Segmentation of anatomical structures from 3d brain mri using automatically-built statistical shape models

Abstract

HAL is a multidisciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L'archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d'enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Statistical shape models for 3D medical image segmentation: A review

“…Instead of a complete training from sample images, some authors prefer to explicitly specify the boundary appearance: Bailleul et al (2004) calculate four features along each profile, which are assumed to accurately describe the border (of the putamen in MRI datasets): maximum intensity difference between inside and outside voxels, intensity means difference, inner voxels regularity and inner-outer voxels regularity difference. All features are normalized to [0..1] and added to give a final estimate of the boundary fit.…”

Segmentation algorithms of subcortical brain structures on MRI for radiotherapy and radiosurgery: A survey

“…Multi atlas based Bondiau [3] Brainstem MR T1, T2 Al Shaikhli [56] Brainstem, cerebellum, ventricles MR T1 Multiple atlas-based Zarpalas [29] Hippocampus MR T1 Artaechevarria [30] Multi-structure MR Collins [31] Hippocampus, amygdala MR T1 Khan [32] Hippocampus MR T1 Kim [33] Hippocampus MR 7T Coupé [34] Multi-structure MR T1 Wang [35] Hippocampus MR Cardoso [36] Hippocampus MR T1 Panda [40] Optic nerve, eye globe CT Heckemann [51] Multi-structure MR T1 Aljabar [52] Multi-structure MR T1 Lötjönen [53] Multi-structure MR T1 Asman [54] Multi-structure MR Active shape models Bailleul [47] Multi-structure MR Tu [48] Multi-structure MR T1 Pitiot [79] Multi-structure MR T1 Zhao [80] Multi-structure MR Rao [81] Multi-structure MR Bernard [82] Subthalamic nucleus MR T1 Olveres [83] Mid brain MR T1, SWI Active appearance models Hu [26] Hippocampus, amygdala MR T1, T2 Duchesne [45] Medial temporal lobe MR T1 Hu [46] Medial temporal lobe MR T1 Cootes [49] Multi-structure MR Brejl [78] Corpus callosum, cerebellum MR Babalola [50,86] Multi-structure MR T1…”

“…Statistical models (SM) have become widely used in the field of computer vision and medical image segmentation over the past decade [26,[45][46][47][48][49][50][73][74][75][76][77][78][79][80][81][82][83][84][85][86][87][88]. Basically, SMs use a priori shape information to learn the variation from a suitably annotated training set, and constrain the search space to only plausible instances defined by the trained model.…”

“…ASM has been widely used for segmentation in medical imaging [73], including segmentation of subcortical structures on brain [47,48,[77][78][79][80][81][82][83][84]. It is based on a statistical shape model (SSM) to constrain the detected organ boundary to plausible shapes (i.e.…”

“…Segmentation of the corpus callosum has been also investigated by using different methods such as deformable models [42,43], or machine learning [44]. Other researchers have focused on a set of different subcortical and cerebellar brain structures, proposing several approaches: active shape and appearance models [45][46][47][48][49][50], atlas-based methods [51][52][53][54][55][56], deformable models [57][58][59] or machine learning approaches [60][61][62][63][64].…”