Method and a system for detection of eye gaze-pattern abnormalities and related neurological diseases

Inventors

De Villers-Sidani, Etienne • DROUIN-PICARO, Paul Alexandre • Desgagne, Yves

Assignees

Innodem Neurosciences

Abstract

The present disclosure relates to a method and a system for detecting a neurological disease and an eye gaze-pattern abnormality related to the neurological disease of a user. The method comprises displaying stimulus videos on a screen of an electronic device and simultaneously filming with a camera of the electronic device to generate a video of the user's face for each one of the stimulus videos, each one of the stimulus videos corresponding to a task. The method further comprises providing a machine learning model for gaze predictions, generating the gaze predictions for each video frame of the recorded video, and determining features for each task to detect the neurological disease using a pre-trained machine learning model.

Core Innovation

The disclosure describes performing a set of tasks for detecting a neurological disease, where each task is distinct from each other and corresponds to a distinct set of features for the task. The set of tasks includes a calibration task, a smooth pursuit task, an anti-saccade task, and at least one of a fixation task and a pro-saccade task. During task performance, stimulus videos are displayed on a screen of an electronic device while simultaneously filming a user’s face with a camera of the electronic device located in proximity to the screen to generate a video for each stimulus video.

The approach uses machine learning to generate gaze predictions for each video frame of the user's face for each task. Based on the generated gaze predictions, values of the set of features for each task are determined, and a pre-trained machine learning model is used to detect the neurological disease. The disclosure also states that based on the generated video for each task, values of the set of features can be determined using a first pre-trained machine learning model, and the neurological disease can be detected using a second pre-trained machine learning model.

The task set is framed with stimulus videos configured for extraction of distinct task-specific features, including a smooth pursuit stimulus video and an anti-saccade stimulus video. In the smooth pursuit task, a target is displayed in a sequence following a predetermined continuous path, appearing moving at a constant speed towards and from extremes of the screen so the user deliberately follows the target movement. In the anti-saccade task, a target is displayed during a fixation period, followed by a blank screen, and then a symbol is displayed at another location pointing to a first direction and together with three other symbols pointing to different directions so the user identifies where the symbol pointed during a stimulus period.

Claims Coverage

The partial document includes three independent methods. Across these, the inventive coverage centers on a distinct multi-task eye-gaze stimulus battery filmed with a camera proximate to the screen, machine learning for gaze predictions and/or feature value determination, and detecting a neurological disease using pre-trained machine learning models.

Across the independent claims, the core inventive approach is to run a multi-task stimulus battery while filming the user’s face with a camera proximate to the screen, use machine learning to obtain task feature values, and detect a neurological disease using pre-trained machine learning models.

Stated Advantages

Documented Applications

No documented applications found