Segmentation of a First Generation Agent Bubbles in the B-Mode Echocardiographic Images

First-generation contrast agents are used in echocardiography to diagnose PFO (Patent Foramen Ovale). The degree of PFO is based on the number of bubbles found in the left atrium. This paper presents research on the preparation of an algorithm for the segmentation of first-generation contrast bubbles. To fulfill the research objective, the images were first preprocessed to reduce the influence of possible noise and artefacts; subsequently, binarization was performed and the segmentation of bubbles then occurred. The results were evaluated by the Dice coefficient. To obtain the best results, the following local thresholding methods were compared: Median, Sauvola, and SDA. The outcome was evaluated and the best values of the Dice coefficient were obtained with two local thresholding methods: the Median and the Sauvola algorithms.