Voxel2Mesh: 3D Mesh Model Generation from Volumetric Data

Wickramasinghe, Udaranga; Remelli, Edoardo; Knott, Graham; Fua, Pascal

doi:10.1007/978-3-030-59719-1_30

Cited by 75 publications

(74 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Implicit methods require a time-consuming topology correction, while explicit methods can pre-define an initial mesh with spherical topology to achieve fast inference. Wickramasinghe et al [19] presented an explicit framework, called Voxel2Mesh, to extract 3D meshes from medical images. Voxel2Mesh employed a series of deformation and unpooling layers to deform an initial mesh while increasing the number of vertices iteratively.…”

Section: Related Workmentioning

confidence: 99%

“…We adopt a local convolutional operation to extract the local feature of a vertex from brain MRI scans. Rather than using a memory-intensive 3D CNN on the entire MRI volume [19], this method only employs a CNN on a cube containing MRI intensity of each vertex and its neighborhood. As illustrated in Figure 1, for each vertex, we find the corresponding position in the brain MRI volume.…”

Section: Deformation Blockmentioning

confidence: 99%

“…As a fast and end-to-end alternative approach, deep learning has shown its advantages in surface reconstruction for general shape objects [7,9,13,14,18] and medical images [2,10,16,19]. Given brain MRI scans, existing deep learning frameworks [2,10] are able to predict cortical surfaces within 30 minutes.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

PialNN: A Fast Deep Learning Framework for Cortical Pial Surface Reconstruction

Ma¹,

Kainz²,

Rueckert³

et al. 2021

Preprint

View full text Add to dashboard Cite

Traditional cortical surface reconstruction is time consuming and limited by the resolution of brain Magnetic Resonance Imaging (MRI). In this work, we introduce Pial Neural Network (PialNN), a 3D deep learning framework for pial surface reconstruction. PialNN is trained end-to-end to deform an initial white matter surface to a target pial surface by a sequence of learned deformation blocks. A local convolutional operation is incorporated in each block to capture the multi-scale MRI information of each vertex and its neighborhood. This is fast and memory-efficient, which allows reconstructing a pial surface mesh with 150k vertices in one second. The performance is evaluated on the Human Connectome Project (HCP) dataset including T1-weighted MRI scans of 300 subjects. The experimental results demonstrate that Pi-alNN reduces the geometric error of the predicted pial surface by 30% compared to state-of-the-art deep learning approaches. The codes are publicly available at https://github.com/m-qiang/PialNN.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Deformation Blockmentioning

confidence: 99%

See 1 more Smart Citation

PialNN: A Fast Deep Learning Framework for Cortical Pial Surface Reconstruction

Ma¹,

Kainz²,

Rueckert³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Polygon meshes are direct and effective surface representations of object shape, when compared to voxels. Geometric learning on meshes has only recently been explored (Kolotouros et al, 2019;Litany et al, 2018;Ranjan et al, 2018;Wang et al, 2018;Wickramasinghe et al, 2020) for shape completion, non-linear facial morphable model generation and classification, 3D surface segmentation, and reconstruction from 2D/3D images. A novel representation learning and generative DL framework using spiral convolution on fixed topology meshes, was established with Neural3DMM by Bouritsas et al (2019) and later improved upon with SpiralNet++ by Gong et al (2019).…”

Section: Introductionmentioning

confidence: 99%

Analyzing Brain Morphology in Alzheimer’s Disease Using Discriminative and Generative Spiral Networks

Azcona

Besson

et al. 2021

Preprint

View full text Add to dashboard Cite

Several patterns of atrophy have been identified and strongly related to Alzheimer's disease (AD) pathology and its progression. Morphological changes in brain shape have been identified up to ten years before clinical diagnoses of AD, making its early diagnosis more desirable. We propose novel geometric deep learning frameworks for the analysis of brain shape in the context of neurodegeneration caused by AD. Our deep neural networks learn low-dimensional shape descriptors of multiple neuroanatomical structures, instead of handcrafted features for each structure. A discriminative network using spiral convolution on 3D meshes is constructed for the in-vivo binary classification of AD from healthy controls (HCs) using a fast and efficient "spiral" convolution operator on 3D triangular mesh surfaces of human brain subcortical structures extracted from T1-weighted magnetic resonance imaging (MRI). Our network architecture consists of modular learning blocks using residual connections to improve overall classifier performance. In this work: (1) a discriminative network is used to analyze the efficacy of disease classification using input data from multiple brain structures and compared to using a single hemisphere or a single structure. It also outperforms prior work using spectral graph convolution on the same the same tasks, as well as alternative methods that operate on intermediate point cloud representations of 3D shapes. (2) Additionally, visual interpretations for regions on the surface of brain structures that are associated to true positive AD predictions are generated and fall in accordance with the current reports on the structural localization of pathological changes associated to AD. (3) A conditional generative network is also implemented to analyze the effects of phenotypic priors given to the model (i.e. AD diagnosis) in generating subcortical structures. The generated surface meshes by our model indicate learned morphological differences in the presence of AD that agrees with the current literature on patterns of atrophy associated to the disease. In particular, our inference results demonstrate an overall reduction in subcortical mesh volume and surface area in the presence of AD, especially in the hippocampus. The low-dimensional shape descriptors obtained by our generative model are also evaluated in our discriminative baseline comparisons versus our discriminative network and the alternative shape-based approaches.

show abstract

“…Binary volumes prohibit interpolation of vertex positions during marching cubes, leading to strong aliasing which can be partially compensated for [3,4,5]. Surface defect detection has been demonstrated using various means such as neural networks [6,7] and local binary patterns [8]. Soussen et al are able to reconstruct perfectly binary objects directly from projections [9].…”

Section: Introductionmentioning

confidence: 99%

Reducing Stair Artifacts in CT Reconstruction

Wedekind

Oertel²,

Castillo

et al. 2021

2021 IEEE International Conference on Image Processing (ICIP)

View full text Add to dashboard Cite

Computed Tomography is increasingly employed for nondestructive evaluation, with the aim of reconstructing a surface mesh of a scanned object from radiographic projections. State-of-the-art algorithms first reconstruct a voxel grid and then extract a surface mesh using existing meshing algorithms, often leading to stair-like aliasing artifacts along the grid axes, due to the grid's orientation-dependent resolution. We circumvent such artifacts in filtered backprojection reconstructions by optimizing the mesh's vertex positions using information taken directly from the projections, rather than from a voxel grid. We show that our approach reduces stair artifacts both visibly and measurably, at relatively little additional computational cost. Our method can be tied into existing mesh extraction algorithms and removes stair artifacts almost entirely.

show abstract

Voxel2Mesh: 3D Mesh Model Generation from Volumetric Data

Cited by 75 publications

References 16 publications

PialNN: A Fast Deep Learning Framework for Cortical Pial Surface Reconstruction

PialNN: A Fast Deep Learning Framework for Cortical Pial Surface Reconstruction

Analyzing Brain Morphology in Alzheimer’s Disease Using Discriminative and Generative Spiral Networks

Reducing Stair Artifacts in CT Reconstruction

Contact Info

Product

Resources

About