Feature selection for classification based on fine motor signs of Parkinson's disease

EMD-Based Method for Supervised Classification of Parkinson’s Disease Patients Using Balance Control Data

There has recently been increasing interest in postural stability aimed at gaining a better understanding of the human postural system. This system controls human balance in quiet standing and during locomotion. Parkinson’s disease (PD) is the most common degenerative movement disorder that affects human stability and causes falls and injuries. This paper proposes a novel methodology to differentiate between healthy individuals and those with PD through the empirical mode decomposition (EMD) method. EMD enables the breaking down of a complex signal into several elementary signals called intrinsic mode functions (IMFs). Three temporal parameters and three spectral parameters are extracted from each stabilometric signal as well as from its IMFs. Next, the best five features are selected using the feature selection method. The classification task is carried out using four known machine-learning methods, KNN, decision tree, Random Forest and SVM classifiers, over 10-fold cross validation. The used dataset consists of 28 healthy subjects (14 young adults and 14 old adults) and 32 PD patients (12 young adults and 20 old adults). The SVM method has a performance of 92% and the Dempster–Sahfer formalism method has an accuracy of 96.51%.

Intelligent Sensory Pen for Aiding in the Diagnosis of Parkinson's Disease from Dynamic Handwriting Analysis

In this paper, we propose a pen device capable of detecting specific features from dynamic handwriting tests for aiding on automatic Parkinson’s disease identification. The method used in this work uses machine learning to compare the raw signals from different sensors in the device coupled to a pen and extract relevant information such as tremors and hand acceleration to diagnose the patient clinically. Additionally, the datasets composed of raw signals from healthy and Parkinson’s disease patients acquired here are made available to further contribute to research related to this topic.

Evaluation of progression markers in the premotor phase of Parkinson's disease: the progression markers in the premotor phase study

BACKGROUND

The clinical diagnosis of Parkinson's disease (PD) is proposed to be too late for the application of beneficial neuroprotective treatment. Thus, it is important to identify and follow individuals at risk for PD in order to gain knowledge about the prodromal course of the disease. Substantia nigra hyperechogenicity (SN+) has been confirmed as a risk factor for PD and appears promising as a predictor of PD, particularly in combination with other putative PD markers. We present the design and initial data of a 2-year longitudinal investigation of subjects proposed to be at high risk of developing PD (HRPD), compared to early PD patients and control subjects. The aim of the presented study is to monitor progression of the neurodegenerative process to motor PD.

METHODS

In total, 40 HRPD, 16 PD and 41 control individuals were recruited. The HRPD subjects had SN+ and additionally either 1 cardinal PD motor sign or 2 further risk (e.g. positive family history) or prodromal markers (e.g. hyposmia). In this cohort, motor function, olfaction, mood and blood markers will be evaluated every 6 months, complemented by a comprehensive clinical, imaging and electrophysiological assessment.

RESULTS

PD, HRPD and control subjects did not differ significantly regarding age, but the HRPD group consisted mainly of males (72.5% of HRPD subjects vs. 43.9% of controls; p = 0.013). Mean disease duration in PD patients was 31 months (range 15-56). HRPD subjects were predominantly recruited according to the occurrence of slight motor signs (HRPD 77.5%, PD 100%, p = 0.05; controls 0%, HRPD vs. controls, p < 0.017). The Unified Parkinson's Disease Rating Scale motor score (mean, range) indicated that the HRPD group (4, 0-12) had values between those of controls (0, 0-2; p < 0.017) and PD subjects (26, 9-55; p < 0.017). Among nonmotor symptoms, hyposmia was more common in both HRPD (47.5%) and PD subjects (75%) than in controls (5.1%; p < 0.017 for both comparisons).

CONCLUSIONS

Here, we describe the recruitment of a highly enriched-risk cohort and a promising study design to assess progression to motor PD. Whether the HRPD group indeed suffers from early, PD-specific neurodegeneration remains to be verified in the ongoing follow-up examinations.

Differentiation between Parkinson disease and other forms of Parkinsonism using support vector machine analysis of susceptibility-weighted imaging (SWI): initial results

OBJECTIVES

To diagnose Parkinson disease (PD) at the individual level using pattern recognition of brain susceptibility-weighted imaging (SWI).

METHODS

We analysed brain SWI in 36 consecutive patients with Parkinsonism suggestive of PD who had (1) SWI at 3 T, (2) brain (123)I-ioflupane SPECT and (3) extensive neurological testing including follow-up (16 PD, 67.4 ± 6.2 years, 11 female; 20 OTHER, a heterogeneous group of atypical Parkinsonism syndromes 65.2 ± 12.5 years, 6 female). Analysis included group-level comparison of SWI values and individual-level support vector machine (SVM) analysis.

RESULTS

At the group level, simple visual analysis yielded no differences between groups. However, the group-level analyses demonstrated increased SWI in the bilateral thalamus and left substantia nigra in PD patients versus other Parkinsonism. The inverse comparison yielded no supra-threshold clusters. At the individual level, SVM correctly classified PD patients with an accuracy above 86 %.

CONCLUSIONS

SVM pattern recognition of SWI data provides accurate discrimination of PD among patients with various forms of Parkinsonism at an individual level, despite the absence of visually detectable alterations. This pilot study warrants further confirmation in a larger cohort of PD patients and with different MR machines and MR parameters.

Individual detection of patients with Parkinson disease using support vector machine analysis of diffusion tensor imaging data: initial results

BACKGROUND AND PURPOSE

Brain MR imaging is routinely performed in the work-up of suspected PD, yet its role is essentially limited to the exclusion of other pathologies. We performed a pattern-recognition analysis based on DTI data to detect subjects with PD at the individual level.

MATERIALS AND METHODS

We included 40 consecutive patients with Parkinsonism suggestive of PD who had DTI at 3T, brain (123)I ioflupane SPECT (DaTSCAN), and extensive neurologic testing including follow-up (17 PD: age range, 67.8 ± 6.7 years; 9 women; 23 Other: consisting of atypical forms of Parkinsonism; age range, 67.2 ± 9.7 years; 7 women). Data analysis included group-level TBSS and individual-level SVM classification.

RESULTS

At the group level, patients with PD versus Other had spatially consistent increase in FA and decrease in RD and MD in a bilateral network, predominantly in the right frontal white matter. At the individual level, SVM correctly classified patients with PD at the individual level with accuracies up to 97%.

CONCLUSIONS

Support vector machine-based pattern recognition of DTI data provides highly accurate detection of patients with PD among those with suspected PD at an individual level, which is potentially clinically applicable. Because most suspected subjects with PD undergo brain MR imaging, already existing MR imaging data may be reused; this practice is very cost-efficient.

Preliminary investigation of test-retest reliability of a robotic assessment for Parkinson's disease

Several systems have been proposed to objectively quantify the motor signs of Parkinson's disease. While validity has been investigated for a subset of these systems, test-retest reliability of automated assessments has been more neglected. We have created a protocol that uses a robotic system to measure force and position as the user performs tracking tasks with and without a simultaneous cognitive or motor task. We present preliminary results for test-retest reliability with eight individuals with early to moderate PD. Based on this data, we identified variables with large, significant correlations between the test and retest data. A total of 21 reliable variables were identified with all portions of the assessment protocol producing reliable variables.

Feature selection for accelerometer-based posture analysis in Parkinson's disease

Posture analysis in quiet standing is a key component of the clinical evaluation of Parkinson's disease (PD), postural instability being one of PD's major symptoms. The aim of this study was to assess the feasibility of using accelerometers to characterize the postural behavior of early mild PD subjects. Twenty PD and 20 control subjects, wearing an accelerometer on the lower back, were tested in five conditions characterized by sensory and attentional perturbation. A total of 175 measures were computed from the signals to quantify tremor, acceleration, and displacement of body sway. Feature selection was implemented to identify the subsets of measures that better characterize the distinctive behavior of PD and control subjects. It was based on different classifiers and on a nested cross validation, to maximize robustness of selection with respect to changes in the training set. Several subsets of three features achieved misclassification rates as low as 5%. Many of them included a tremor-related measure, a postural measure in the frequency domain, and a postural displacement measure. Results suggest that quantitative posture analysis using a single accelerometer and a simple test protocol may provide useful information to characterize early PD subjects. This protocol is potentially usable to monitor the disease's progression.