Interactive 3D brain vessel segmentation from an example

Abstract

Segmentation of cerebral vascular networks from 3D angiographic data remains a challenge. Automation generally induces a high computational cost and possible errors, while interactive methods are hard to use due to the dimension and complexity of images. This article presents a compromise between both approaches, by using the concept of examplebased segmentation. Segmentation examples of vascular structures are involved in a scheme relying on connected filtering, in order to obtain an interactive -but strongly assistedsegmentation method. This strategy, which uses componenttrees in a non-standard fashion, leads to promising results, when applied on cerebral MR angiographic data.

Filtering and segmentation of 3D angiographic data: Advances based on mathematical morphology

“…The (discrete) direction e 1 of B(x) is approximated from the images of regularized direction fields, I The SE-based adjunct dilation, resulting in the image δ B (I in ), is followed by the adjunct erosion ε B . Both computations (δ B and ε B ), whose results are formally defined by Equations (19) and (16) respectively, then provide the final filtering result F(I in ) = ϕ B (I in ) with a low algorithmic cost (see Proposition 4). Also note that this processing ensures idempotence, guaranteeing that the filter obeys morphological rules.…”

“…Section 4, which is an extended and improved version of the conference articles [71,70], describes a vessel filtering method based on a hybrid strategy, merging both new spatially variant mathematical morphology algorithms and derivative-based approaches. Section 5, which is an extended and improved version of the conference articles [56,19], describes an example-based interactive vessel segmentation method relying on a component-tree-based technique. Section 6 describes and discusses experimental results related to vessel segmentation and filtering performed on angiographic phantom images and in vivo cerebral MRA data.…”

From 2D markers in MIP to 3D vessel segmentation: A fuzzy paradigm for connected filtering

“…Let M ⊆ Ω be a binary vascular marker defined on the support Ω of the image I. In the method initially proposed in [10], the vascular volume S ⊆ Ω segmented from I is obtained by computing the subset N ⊆ N of nodes of T that "fit at best" the vascular marker M , which is assumed to provide a first approximation of the sought vessels.…”

“…In [9], Caldairou et al proposed to consider vectorial attributes for node selection, leading a segmentation method that requires a classification process to discriminate the relevant nodes in a wide parameter space. In [10], we investigated a different methodology, that no longer uses a "local" description of each node. The main idea is to come back to the definition of the component tree, and to consider -more globally-the nodes as the generators of the image.…”

“…To this end, we first show (Sec. 2.1) that the energy-based formulation of the method developed in [10] can be extended to fuzzy examples/markers. We then explain (Sec.…”