MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Cluster Analysis On Longitudinal Data of Parkinson Disease Subjects

M Salmanpourpaeenafrakati1,2*, A Saberi3, G Hajianfar4, M Shamsaei1, H Soltanian-zadeh6, A Rahmim1, (1) University of British Columbia, BC cancer, Vancouver, BC, CA, (2) Islamic Azad University,Tehran ,Iran ,(3) Rajaie Cardiovascular Medical And Research Center, Islamic Azad University,Tehran ,Iran,(4) Amirkabir University Of Technology, Islamic Azad University,Tehran ,Iran,(5) Henry Ford Health System, Detroit, USA

Presentations

(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose:
We aimed to identify distinct disease progression pathways in Parkinson’s disease (PD), making use of imaging features, for appropriate understanding of disease and powering of clinical trials.

Methods:
We studied 885 PD-subjects derived from longitudinal datasets (years 0,1,2&4; Parkinson’s Progressive Marker Initiative), with 980 features, e.g. Movement Disorder Society’s Unified Parkinson's Disease Rating Scale (MDS-UPDRS) measures, a range of task/exam performances, socioeconomic/family histories and SPECT image features. Segmentation of regions-of-interest (ROIs; caudate and putamen) on DaT SPECT images were performed via MRI images. Radiomic features (RFs) were extracted for each ROI using our standardized SERA software. We first performed unsupervised clustering to identify disease subtypes in a given year (3 clusters robustly identified in another work of ours, applicable to all years, categorized as i) motor-dominant, ii) non-motor-dominant, and ii) mixed-motor/non-motor). We then created 2 datasets with same patients followed longitudinally. First dataset included 84 patients which had all features in each year. Second dataset consisted of 143 patients (based on year 4) with some missing data in some years that we filled using ensemble hybrid machine-learning (majority-voting) system consisting of 8 feature-selection algorithms (FSAs), 8 dimensionality-reduction algorithms, and 6 classifiers. We finally performed longitudinal-clustering of disease progression pathways, using K-Means Longitudinal Clustering (KMLC), an extension of standard K-mean clustering. Ray-Turi clustering evaluation method was used for optimal selection of number of pathway clusters.
Results:
Our analysis revealed significant heterogeneity in disease projection. We identified 7 distinct progression trajectories/clusters from initial dataset, confirmed by analysis of second dataset. The pathways included those with consistent disease escalation (2 pathways), inconsistent progression (2 pathway), improvement (1 pathways), and slow improvement (2 pathways).
Conclusion:
Advanced longitudinal missing-data filling and unsupervised-clustering demonstrated 7 distinct longitudinal clusters, depicting significant heterogeneity in PD disease progression.

Keywords

Not Applicable / None Entered.

Taxonomy

IM- Multi-Modality Imaging Systems: MRI/SPECT - human

Contact Email