Classification of Huntington's Disease Stage with Features Derived from Structural and Diffusion-Weighted Imaging.

The purpose of this study was to classify Huntington's disease (HD) stage using support vector machines and measures derived from T1- and diffusion-weighted imaging. The effects of feature selection approach and combination of imaging modalities are assessed. Fourteen premanifest-HD individuals (Pre-HD; on average > 20 years from estimated disease onset), eleven early-manifest HD (Early-HD) patients, and eighteen healthy controls (HC) participated in the study. We compared three feature selection approaches: (i) whole-brain segmented grey matter (GM; voxel-based measure) or fractional anisotropy (FA) values; (ii) GM or FA values from subcortical regions-of-interest (caudate, putamen, pallidum); and (iii) automated selection of GM or FA values with the algorithm Relief-F. We assessed single- and multi-kernel approaches to classify combined GM and FA measures. Significant classifications were achieved between Early-HD and Pre-HD or HC individuals (accuracy: generally, 85% to 95%), and between Pre-HD and controls for the feature FA of the caudate ROI (74% accuracy). The combination of GM and FA measures did not result in higher performances. We demonstrate evidence on the high sensitivity of FA for the classification of the earliest Pre-HD stages, and successful distinction between HD stages.

Processing Time Reduction: an Application in Living Human High-Resolution Diffusion Magnetic Resonance Imaging Data.

High Angular Resolution Diffusion Imaging (HARDI) is a type of brain imaging that collects a very large amount of data, and if many subjects are considered then it amounts to a big data framework (e.g., the human connectome project has 20 Terabytes of data). HARDI is also becoming increasingly relevant for clinical settings (e.g., detecting early cerebral ischemic changes in acute stroke, and in pre-clinical assessment of white matter-WM anatomy using tractography). Thus, this method is becoming a routine assessment in clinical settings. In such settings, the computation time is critical, and finding forms of reducing the processing time in high computation processes such as Diffusion Spectrum Imaging (DSI), a form of HARDI data, is very relevant to increase data-processing speed. Here we analyze a method for reducing the computation time of the dMRI-based axonal orientation distribution function h by using Monte Carlo sampling-based methods for voxel selection. Results evidenced a robust reduction in required data sampling of about 50 % without losing signal's quality. Moreover, we show that the convergence to the correct value in this type of Monte Carlo HARDI/DSI data-processing has a linear improvement in data-processing speed of the ODF determination. Although further improvements are needed, our results represent a promissory step for future processing time reduction in big data.

Classification of Huntington's disease stage with support vector machines: A study on oculomotor performance.

Alterations in oculomotor performance are among the first observable physical alterations during presymptomatic stages of Huntington's disease (HD). Quantifiable measurements of oculomotor performance have been studied as possible markers of disease status and progression in presymptomatic and early symptomatic stages of HD, on the basis of traditional analysis methods. Whether oculomotor performance can be used to classify individuals according to HD disease stage has yet to be explored via the application of machine-learning methods. In the present study, we report the application of the support vector machine (SVM) algorithm to oculomotor features pooled from a four-task psychophysical experiment. We were able to automatically distinguish control participants from presymptomatic HD (pre-HD) participants with an accuracy of 73.47 %, a sensitivity of 74.31 %, and a specificity of 72.64 %; to distinguish control participants from HD patients with an accuracy of 81.84 %, a sensitivity of 76.19 %, and a specificity of 87.48 %; and to distinguish pre-HD participants from HD patients with an accuracy of 83.54 %, a sensitivity of 92.62 %, and a specificity of 74.45 %. These results demonstrate that the application of supervised classification methods to oculomotor features is a valuable and promising approach to the automatic detection of disease stage in HD.