Abstract:Quality control of structures segmentation in volumetric medical images is important for identifying segmentation errors in clinical practice and for facilitating model development. 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 segmentation error regions within a slice. The key components include: 1. SegQC-Net, a deep network that inputs a scan and its segmentation mask and outputs segmentation error probabilities for each voxel in the scan; 2. three new segmentation quality metrics, two overlap metrics and a structure size metric, computed from the segmentation error probabilities; 3. a new method for detecting possible segmentation errors in scan slices computed from the segmentation error probabilities. We introduce a new evaluation scheme to measure segmentation error discrepancies based on an expert radiologist corrections of automatically produced segmentations that yields smaller observer variability and is closer to actual segmentation errors. 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 quality estimation. Our studies indicate that SegQC outperforms TTA-based quality estimation in terms of Pearson correlation and MAE for fetal body and fetal brain structures segmentation. Our segmentation error detection method achieved recall and precision rates of 0.77 and 0.48 for fetal body, and 0.74 and 0.55 for fetal brain segmentation error detection respectively. SegQC enhances segmentation metrics estimation for whole scans and individual slices, as well as provides error regions detection.

An automatic segmentation framework of quasi-periodic time series through graph structure

A compact hybrid fully convolutional and BiLSTM network with squeeze and temporal excitation approach for semantic segmentation of ECG signal

Anomaly Detection in Quasi-Periodic Time Series based on Automatic Data Segmentation and Attentional LSTM-CNN

Knowledge Discovery Based Auxiliary Diagnosis Algorithm for Electrocardiograph On-Line Data

A Novel Quality Evaluating Method for Over-Segmentation Approaches Using Real-Time Boundary Information

[A Study on Clustering Analysis of Arrhythmias].

Epileptic State Segmentation with Temporal-Constrained Clustering.

Integrating Data-Driven Segmentation, Local Feature Extraction and Fisher Kernel Encoding to Improve Time Series Classification

ECG_SegNet: An ECG delineation model based on the encoder-decoder structure

Recurrent Aggregation Learning for Multi-View Echocardiographic Sequences Segmentation

A Causal Intervention Scheme for Semantic Segmentation of Quasi-periodic Cardiovascular Signals

On-line segmentation of time-series data

SegQC: a segmentation network-based framework for multi-metric segmentation quality control and segmentation error detection in volumetric medical images

Visibility graph-based segmentation of multivariate time series data and its application

An improved fast shapelet selection algorithm and its application to pervasive EEG

Quality Control-Driven Image Segmentation Towards Reliable Automatic Image Analysis in Large-Scale Cardiovascular Magnetic Resonance Aortic Cine Imaging

An adaptive time series segmentation algorithm based on visibility graph and particle swarm optimization

Time Series Segmentation Clustering: A New Method for S-Phase Picking in Microseismic Data

Robust Group Anomaly Detection for Quasi-Periodic Network Time Series

SIMTSeg: A self-supervised multivariate time series segmentation method with periodic subspace projection and reverse diffusion for industrial process

A Structure Constraint Matrix Factorization Framework for Human Behavior Segmentation