Image Segmentation with Partial Convexity Shape Prior Using Discrete Conformality Structures

Siu, Chun Yin; Chan, Hei Long; Lui, Ronald Lok Ming

doi:10.1137/19m129718x

Cited by 11 publications

(2 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The process consists in learning the weights of the network through the optimisation of a differentiable loss function, such as the Dice loss or the cross-entropy, with a gradient descent. Nevertheless, incorporating a small amount of high-level information in the segmentation process (shape prior knowledge [2], topology preservation enforcement [3], prescription of the number of connected components/holes [4], (partial) convexity [5]) proves to achieve more accurate results. Motivated by this observation and inspired by the work of Peng et al [6], we propose enforcing geometrical constraints in the training of the convolutional neural network by designing a suitable loss function which, in addition to intensity pairing, includes a criterion of edge alignment through a weighted total variation term, an area penalisation, and a component ensuring intensity homogeneity, yielding a non-smooth non-convex optimisation problem.…”

Section: Introductionmentioning

confidence: 99%

A Geometrically-Constrained Deep Network For Ct Image Segmentation

Lambert

Guyader²,

Petitjean³

2021

2021 IEEE 18th International Symposium on Biomedical Imaging (ISBI)

View full text Add to dashboard Cite

Incorporating prior knowledge into a segmentation process, whether it be geometrical constraints such as volume penalisation, (partial) convexity enforcement, or topological prescriptions to preserve the contextual relations between objects, proves to improve accuracy in medical image segmentation, in particular when addressing the issue of weak boundary definition. Motivated by this observation, the proposed contribution aims to include geometrical constraints in the training of convolutional neural networks in the form of a penalty in the loss function. These geometrical constraints take several forms and encompass level curve alignment through the weighted total variation component, an area penalisation phrased as a hard constraint in the modelling, and an intensity homogeneity criterion based on a combination of the standard Dice loss with the piecewise constant Mumford-Shah model. The mathematical formulation yields a non-smooth non-convex optimisation problem, which rules out conventional smooth optimisation techniques and leads us to adopt a Lagrangian setting. The application falls within the scope of organ-at-risk segmentation in CT (Computed Tomography) images, in the context of radiotherapy planning. Experiments demonstrate that our method provides significant improvements over existing non-constrained approaches.

show abstract

Section: Introductionmentioning

confidence: 99%

A Geometrically-Constrained Deep Network For Ct Image Segmentation

Lambert

Guyader²,

Petitjean³

2021

2021 IEEE 18th International Symposium on Biomedical Imaging (ISBI)

View full text Add to dashboard Cite

show abstract

“…This method has shown to be successful to achieve accurate segmentation results, even for degraded 2D images. The method is further extended to image segmentation problem with convexity prior enforced [55]. Nevertheless, this model suffers from a major limitation.…”

Section: Introductionmentioning

confidence: 99%

Topology-Preserving 3D Image Segmentation Based On Hyperelastic Regularization

Zhang

Lui

2021

Preprint

View full text Add to dashboard Cite

Image segmentation is to extract meaningful objects from a given image. For degraded images due to occlusions, obscurities or noises, the accuracy of the segmentation result can be severely affected. To alleviate this problem, prior information about the target object is usually introduced. In [10], a topology-preserving registration-based segmentation model was proposed, which is restricted to segment 2D images only. In this paper, we propose a novel 3D topologypreserving registration-based segmentation model with the hyperelastic regularization, which can handle both 2D and 3D images. The existence of the solution of the proposed model is established. We also propose a converging iterative scheme to solve the proposed model. Numerical experiments have been carried out on the synthetic and real images, which demonstrate the effectiveness of our proposed model.

show abstract

Recent Development of Medical Shape Analysis via Computational Quasi-Conformal Geometry

Chan¹,

Lui²

2021

Handbook of Mathematical Models and Algorithms in Computer Vision and Imaging

View full text Add to dashboard Cite

Image Segmentation with Partial Convexity Shape Prior Using Discrete Conformality Structures

Cited by 11 publications

References 32 publications

A Geometrically-Constrained Deep Network For Ct Image Segmentation

A Geometrically-Constrained Deep Network For Ct Image Segmentation

Topology-Preserving 3D Image Segmentation Based On Hyperelastic Regularization

Recent Development of Medical Shape Analysis via Computational Quasi-Conformal Geometry

Contact Info

Product

Resources

About