Onurcan ŞAHİN, Erdinç ALTUĞ

Design and control of a visual servomechanism for automating corneal cross-linking treatment in keratoconus patients

A novel mechatronic system designed for automated corneal cross-linking (CCL) treatment in keratoconus patients is introduced. Keratoconus is a serious illness that if not treated may cause serious distortion of vision. Currently, the CCL operation, which is the most promising treatment of this disease, is performed manually. The automated system developed is the first of its kind to automate the treatment with visual feedback, and it aims to increase the efficiency in treatment and to eliminate any potential side effects and risks of the treatment. To track the eye of the patient, the system consists of a camera, an image processing algorithm developed on OpenCV sharp, a planar servomechanism system consisting of various mechanical and electronic components, and a PIC microcontroller that contains digital PID controllers. The proposed system and the algorithms are designed and simulated in Matlab, and then the system is manufactured and various experiments with an eye pattern and animal eyes are performed. The results are shown and discussed.

Anahtar Kelimeler:

Keratoconus treatment, eye tracking, system identification, digital PID design, corneal cross-linking

Design and control of a visual servomechanism for automating corneal cross-linking treatment in keratoconus patients

Keywords:

Keratoconus treatment, eye tracking, system identification, digital PID design, corneal cross-linking,

PDF

___

Figure 16. Processed frame under UVA irradiation.
Figure 17. Field experiment eye detection.
The developed prototype of the system is complete and working efficiently. Moreover, based on opinions
of our medical advisors, obtained response characteristics satisfy the demands of the CCL treatment. As far
as we know, this is the only automated CCL treatment system available in the literature; because of this, no
performance comparison could be performed. As a future work, we plan to improve the system with other sensors. With the addition of encoders,
additional and more accurate information to control position data can be obtained. Furthermore, this will
provide a better control ability to the system. Moreover, we plan to implement an IR camera in the system.
The pupil (black part of the eye) can easily be separated under IR LED illumination and the IR cameras would
give clearer visual data for the image processing algorithm to work with. In order to eliminate the calibration
step for each patient, a sensor will be used to estimate the correct distance to the patient’s eye. Alternatively,
the size of the detected eye could as well be used within the vision algorithm to perform self-calibration. One
other planned addition to the system is a single-board vision computer to eliminate the dependence on a PC in the field. Due to the nature of the treatment, the system needs to track only a single eye. For those patients
requiring treatment for both eyes, an alternative system can be developed. A system that can track two eyes
at the same time and compensate for movements would be a nice improvement. The real success of the prototype can only be seen by a comparable clinical study, where the treatment
with the proposed automated system will be compared to that of the classical manual treatment in a medical
setting. This will be the topic of further research, and has significant importance in refractive surgery. The
developed automated vision-based system will also give a chance to future medical clinical studies to improve
their results by eliminating human errors involved.