Detecting cognitive impairment by eye movement analysis using automatic classification algorithms

Abstract

The Visual Paired Comparison (VPC) task is a recognition memory test that has shown promise for the detection of memory impairments associated with mild cognitive impairment (MCI). Because patients with MCI often progress to Alzheimer’s Disease (AD), the VPC may be useful in predicting the onset of AD. VPC uses noninvasive eye tracking to identify how subjects view novel and repeated visual stimuli. Healthy control subjects demonstrate memory for the repeated stimuli by spending more time looking at the novel images, i.e., novelty preference. Here, we report an application of machine learning methods from computer science to improve the accuracy of detecting MCI by modeling eye movement characteristics such as fixations, saccades, and re-fixations during the VPC task. These characteristics are represented as features provided to automatic classification algorithms such as Support Vector Machines (SVMs). Using the SVM classification algorithm, in tandem with modeling the patterns of fixations, saccade orientation, and regression patterns, our algorithm was able to automatically distinguish age-matched normal control subjects from MCI subjects with 87% accuracy, 97% sensitivity and 77% specificity, compared to the best available classification performance of 67% accuracy, 60% sensitivity, and 73% specificity when using only the novelty preference information. These results demonstrate the effectiveness of applying machine-learning techniques to the detection of MCI, and suggest a promising approach for detection of cognitive impairments associated with other disorders.

On the early diagnosis of Alzheimer's Disease from multimodal signals: A survey

“…It is thought that saccades are of particular interest because they are very related to attention and thus, they are likely to be disturbed by cognitive impairments associated with neurodegenerative disorders such as AD, as well as by dysfunctions related purely to oculomotor execution [219]. All these behaviours can be measured easily and in a non-invasive way in a laboratory, making the patients carry out specific tasks like the reflexive paradigm, the memory-guided paradigm [219], the gap/overlap task or the antisaccade task [220] while their eyes are being tracked by cameras or infrared systems [221] and image processing techniques. Reading tasks have also been used [222].…”

“…Lagun et al [221] are about the only ones taking advantage of these biomarkers for AD recognition. They used the visual paired comparison task, which consists on identifying the eye movement pattern of the patients while introducing new visual stimulus.…”

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

“…The final results of the classification algorithm performance are reported, for each task, determining the mean values of a priori defined evaluation metrics (Costa et al, 2007;Lagun et al, 2011;Sokolova & Lapalme, 2009). …”

“…However, whether oculomotor performance parameters can be used to classify individuals according to their disease stage remains to be seen. Recently, this question has been studied in other neurodegenerative and neuropsychiatric disorders (Lagun, Manzanares, Zola, Buffalo, & Agichtein, 2011;Tseng et al, 2013), but it still remains unexplored in HD. A variety of experimental paradigms and several machinelearning algorithms have been applied to resolve these classification problems: (1) support vector machines (SVMs; Coco & Keller, 2014;Jin, Cheng, Zhou, & Li, 2012;Lagun et al, 2011;Rivera et al, 2012;Xiang, Yan, & Chen, 2009), (2) support vector machines with recursive feature elimination (SVM-RFE; Tseng et al, 2013), (3) Fisher linear discriminant analysis (Jin et al, 2012), (4) the multinomial log-linear neuronal networks model (Jin et al, 2012), (5) least-square angle regression (Coco & Keller, 2014;Jin et al, 2012), (6) naive Bayes (Lagun et al, 2011), and (7) logistic regression (Lagun et al, 2011).…”

“…For example, Tseng et al (2013) achieved an accuracy of 89.6 % with Parkinson's disease, and Lagun et al (2011) of 87 % with mild cognitive impairment. The variety of paradigms used is also reflected in these studies: The former (Tseng et al, 2013) developed a novel approach to analyze eye movements recorded in natural viewing settings, whereas in the latter (Lagun et al, 2011) a visual paired recognition task was applied, proving its sensitivity in identifying functional changes in medial temporal lobe structures.…”