G. Longoni
The Hospital for Sick Children NeurologyAuthor Of 2 Presentations
PS07.04 - Fibre-specific white matter differences in children with pediatric acquired demyelinating syndromes compared to healthy children
Abstract
Background
White matter (WM) microstructural changes occur in youth with multiple sclerosis (MS) and myelin oligodendrocyte glyoprotein (MOG)-associated disorders. While diffusion tensor imaging has been extensively used to characterize white matter, this method lacks microstructural and pathological specificity. ‘Fixel Based Analysis’ (FBA) statistically estimates changes in diffusion MRI connectivity that is specific to micro and macro-structure. WM damage that leads to less densely packed axons in a fiber bundle causes a decrease in fibre density (FD). If the number of axons is not reduced but occupies less area, then fibre cross-section (FC) will decrease. Last, if the density of axons within a fibre bundle and the area the bundle occupies are reduced, then fibre density and cross-section (FDC) will decrease.
Objectives
To use whole-brain FBA to measure differences in FD, FC, FDC in youth with demyelinating syndromes compared to healthy controls.
Methods
We evaluated group differences in the FBA metrics between 28 typically developing children (17F; age 15.0±2.6y), 19 children with MS (13F; 16.9±1.1y; disease duration (DD)=0.1-11.7y; expanded disability status scale(EDSS):median=1.5,range=0-4.5), and 11 children with MOG (8F;12.1±2.8y; DD=0.5-6.4y;EDSS:m=1.0,r=0-3). Multi-shell diffusion-weighted imaging of the brain was acquired with echo planar imaging on a 3T MRI scanner and was pre-processed to correct for distortions and movement. Whole-brain group FBA was performed on FD, FC and FDC to test differences between groups adjusting for age, sex, total intracranial volume, EDSS and DD (p<0.05, family-wise error (FWE) corrected).
Results
Participants with MS and MOG showed reduced FD, FC and FDC relative to typically developing children (FWE corrected p<0.05). Differences in FD were found within splenium, superior longitudinal fasciculus and optic radiations. MS patients had reduced FDC within the corticospinal tract and cerebellar peduncle compared to MOG patients. In participants with MS and MOG, decreased FD within the brain stem, cerebellar peduncles and corona radiata was associated with increased DD and EDSS.
Conclusions
Our preliminary findings showed that patients with demyelinating disorders display decreased axonal density and fibre bundle size in multiple WM tracts relative to typically developing children, which were related to clinical outcomes (EDSS, DD). These changes were more pronounced in MS compared to MOG participants in selected WM tracts.
PS16.03 - Use of machine learning classifiers based on structural and functional visual metrics to predict diagnosis in children with acquired demyelination.
Abstract
Background
Predicting diagnosis in youth at the first episode of demyelination is feasible in some but not all cases. Machine learning classifiers (MLC) can be trained to identify relationships between numerous multimodal input features and disease classifications to provide highly accurate predictions.
Objectives
To assess performance of machine learning classifiers for early disease diagnosis based on visual metrics in youth with demyelination.
Methods
Standardized clinical and visual data was prospectively collected at disease onset from 141 pediatric subjects with acquired demyelinating syndromes (ADS) and 75 healthy controls (HC). Participants were recruited through The Hospital for Sick Children (Toronto, Ontario (2010-2020)) and University of Calgary (2010-2017). Patients were classified using consensus definitions of demyelinating disorders and serum antibody testing for myelin oligodendrocyte glycoprotein (MOG) and aquaporin 4 (AQP4). Twenty-two auto-segmented Optical Coherence Tomography (OCT) features, 4 functional visual and 4 clinical features were used in a stratified manner alone or in combination to identify which combination of features provided the highest predictive accuracy. These input features were analyzed using 9 supervised MLC (Random Forest (RF), AdaBoost, XGBoost, Decision Tree (DT), Logistic Regression, Support Vector Machines (SVM), k-Nearest Neighbors, Stochastic Gradient Descent, Multinomial Naive Bayes). Data was split 80/20 between training and test sets. Backward feature selection was performed to re-run the analysis with best scoring predictor features in the MLC with highest predictive accuracy.
Results
AdaBoost, SVM, and DT were the best performing MLC with a test set accuracy between 82-88% in distinguishing between ADS and HC eyes. Multiple sclerosis (MS) was distinguished from HC with 92% accuracy. In descending order, fovea thickness, inferotemporal ganglion cell layer (GCL) thickness, low contrast visual acuity, outer inferior macular thickness, temporal peripapillary retinal nerve fiber layer and superior GCL thicknesses were the most important contributors to disease classification.
Conclusions
MLC can be used to combine visual metrics and clinical parameters to distinguish ADS from HC, and to predict MS. In addition to commonly used clinical metrics, we identified other structural and functional metrics that contribute importantly to classification. Among the machine learning algorithms tested, AdaBoost, SVM and DT performed best for this model.