An Averaged Intensity Difference Detection Algorithm for Identification of Human Gingival Sulcus in Optical Coherence Tomography Images

In the past decade, there has been an increase in the development of sensitive, high-resolution, non-invasive diagnostic methods for periodontic diseases. Optical coherence tomography (OCT) has attracted considerable attention in clinical settings. In this paper, a reliable, robust algorithm for the detection of gingival sulcus in 2D OCT cross-sectional images is proposed. Previously, the measurement of gingival sulcus in OCT images has been performed by manual identification using two-dimensional (2D) cross-sectional images. The automated detection of gingival sulcus continuity in 2D OCT images may help medical practitioners to assess important features of gingival tissues. The Sobel and canny operators have mainly been used for boundary and edge detection in OCT images. However, these algorithms are highly sensitive to noise and speckle in OCT images. To overcome these limitations, we propose an algorithm for the quantitative depth measurement of the human gingival sulcus, based on averaged intensity difference. In this paper, we utilized two commercially-available swept-source OCT systems operating at center wavelengths of 1310 and 1060 nm to image gingival sulcus of human samples in vivo. The images were processed using three algorithms: canny, Sobel, and averaged intensity difference.