2020/05/25 by Konstantin Dobratulin, Dobratulin, Konstantin, Andrey Gaidel +9
Engineering · Medicine · #Engineering Technology and Methodologies #Medical and Biological Sciences
paper · pdf · doi:10.48550/arxiv.2005.12110
In this article we investigate the efficiency of deep learning algorithms in\nsolving the task of detecting anatomical reference points on radiological\nimages of the head in lateral projection using a fully convolutional neural\nnetwork and a fully convolutional neural network with an extended architecture\nfor biomedical image segmentation - U-Net. A comparison is made for the results\nof detection anatomical reference points for each of the selected neural\nnetwork architectures and their comparison with the results obtained when\northodontists detected anatomical reference points. Based on the obtained\nresults, it was concluded that a U-Net neural network allows performing the\ndetection of anatomical reference points more accurately than a fully\nconvolutional neural network. The results of the detection of anatomical\nreference points by the U-Net neural network are closer to the average results\nof the detection of reference points by a group of orthodontists.\n