The length and thickness of the uterus and endometrium are morphology characteristics as important measuresfor uterine diagnosis. In diagnosing uterine, doctors mark anatomical landmark points of uterus and endometriumin order to measure their length and thickness. However, it is difficult to reliably detect the landmarks of theuterus and endometrium due to the ambiguous boundaries and heterogeneous textures of uterus transvaginalultrasound image. In this paper, we propose a novel region-guided adversarial learning framework for anatomicallandmark detection in transvaginal ultrasound image, aiming at automatically detecting the landmark points ofuterus and endometrium of transvaginal ultrasound image to a diagnostical precision. In the proposed adver-sarial learning scheme, the proposed framework consists of a landmark predictor and two discriminators for theuterus and endometrium. The proposed landmark predictor is to detect the desired landmarks of both uterusand endometrium regions from transvaginal ultrasound image. The discriminator is to determine whether thepredicted landmarks of uterus and endometrium are related with their regions or not (i.e., whether the predictedlandmark points are on the region boundaries or not.). By adversarial learning between the predictor and thediscriminators with uterus and endometrium region images, the performance of the landmark predictor can beimproved. In testing, with the trained predictor only, uterus and endometrium landmarks are predicted. Exper-imental results demonstrated that the proposed method achieved a high accuracy in detecting landmarks of theuterus and endometrium in the ultrasound image.
展开▼
