Quality control (QC) of structures segmentation in volumetric medical images is important for identifying segmentationerrors in clinical practice and for facilitating model development by enhancing network performanc...
详细信息
Quality control (QC) of structures segmentation in volumetric medical images is important for identifying segmentationerrors in clinical practice and for facilitating model development by enhancing network performance in semi-supervised and active learning scenarios. This paper introduces SegQC, a novel framework for segmentation quality estimation and segmentation error detection. SegQC computes an estimate measure of the quality of a segmentation in volumetric scans and in their individual slices and identifies possible segmentationerror regions within a slice. The key components of SegQC include: 1) SegQC-Net, a deep network that inputs a scan and its segmentation mask and outputs segmentationerror probabilities for each voxel in the scan;2) three new segmentation quality metrics computed from the segmentationerror probabilities;3) a new method for detecting possible segmentationerrors in scan slices computed from the segmentationerror probabilities. We introduce a novel evaluation scheme to measure segmentationerror discrepancies based on an expert radiologist's corrections of automatically produced segmentations that yields smaller observer variability and is closer to actual segmentationerrors. We demonstrate SegQC on three fetal structures in 198 fetal MRI scans - fetal brain, fetal body and the placenta. To assess the benefits of SegQC, we compare it to the unsupervised Test Time Augmentation (TTA)-based QC and to supervised autoencoder (AE)-based QC. Our studies indicate that SegQC outperforms TTA-based quality estimation for whole scans and individual slices in terms of Pearson correlation and MAE for fetal body and fetal brain structures segmentation as well as for volumetric overlap metrics estimation of the placenta structure. Compared to both unsupervised TTA and supervised AE methods, SegQC achieves lower MAE for both 3D and 2D Dice estimates and higher Pearson correlation for volumetric Dice. Our segmentation error detection method achieved r
Automatic segmentation of organs-at-risk (OARs) in CT scans using convolutional neural networks (CNNs) is being introduced into the radiotherapy workflow. However, these segmentations still require manual editing and ...
详细信息
ISBN:
(纸本)9783031164439;9783031164422
Automatic segmentation of organs-at-risk (OARs) in CT scans using convolutional neural networks (CNNs) is being introduced into the radiotherapy workflow. However, these segmentations still require manual editing and approval by clinicians prior to clinical use, which can be time consuming. The aim of this work was to develop a tool to automatically identify errors in 3D OAR segmentations without a ground truth. Our tool uses a novel architecture combining a CNN and graph neural network (GNN) to leverage the segmentation's appearance and shape. The proposed model was trained using data-efficient learning using a synthetically-generated dataset of segmentations of the parotid gland with realistic contouring errors. The effectiveness of our model was assessed with ablation tests, evaluating the efficacy of different portions of the architecture as well as the use of transfer learning from a custom pretext task. Our best performing model predicted errors on the parotid gland with a precision of 85.0% & 89.7% for internal and external errors respectively, and recall of 66.5% & 68.6%. This offline QA tool could be used in the clinical pathway, potentially decreasing the time clinicians spend correcting contours by detecting regions which require their attention. All our code is publicly available at https://*** rrr-uom-projects/contour_auto_QATool.
Medical diagnosis software and computer-assisted surgical systems often use segmented image data to help clinicians make decisions. The segmentation extracts the region of interest from the background, which makes the...
详细信息
Medical diagnosis software and computer-assisted surgical systems often use segmented image data to help clinicians make decisions. The segmentation extracts the region of interest from the background, which makes the visualization clearer. However, no segmentation method can guarantee accurate results under all circumstances. As a result, the clinicians need a solution that enables them to check and validate the segmentation accuracy as well as displaying the segmented area without ambiguities.
With the method presented in this paper, the real CT or MR image is displayed within the segmented region and the segmented boundaries can be expanded or contracted interactively. By this way, the clinicians are able to check and validate the segmentation visually and make more reliable decisions. After experiments with real data from a hospital, the presented method is proved to be suitable for efficiently detecting segmentationerrors. The new algorithm uses new graphic processing uint （GPU） shading functions recently introduced in graphic cards and is fast enough to interact oil the segmented area, which was not possible with previous methods.
暂无评论