageous to increase data diversity and size but may lead to spurious correlations between site-related biological and non-biological image features and target labels, which machine learning (ML) models may exploit as shortcuts. To date, studies analyzing how and if deep learning models may use such effects as a shortcut are scarce. Thus, the aim of this work was to investigate if site-related effects are encoded in the feature space of an established deep learning model designed for Parkinson's disease (PD) classification based on T1-weighted MRI datasets. Therefore, all layers of the PD classifier were frozen, except for the last layer of the network, which was replaced by a linear layer that was exclusively re-trained to predict three potential bias types (biological sex, scanner type, and originating site). Our findings based on a large database consisting of 1880 MRI scans collected across 41 centers show that the feature space of the established PD model (74% accuracy) can be used to classify sex (75% accuracy), scanner type (79% accuracy), and site location (71% accuracy) with high accuracies despite this information never being explicitly provided to the PD model during original training. Overall, the results of this study suggest that trained image-based classifiers may use unwanted shortcuts that are not meaningful for the actual clinical task at hand. This finding may explain why many image-based deep learning models do not perform well when applied to data from centers not contributing to the training set.

Image-encoded biological and non-biological variables may be used as shortcuts in deep learning models trained on multisite neuroimaging data.

OBJECTIVE: This work investigates if deep learning (DL) models can classify originating site locations directly from magnetic resonance imaging (MRI) scans with and without correction for intensity differences. MATERIAL AND METHODS: A large database of 1880 T1-weighted MRI scans collected across 41 sites originally for Parkinson's disease (PD) classification was used to classify sites in this study. Forty-six percent of the datasets are from PD patients, while 54% are from healthy participants. After preprocessing the T1-weighted scans, 2 additional data types were generated: intensity-harmonized T1-weighted scans and log-Jacobian deformation maps resulting from nonlinear atlas registration. Corresponding DL models were trained to classify sites for each data type. Additionally, logistic regression models were used to investigate the contribution of biological (age, sex, disease status) and non-biological (scanner type) variables to the models' decision. RESULTS: A comparison of the 3 different types of data revealed that DL models trained using T1-weighted and intensity-harmonized T1-weighted scans can classify sites with an accuracy of 85%, while the model using log-Jacobian deformation maps achieved a site classification accuracy of 54%. Disease status and scanner type were found to be significant confounders. DISCUSSION: Our results demonstrate that MRI scans encode relevant site-specific information that models could use as shortcuts that cannot be removed using simple intensity harmonization methods. CONCLUSION: The ability of DL models to exploit site-specific biases as shortcuts raises concerns about their reliability, generalization, and deployability in clinical settings.

Machine learning-based prediction of longitudinal cognitive decline in early Parkinson's disease using multimodal features.

Patients with Parkinson's Disease (PD) often suffer from cognitive decline. Accurate prediction of cognitive decline is essential for early treatment of at-risk patients. The aim of this study was to develop and evaluate a multimodal machine learning model for the prediction of continuous cognitive decline in patients with early PD. We included 213 PD patients from the Parkinson's Progression Markers Initiative (PPMI) database. Machine learning was used to predict change in Montreal Cognitive Assessment (MoCA) score using the difference between baseline and 4-years follow-up data as outcome. Input features were categorized into four sets: clinical test scores, cerebrospinal fluid (CSF) biomarkers, brain volumes, and genetic variants. All combinations of input feature sets were added to a basic model, which consisted of demographics and baseline cognition. An iterative scheme using RReliefF-based feature ranking and support vector regression in combination with tenfold cross validation was used to determine the optimal number of predictive features and to evaluate model performance for each combination of input feature sets. Our best performing model consisted of a combination of the basic model, clinical test scores and CSF-based biomarkers. This model had 12 features, which included baseline cognition, CSF phosphorylated tau, CSF total tau, CSF amyloid-beta1-42, geriatric depression scale (GDS) scores, and anxiety scores. Interestingly, many of the predictive features in our model have previously been associated with Alzheimer's disease, showing the importance of assessing Alzheimer's disease pathology in patients with Parkinson's disease.

Explainable classification of Parkinson's disease using deep learning trained on a large multi-center database of T1-weighted MRI datasets.

INTRODUCTION: Parkinson's disease (PD) is a severe neurodegenerative disease that affects millions of people. Early diagnosis is important to facilitate prompt interventions to slow down disease progression. However, accurate PD diagnosis can be challenging, especially in the early disease stages. The aim of this work was to develop and evaluate a robust explainable deep learning model for PD classification trained from one of the largest collections of T1-weighted magnetic resonance imaging datasets. MATERIALS AND METHODS: A total of 2,041 T1-weighted MRI datasets from 13 different studies were collected, including 1,024 datasets from PD patients and 1,017 datasets from age- and sex-matched healthy controls (HC). The datasets were skull stripped, resampled to isotropic resolution, bias field corrected, and non-linearly registered to the MNI PD25 atlas. The Jacobian maps derived from the deformation fields together with basic clinical parameters were used to train a state-of-the-art convolutional neural network (CNN) to classify PD and HC subjects. Saliency maps were generated to display the brain regions contributing the most to the classification task as a means of explainable artificial intelligence. RESULTS: The CNN model was trained using an 85%/5%/10% train/validation/test split stratified by diagnosis, sex, and study. The model achieved an accuracy of 79.3%, precision of 80.2%, specificity of 81.3%, sensitivity of 77.7%, and AUC-ROC of 0.87 on the test set while performing similarly on an independent test set. Saliency maps computed for the test set data highlighted frontotemporal regions, the orbital-frontal cortex, and multiple deep gray matter structures as most important. CONCLUSION: The developed CNN model, trained on a large heterogenous database, was able to differentiate PD patients from HC subjects with high accuracy with clinically feasible classification explanations. Future research should aim to investigate the combination of multiple imaging modalities with deep learning and on validating these results in a prospective trial as a clinical decision support system.

Motor symptoms in Parkinson's disease are related to the interplay between cortical curvature and thickness.

INTRODUCTION: Brain atrophy in Parkinson's disease occurs to varying degrees in different brain regions, even at the early stage of the disease. While cortical morphological features are often considered independently in structural brain imaging studies, research on the co-progression of different cortical morphological measurements could provide new insights regarding the progression of PD. This study's aim was to examine the interplay between cortical curvature and thickness as a function of PD diagnosis, motor symptoms, and cognitive performance. METHODS: A total of 359 de novo PD patients and 159 healthy controls (HC) from the Parkinson's Progression Markers Initiative (PPMI) database were included in this study. Additionally, an independent cohort from four databases (182 PD, 132 HC) with longer disease durations was included to assess the effects of PD diagnosis in more advanced cases. Pearson correlation was used to determine subject-specific associations between cortical curvature and thickness estimated from T1-weighted MRI images. General linear modeling (GLM) was then used to assess the effect of PD diagnosis, motor symptoms, and cognitive performance on the curvature-thickness association. Next, longitudinal changes in the curvature-thickness correlation as well as the predictive effect of the cortical curvature-thickness association on changes in motor symptoms and cognitive performance across four years were investigated. Finally, Akaike information criterion (AIC) was used to build a GLM to model PD motor symptom severity cross-sectionally. RESULTS: A significant interaction effect between PD motor symptoms and age on the curvature-thickness correlation was found (ßstandardized = 0.11; t(350) = 2.12; p = 0.03). This interaction effect showed that motor symptoms in older patients were related to an attenuated curvature-thickness association. No significant effect of PD diagnosis was observed for the PPMI database (ß = 0.03; t(510) = 0.35; p = 0.72). However, in patients with a longer disease duration, a significant effect of diagnosis on the curvature-thickness association was found (ßstandardized = 0.31; t(306.7) = 3.49; p = 0.0006). Moreover, rigidity, but not tremor, in PD was significantly related to the curvature-thickness correlation (ßstandardized = 0.11, t(350) = 2.24, p = 0.03; ßstandardized = -0.03, t(350) = -0.58, p = 0.56, respectively). The curvature-thickness association was attenuated over time in both PD and HC, but the two groups did not show a significantly different effect (ßstandardized = 0.03, t(184.7) = 0.78, p = 0.44). No predictive effects of the CC-CT correlation on longitudinal changes in cognitive performance or motor symptoms were observed (all p-values > 0.05). The best cross-sectional model for PD motor symptoms included the curvature-thickness correlation, cognitive performance, and putamen dopamine transporter (DAT) binding, which together explained 14 % of variance. CONCLUSION: The association between cortical curvature and thickness is related to PD motor symptoms and age. This research shows the potential of modeling the curvature-thickness interplay in PD.