Identification of saccadic components in spinocerebellar ataxia applying an independent component analysis algorithm

Abstract

Anomalies in the oculomotor system are well known symptoms in different neurodegenerative diseases. It has been found that patients suffering from severe spino cerebellar ataxia type 2 show deterioration in the main parameters used to describe saccadic movements, specifically the slowing of horizontal saccadic eye movements. Besides, a combination of two components, named pulse and step, constitutes an accepted model of the saccadic generation system. In the present work, independent component analysis is applied in order to separate both pulse and step components, revealing significant differences in several parameters related to the morphology of these components between patients and control responses. Ten electro-oculographic records of spino cerebellar ataxia type 2 patients and ten control subjects were processed with the proposed algorithm with the aim of obtaining a correct diagnosis. The results obtained from these real experiments reveal the validity of the proposed approach as a classification tool for the diagnosis of this disease.

Introduction

There are almost 800 patients and 8000 presymptomatic relatives in Cuba at risk of developing some autosomal dominant cerebellar ataxia (ADCAs), in the next few years. This is a heterogeneous group of dominantly inherited neurological disorders characterized by progressive ataxia that results from degeneration of the cerebellum and its afferent connections [1], [2].

Worldwide prevalence of SCAs is estimated at about 6 cases per 100,000 people, although much higher figures have been reported in particular populations [3]. The variant of SCA2 is between the three most prevalent subtype worldwide among the almost 30 SCAs, together with SCA3 and SCA6 [3], [4]. However, exact global prevalence of SCA2 is undetermined as the few existing epidemiological studies have been performed in isolated geographical regions [4], [5]. SCA2 is the most predominant SCA variant in Italy, England, Spain, India, South Korea, and especially in Cuba [4], [6].

Cuba is the country that presents the highest concentration of patients with hereditary ataxias over the world. Particularly, the spinocerebellar ataxia type 2 (SCA2) molecular form is predominant. SCA2 reaches the highest prevalence in the province of Holguin, located in the north-east of Cuba, getting to a rate of 40 patients for each 100,000 inhabitants [7]. Reported high prevalence is probably the result of a founder effect, but might be due to an interaction between a mutant gene and an unidentified environmental neurotoxin [8], [9]. The stabilization of the prevalence along time suggests the existence of mutated chromosomes, acting as reservoir for further expansions [10].

Ocular saccades are the rapid eyes movements followed by a stable fixation that allow animals with an area of high acuity in the retina to inspect peripheral stimuli. Saccade trajectories tend to be remarkably stereotyped both within and across the different species. The saccade duration and its peak velocity increase monotonically with the amplitude of the movement in a regular way, which has been called the “main sequence” as a resemblance to the astronomical term [11], [12].

A commonly accepted model of the saccadic system states that the programming of the saccades consists of a pulse and a step component. The pulse consists in a high frequency burst that creates the muscle forces which rapidly move the eye to a new position, whereas the step component keeps the eyes stable at the new position [13], [14].

The ocular movement records have been widely used in processing and classification of biological signals and pathological conditions. Specifically, saccadic movements in response to a visual stimulation are considered amongst the most useful tools in the study of neurological pathologies [15], [16], [17], [18], [19], [20], [21], [22], [23]. Several studies have reported oculomotor abnormalities in ADCA [3], [24]; among them, slowness of saccades has been suggested as a relatively characteristic finding in SCA2 patients [5], [25], [26].

In order to quantify and to establish the possibility of comparing clinical findings, a variety of clinical testing scales were developed. The most frequently used scale is the International Cooperative Ataxia Rating Scales (ICARS) [27], which involves a quantification of postural and stance disorders, limb ataxia, dysarthria and oculomotor disorders. The Scale for the Assessment and Rating of Ataxia (SARA) [28], [29] is proposed as an alternative to ICARS, claiming to be easier and faster to assess and eliminating redundant items form ICARS. However, the utilization of the results from these scales in computer-aided diagnosis of SCA2 is difficult, since the obtained data is semiquantitative and based on subjective estimates made by medical experts.

There are other symptoms of SCA2 and similar ataxias apart from the cited abnormalities in ocular saccades which have been studied and quantified for SCA2 diagnosis. In particular, deficits in ocular and manual tracking [24], postural stability [30], and stair-climbing [31] have been studied. Comparison and reproduction of the results are complicated since the experiment equipment is not easily accessible and the number of tested subjects is small due to the low prevalence of SCA2 world-wide.

In the present work, the application of an independent component analysis (ICA) based approach is proposed in order to separate both pulse and step components from the original saccadic responses to a given visual stimuli. Results reveal significant differences between patients and control responses. A k-means classification algorithm will be applied in order to identify a given experiment according to the values of the parameters related to its pulse and step components.

Through the use of our ICA based methodology, we expect to obtain a diagnosis process which is meant to be easy to apply, independent from the medical experts' subjective opinion, and adequate to the Cuban sanitary system resources and, specifically, to the Centre for the Research and Rehabilitation of Hereditary Ataxias (CIRAH) at Holguín (Cuba). ICA eliminates noise components, focusing only in the pulse and step components that form an ocular saccade. Section 2 of this paper describes the materials and methods needed for the application of the diagnosis algorithm, starting from an outline of the algorithm itself. Section 3 shows the results over 20 experiments, where 10 of them belong to a group of control subjects and the 10 remaining are SCA2 patients. 4 Discussion, 5 Conclusions present the discussion and conclusions, respectively.