Abstract:Anatomically diffused tissues (ADTs) refer to soft tissues containing many anatomical regions that are spatially dispersed and structurally irregular. In magnetic resonance images, ADTs exhibit blurred morphology and heterogeneous texture, making the accurate extraction of their 3D anatomy challenging. Center-free fuzzy C-means (FCM) can effectively partition nonlinear or nonspherical clusters, providing a promising scheme for ADT segmentation. It solves the uncertainty arising from unreliable center estimation by introducing a similarity criterion. However, the similarity criterion is sensitive to the number of target objects and their adjacent members in the images. Moreover, memberships of the existing algorithms are susceptible to losing real ADT details. To handle these issues, we propose an edge-preserving constrained center-free FCM algorithm for segmenting 3D ADTs in magnetic resonance images. To overcome the sensitivity of the similarity criterion, a novel object-to-cluster similarity measure is first proposed to utilize refined member-toobject adjacency. Specifically, the similarity measure focuses on members in the feature space, which share approximately homogeneous characteristics with each target object. Gradient-domain edge-preserving filtering is then combined with the improved similarity criterion to construct the novel objective function of center-free FCM. With the assistance of the designed imagedriven edge-preserving regularization, the gradient information of clusters is constrained, eventually approaching that of ADTs in the guidance image. Experiments are conducted on two public brain datasets and one local intrahepatic vein dataset. The results demonstrate that the proposed algorithm is more effective for ADT segmentation than the state-of-the-art peers, exhibiting superior generalization capability.

A segmentation system based on clustering method for pediatric DTI images

Segmentation of Pediatric Diffusion Tensor Magnetic Resonance Imaging Images Based on Adaptive Mean Shift Clustering Method

Progress in Segmentation of Diffusion Tensor Magnetic Resonance Images

Unifying the analyses of anatomical and diffusion tensor images using volume-preserved warping.

Robust Fiber Clustering Of Cerebral Fiber Bundles In White Matter

Neonatal White Matter Damage Analysis Using DTI Super-Resolution and Multi-Modality Image Registration.

Brain Tissue Classification Based on DTI Using an Improved Fuzzy C-means Algorithm with Spatial Constraints.

DTI Registration in Atlas Based Fiber Analysis of Infantile Krabbe Disease

A hybrid approach to automatic clustering of white matter fibers.

Accurate Corresponding Fiber Tract Segmentation Via FiberGeoMap Learner with Application to Autism.

Fiber Segmentation Using A Density-Peaks Clustering Algorithm

A Robust Fiber Clustering Method for Tract Refinement

Review of Fast Density-Peaks Clustering and Its Application to Pediatric White Matter Tracts.

Brain Tissue Segmentation Based on DTI Data.

A Statistical Approach to Segmentation of Diffusion Tensor Imaging

Segmentation of 3D Anatomically Diffused Tissues in Magnetic Resonance Images Through Edge-Preserving Constrained Center-Free Fuzzy C-Means

Multimodality White Matter Tract Segmentation Using CNN.

Analysis of fiber similarity metric for fiber tract clustering in white matter of human brain

Multi-channel features based automated segmentation of diffusion tensor imaging using an improved FCM with spatial constraints.

Hierarchical Density-Based Clustering of White Matter Tracts in the Human Brain

Direct segmentation of brain white matter tracts in diffusion MRI