A Novel Distance-Based Metric for Quality Assessment in Image Segmentation

2025-03-28

AI Summary

Quantitative evaluation of image segmentation has long suffered from limitations: pixel-counting metrics ignore geometric structure, while distance-based measures (e.g., Hausdorff distance) lack interpretability and cross-dataset comparability. To address these issues, we propose the Surface Consistency Coefficient (SCC), a novel surface-distance-field–based metric that explicitly models segmentation errors weighted by their geodesic distance to the ground-truth surface—thereby emphasizing spatial distribution characteristics in boundary-adjacent regions. SCC achieves explicit physical interpretability and scale invariance through principled error weighting and normalization. Extensive experiments on multi-scale synthetic and real-world medical images demonstrate that SCC effectively discriminates boundary errors from interior errors, achieves superior correlation with expert clinical ratings compared to Hausdorff distance and other state-of-the-art metrics, and exhibits numerical stability and consistent cross-structure comparability.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Calibration & Uncertainty Quantification
• Knowledge Representation and Reasoning: Qualitative Reasoning

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Security and Privacy: Large-scale security measurements
• User Modeling, Personalization and Recommendation: Metrics for user behavior and evaluating success

Abstract

The assessment of segmentation quality plays a fundamental role in the development, optimization, and comparison of segmentation methods which are used in a wide range of applications. With few exceptions, quality assessment is performed using traditional metrics, which are based on counting the number of erroneous pixels but do not capture the spatial distribution of errors. Established distance-based metrics such as the average Hausdorff distance are difficult to interpret and compare for different methods and datasets. In this paper, we introduce the Surface Consistency Coefficient (SCC), a novel distance-based quality metric that quantifies the spatial distribution of errors based on their proximity to the surface of the structure. Through a rigorous analysis using synthetic data and real segmentation results, we demonstrate the robustness and effectiveness of SCC in distinguishing errors near the surface from those further away. At the same time, SCC is easy to interpret and comparable across different structural contexts.

Problem

• Proposes a new metric for evaluating image segmentation quality
• Addresses limitations of traditional pixel-counting error metrics
• Measures spatial error distribution relative to structure surfaces

Innovation

• Introduces Surface Consistency Coefficient (SCC)
• Quantifies spatial error distribution near surfaces
• Ensures interpretability and cross-context comparability