Patch-based segmentation using expert priors: Application to hippocampus and ventricle segmentation

By Pierrick Coupé, José Vicente Manjón, Vladimir S Fonov, Jens C. Pruessner, Montserrat Robles, D. Louis Collins. NeuroImage, 2011

737 citations recorded by the source corpus.

Quantitative magnetic resonance analysis often requires accurate, robust, and reliable automatic extraction of anatomical structures. Recently, template-warping methods incorporating a label fusion strategy have demonstrated high accuracy in segmenting cerebral structures. In this study, we propose a novel patch-based method using expert manual segmentations as priors to achieve this task. Inspired by recent work in image denoising, the proposed nonlocal patch-based label fusion produces accurate and robust segmentation. Validation with two different datasets is presented. In our experiments, the hippocampi of 80 healthy subjects and the lateral ventricles of 80 patients with Alzheimer's disease were segmented. The influence on segmentation accuracy of different parameters such as patch size and number of training subjects was also studied. A comparison with an appearance-based method and a template-based method was also carried out. The highest median kappa index values obtained with the proposed method were 0.884 for hippocampus segmentation and 0.959 for lateral ventricle segmentation.

Subfields: Biophysics, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/67082b10fc39ae4262170facfe5f517d02416231

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Biophysics

In Computer Vision and Pattern Recognition

Citations in the corpus

BEaST: Brain extraction based on nonlocal segmentation technique (2012, methodology): “As shown in Coupé et al. (2011), increasing the number of selected priors improves the segmentation accuracy with the average DSC increasing from 0.9797 ( N =2) to 0.9856 ( N =40).”
MUSE: MUlti-atlas region Segmentation utilizing Ensembles of registration algorithms and parameters, and locally optimal atlas selection (2015, positive): “With the exception of patch-based approaches that use affine registration to align atlas images (Coupé et al., 2011; Konukoglu et al., 2013), deformable image registration (Sotiras et al.”
SynthStrip: skull-stripping for any brain image (2022, positive): “The skull-stripping tool BEaST (Eskildsen et al., 2012) builds on patch-based, non-local segmentation techniques (Coup´e et al., 2010, 2011; Roy et al., 2017) and assigns a label to each voxel by comparing its local neighborhood to patches in a reference set with prior labels.”
DeepNAT: Deep Convolutional Neural Network for Segmenting Neuroanatomy (2017, methodology): “Good segmentation performance was reported for patch-based approaches following a nonlocal means strategy (Coupé et al., 2011; Rousseau et al., Preprint submitted to NeuroImage ar X iv :1 70 2.”
A review on automatic fetal and neonatal brain MRI segmentation (2017, positive): “(2012) method is a multi-resolution extension of the patch-based technique of (Coupé et al., 2011) for the brain extraction from adult MRI. Multi-atlas segmentation was also used by Tourbier et al. (2015). They applied multi-atlas fusion with global weighting based on the normalized correlation coef”
Evaluation of prostate segmentation algorithms for MRI: the PROMISE12 challenge (2013, methodology): “Finally, a patch-based local voxel weighting strategy is introduced, which was recently proposed for use in patch-based brain segmentation (Coupé et al., 2011) and improved by introducing the weight of the mapping agreement from atlas to target.”
Unsupervised domain adaptation for medical imaging segmentation with self-ensembling (2018, methodology): “Before the development of deep learning models, initial work was focused mostly on patch-based (Coup (cid:1) e et al., 2011) segmentation.”
The Multimodal Brain Tumor Image Segmentation Benchmark (BRATS) (2015, methodology): “We use exponential weights, based on patch similarity [90], [95], [97]; the decay parameter σ2 depends on the test case, and is set to the maximum of SSD between every voxel in the test case and every first-neighbor training patch.”
Deep 3D Convolutional Encoder Networks With Shortcuts for Multiscale Feature Integration Applied to Multiple Sclerosis Lesion Segmentation (2016, methodology): “[14] proposed a segmentation method based on an extension of the non-local means algorithms [8].”
A supervised patch-based approach for human brain labeling (2011, neutral): “The method described in this article has been developed irrespective of a similar patch-based approach recently proposed by Coupé et al. [10].”
Local Temperature Scaling for Probability Calibration (2020, methodology): “[8] Coupé, P.”
Multi-Atlas Segmentation of Biomedical Images: A Survey (2014, methodology): “An alternative label fusion strategy involves the use of patches to compute weights at each voxel, which can be used with a conventional label fusion method (Coup´e et al., 2011; Fonov et al., 2012).”