Abstract:Measuring cross-sectional areas in ultrasound images is a standard tool to evaluate disease progress or treatment response. Often addressed today with supervised deep-learning segmentation approaches, existing solutions highly depend upon the quality of experts' annotations. However, the annotation quality in ultrasound is anisotropic and position-variant due to the inherent physical imaging principles, including attenuation, shadows, and missing boundaries, commonly exacerbated with depth. This work proposes a novel approach that guides ultrasound segmentation networks to account for sonographers' uncertainties and generate predictions with variability similar to the experts. We claim that realistic variability can reduce overconfident predictions and improve physicians' acceptance of deep-learning cross-sectional segmentation solutions. Our method provides CM's certainty for each pixel for minimal computational overhead as it can be precalculated directly from the image. We show that there is a correlation between low values in the confidence maps and expert's label uncertainty. Therefore, we propose to give the confidence maps as additional information to the networks. We study the effect of the proposed use of ultrasound CMs in combination with four state-of-the-art neural networks and in two configurations: as a second input channel and as part of the loss. We evaluate our method on 3D ultrasound datasets of the thyroid and lower limb muscles. Our results show ultrasound CMs increase the Dice score, improve the Hausdorff and Average Surface Distances, and decrease the number of isolated pixel predictions. Furthermore, our findings suggest that ultrasound CMs improve the penalization of uncertain areas in the ground truth data, thereby improving problematic interpolations. Our code and example data will be made public at <a class="link-external link-https" href="https://github.com/IFL-CAMP/Confidence-segmentation" rel="external noopener nofollow">this https URL</a>.

Fusion for Evaluation of Image Classification in Uncertain Environments

Fusion de classifieurs pour la classification d'images sonar

An Image Segmentation Method Based on Transformer and Multi-Scale Feature Fusion for UAV Marine Environment Monitoring

Experts Fusion and Multilayer Perceptron Based on Belief Learning for Sonar Image Classification

Look and Listen: A Multi-modality Late Fusion Approach to Scene Classification for Autonomous Machines

Ultrafusion: Optimal Fuzzy Fusion in Land-Cover Segmentation Using Multiple Panchromatic Satellite Images

Navigating Uncertainty: Semantic-Powered Image Enhancement and Fusion

Comparison of two data fusion approaches for land use classification

A Quantitative Validation of Multi-Modal Image Fusion and Segmentation for Object Detection and Tracking

Classification of High Spatial Resolution Remote Sensing Images Based on Decision Fusion

Information Fusion on Oversegmented Images: An Application for Urban Scene Understanding.

A Novel Fusion Strategy for Segmentation of Multisensor Remote Sensing Images

Can ultrasound confidence maps predict sonographers' labeling variability?

On the Overconfidence Problem in Semantic 3D Mapping

Rethinking the Evaluation of Visible and Infrared Image Fusion

On the Calibration of Uncertainty Estimation in LiDAR-based Semantic Segmentation

A Novel Image Fusion Method of Multi-Spectral and SAR Images for Land Cover Classification

Elevation Information-Guided Multimodal Fusion Robust Framework for Remote Sensing Image Segmentation

Visible and Near Infrared Image Fusion Based on Texture Information

Improving Semantic Image Segmentation via Label Fusion in Semantically Textured Meshes

A Novel Remote Sensing Image Classification Scheme Based on Data Fusion, Multiple Features and Ensemble Learning