Abstract:

One of the most significant factors affecting supervised neural network training is the precision of the annotations. Also, in a case of expert group, the problem of inconsistent data annotations is an integral part of real-world supervised learning processes, well-known to researchers. One practical example is a weak ground truth delineation for medical image segmentation.
In this paper, we have developed a new method of accurate segmentation of blood vessels based on a convolutional neural network. We focused on imperfect annotations for the semantic segmentation of blood vessels and introduced a concept of uncertainty masks and loss masking. These uncertainty masks can be created roughly by non-experts, which makes annotation process cheaper and faster. Quantitative results of our method on a real-world problem with missing annotations and on a perfectly labeled data set with artificially introduced noise are presented. Models trained with loss masking seem to be more robust regardless of the number of removed vessels. Noise robustness of four different model architectures has been tested and compared to the loss masking method, which turned out