Diabetic macular edema is one of the most serious complications of diabetic retinopathy and the leading cause of central visual acuity loss in diabetic patients. Optical coherence tomography is the current standard for monitoring this condition and evaluating the efficacy of anti-VEGF therapy. This master's thesis focuses on the design and implementation of a method for automatic segmentation of the internal limiting membrane and retinal pigment epithelium in image data acquired with the Zeiss Cirrus HD-OCT device.

The proposed algorithm utilizes a combination of bilateral filtering for speckle noise reduction and adaptive thresholding to identify key retinal interfaces within axial intensity profiles. The method includes a specific strategy to eliminate the influence of the photoreceptor layer and a multi-stage post-processing routine to ensure lateral continuity of the profiles even in the presence of pathological changes. Based on the segmented boundaries, a retinal thickness map is generated, and clinical parameters are calculated within the sectors of the ETDRS grid.

Validation was performed on a dataset of 27 longitudinal clinical records from the ophthalmology department of the University Hospital Brno. The results demonstrated a high degree of correlation with the manufacturer's commercial software, with a correlation coefficient of $r = 0.994$ for central subfield thickness in records with sufficient signal quality. The thesis further analyzes the impact of signal quality on segmentation accuracy and discusses the causes of the identified systematic bias. The implemented method reliably captures the treatment response trend in all monitored patients, confirming its utility for objective quantification of disease progression in clinical practice and research.