Purpose:
To model patterns of visual distortion caused by metamorphopsia in a patient, by use of a deformable Amsler grid, based on cubic B-splines; and to localize the affected area in macular OCT images for further analysis.
Methods:
In order to model distorted vision, we display in a computer monitor an interactively deformable Amsler Grid to the patient at a distance of 30 cm. One eye is covered and the other fixates the grid center. The perceived distorted lines will be deformed in the opposite direction of patient’s distortion until vision is corrected and the lines are seen straight again. The grid is based on cubic B-splines in order to have a smooth deformation. Once the metamorphopsia model is acquired, we align it with the corresponding OCT images of the patient’s macula for further analysis. This was tested on seven patients with macular disease.
Results:
Five patients could reliably fulfill the task. The system provides the model of distorted vision in a range of 47.55º horizontal and 31.38º vertical visual angles. The grid lines are placed every 5.52º horizontal and 3.62º vertical visual angle. The perceived distorted vision was modeled and the affected area was localized in macular OCT and fundus images (see figure). The pathology matched the location of the distortion as graded by an experienced ophthalmologist. The system provides a measurement of the deformation seen in the grid (i.e., the average displacement of pixels) in order to have an estimation of the size of the affected area in the macula (in mm) according to the patients’ visual perception; which helps to evaluate the progression of the disease. The resolution of the system can be adapted to the patient’s degree of distortion.
Conclusions:
A deformable Amsler grid based system provides a simple and useful method for modeling distorted vision in patients with metamorphopsia. By aligning the model with macular OCT images, we can identify macular features, which could help in the development of an automatic method to model metamorphopsia.
Keywords: computational modeling • low vision • image processing