A Precision-Recall Criterion Based Consensus Model For Fusing Multiple Segmentations

Abstract

This paper presents a general framework for seamlessly combining multiple low cost and inaccurate estimated segmentation maps (with an arbitrary number of regions) of the same scene to achieve a final improved segmentation. The proposed fusion model is derived from the well-known precision-recall criterion, specially dedicated to the specific clustering problem of any spatially indexed data and which is also efficient and widely used in the vision community for evaluating both a region-based segmentation and the quality of contours produced by this segmentation map compared to one or multiple ground-truth segmentations of the same image. The proposed combination framework is here specifically designed to be robust with respect to outlier segmentations (that appear to be inconsistent with the remainder of the segmentation ensemble) and includes an explicit internal regularization factor reflecting the inherent ill-posed nature of the segmentation problem. We propose also a hierarchical and efficient way to optimize the consensus energy function related to this fusion model that exploits a simple and deterministic iterative relaxation strategy combining the different segments or individual regions belonging to the segmentation ensemble in the final solution. The experimental results on the Berkeley database with manual ground truth segmentations show the effectiveness of our combination model.

Conflict management in the fusion of complementary segmentations of deformed kidneys and nephroblastoma

“…This metric is commonly used in the medical field. The F-Measure (Martin et al (2004)) is also based on the Dice for the segmentation fusion (Mignotte & Hélou (2014)) at the region level. In the present approach, 220 the Dice criterion is used for the segmentation fusion at the pixel level.…”

“…The Shape-Based Averaging (SBA) method (Rohlfing & Maurer (2007)) uses the signed Euclidean distance maps in order to determine the segmentation contours of each possible label. F-Measure Martin et al (2004) is a criterion used in Mignotte & Hélou (2014) which evaluates the quality of the resulting contours and uses the notions of precision and recall.…”

Fusion of multiple segmentations of medical images using OV2ASSION and Deep Learning methods: Application to CT-Scans for tumoral kidney

“…Many methods have emerged with the use of different metrics, via an iterative algorithm (Iterated conditional modes) with the Variation of Information 4 (VoI) criterion [12,13]; with the F-Measure (or precision-recall criterion) [14];…”

A fusion method based on Deep Learning and Case-Based Reasoning which improves the resulting medical image segmentations

“…However, these methods are limited. Many methods have emerged with the use of different metrics, via an iterative algorithm, for merging segmentations, such as using the Variation of Information (VoI) criterion (Mignotte (2014); Nguyen et al (2018)), the F-Measure (or precision-recall criterion) (Mignotte & Hélou (2014)), the Global Consistency Error (GCE) (Khelifi & Mignotte (2016)) or the Probabilistic Rand Index (PRI) measure (Mignotte (2010)). More recently, other fusion approaches use the combination of multiple metrics like the VoI and the F-Measure criteria and the GCE and the F-Measure criteria (Khelifi & Mignotte (2017a,b)).…”