Robust Segmentation Models Using an Uncertainty Slice Sampling-Based Annotation Workflow

Semantic segmentation neural networks require pixel-level annotations in large quantities to achieve a good performance. In the medical domain, such annotations are expensive because they are time-consuming and require expert knowledge. Active learning optimizes the annotation effort by devising strategies to select cases for labeling that are the most informative to the model. In this work, we propose an uncertainty slice sampling (USS) strategy for the semantic segmentation of 3D medical volumes that selects 2D image slices for annotation and we compare it with various other strategies. We demonstrate the efficiency of USS on a CT liver segmentation task using multisite data. After five iterations, the training data resulting from USS consisted of 2410 slices (4% of all slices in the data pool) compared to 8121(13%), 8641(14%), and 3730(6%) slices for uncertainty volume (UVS), random volume (RVS), and random slice (RSS) sampling, respectively. Despite being trained on the smallest amount of data, the model based on the USS strategy evaluated on 234 test volumes significantly outperformed models trained according to the UVS, RVS, and RSS strategies and achieved a mean Dice index of 0.964, a relative volume error of 4.2%, a mean surface distance of 1.35mm, and a Hausdorff distance of 23.4mm. This was only slightly inferior to 0.967, 3.8%, 1.18mm, and 22.9mm achieved by a model trained on all available data. Our robustness analysis using the 5th percentile of Dice and the 95th percentile of the remaining metrics demonstrated that USS not only resulted in the most robust model compared to other strategies, but also outperformed the model trained on all data according to the 5th percentile of Dice (0.946 vs. 0.945) and the 95th percentile of mean surface distance (1.92mm vs. 2.03mm).