Active Contour Method Modification for Diabetic Macular Edema Area Quantification in OCT Images

A new method for automated quantification of diabetic macular edema on optical coherence tomograms based on modification of the active contour model is proposed. Traditional active contour models require manual initialization of the contour and are computationally expensive. This paper presents a modification of the IAC that uses a fast Fourier transform to accelerate the iterative segmentation process and automatic loop initialization. The model is reformulated using a threshold function that defines the contour as a region boundary, and includes a Fourier function of the pressure force with sign to account for global image features and a gradient force for local adaptation. A spectral residual approach based on the analysis of the frequency spectrum of the image is applied for automatic initialization. The proposed method combines the advantages of the frequency approach and regional segmentation, providing automatic initialization, robustness to background inhomogeneities, and high-speed performance comparable to traditional active contour models despite processing the entire image. This modification of the active contour model allows efficient and automatic extraction of diabetic macular edema boundaries on optical coherence tomograms, eliminating the need for manual initialization and ensuring the practical applicability of the method for quantitative assessment of the diabetic macular edema area.