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

• Published in: NeuroImage (Orlando, Fla.) Vol. 54; no. 2; pp. 940 - 954
• Main Authors: Coupé, Pierrick, Manjón, José V., Fonov, Vladimir, Pruessner, Jens, Robles, Montserrat, Collins, D. Louis
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 15.01.2011; Elsevier Limited; Elsevier
• Subjects: Alzheimer Disease - pathology; Alzheimer's disease; Applications; Bioengineering; Brain; Computer Science; Engineering Sciences; Experiments; Hippocampus; Hippocampus - anatomy & histology; Humans; Image Interpretation, Computer-Assisted - methods; Image Processing; Lateral ventricles; Lateral Ventricles - anatomy & histology; Libraries; Life Sciences; Machine Learning; Magnetic Resonance Imaging; Medical Imaging; Medical research; Methods; MRI; Neurodegenerative diseases; Pathology; Segmentation; Signal and Image Processing; Statistics; Structure segmentation; Ventricle; Ventricle (lateral); Ventricles (cerebral); Brain; Image processing; MRI; Structure segmentation; Alzheimer's disease; Hippocampus; Lateral ventricles; image processing; patch-based segmentation; hippocampus; nonlocal means; segmentation; ventricle; aging; brain
• ISSN: 1053-8119; 1095-9572; 1095-9572
• DOI: 10.1016/j.neuroimage.2010.09.018

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. ►The Nonlocal means estimator can be used to accurately segment anatomical brain structures such as hippocampus and lateral ventricles. ►Contrary to template warping methods working at the structure level, the proposed method handles a finer scale by using patches. ►Nonlocal patch-based segmentation does not require nonlinear registrations while providing state-of-the-art results.