(1) To use electrooculogram (EOG) to differentiate patients with Myasthenia Gravis (MG) from those with extra-ocular movement abnormalities due to other etiologies (non-MG group). (2) To test the sensitivity of EOG features to classifying the EOG data into Controls (free of extra-ocular abnormalities), MG patients, and a non-MG group.

Current electrodiagnostic methods used to diagnose MG vary but can often be invasive, difficult to access or have limited sensitivity and specificity. Our research group has previously demonstrated the potential of using features extracted from EOG signals as a screening method for MG detection.

Data was collected from 51 patients (21 controls, 10 non-MG, 20 MG) based on clinical assessment. Patients followed a video protocol, and we focused on horizontal saccades, extracting time and time-frequency domain features from EOG signals. Time and spectral features from left and right saccades were averaged. We evaluated the ability of each signal feature to differentiate MG patients from controls and non-MG patients (case A) and subclassified controls and non-MG patients (case B). Group separation was tested using a Student’s t-test.

We found a difference in the median duration of the saccadic eye movement. For case A, we found that the range of horizontal eye movements in MG patients was restricted, resulting in the width of the eye movements being smaller than in the controls and the non-MG group. Likewise, we found the non-MG group to exhibit a relative reduction in horizontal eye movements compared to the controls. Statistical significance was established in both cases with a p-value ≤ 0.05.

Time and spectral domain features from EOG signals can detect MG and differentiate it from other conditions with extra-ocular abnormalities. This expands our previous work, using new EOG signal features and streamlined data collection.