Optical coherence tomography (OCT) has had significant success in the field of ophthalmology, where it is essential for both screening and diagnosis. Clinical ophthalmic OCT systems are primarily used as table-top instruments that requires the subject to align with the chinrest and be operated by qualified personnel. In order to perform OCT imaging on bedridden patients or on babies, a handheld model is essential. In handheld devices, eye movements and probe movements cause artifacts while recording OCT images, making interpretation and registration more difficult. As a result, there is a need for an OCT scanner with an automatic real-time eye-tracking system and a correction mechanism to compensate for such movements. This work aims at developing a scanner head employing a cutting-edge stereo edge camera equipped with an inertial measurement unit (IMU) for detecting rotations and motions with six degrees of freedom. In this work, an Intel RealSense D435i depth camera-based eye tracking is performed. A python-based code was developed to image the eye continuously, detect face landmarks with media pipe, process the eye features in each frame, identify the iris in each frame and a circle is marked over the iris which would move along with the iris. The algorithm is tested on various scenarios of face angle and head motion. The eye movement identification and tracking capability of the developed algorithm and its performance results are presented in this study.
|