Effects of Registration Regularization and Atlas Sharpness on Segmentation Accuracy

Yeo, B. T. Thomas; Sabuncu, Mert R.; Desikan, Rahul S.; Fischl, Bruce; Golland, Polina

doi:10.1007/978-3-540-75757-3_83

Cited by 37 publications

(46 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A more elaborate way of tuning this parameter would be to evaluate the registration result for a variety of values and choose the best tradeoff. Such an approach is for example clearly presented in (Yeo et al, 2008a). The same set of parameters is used for all the experiments: a maximum step …”

Section: Experiments: Practical Advantage Of the Symmetric Forcesmentioning

confidence: 99%

Diffeomorphic demons: Efficient non-parametric image registration

et al. 2009

View full text Add to dashboard Cite

We propose an efficient non-parametric diffeomorphic image registration algorithm based on Thirion's demons algorithm. In the first part of this paper, we show that Thirion's demons algorithm can be seen as an optimization procedure on the entire space of displacement fields. We provide strong theoretical roots to the different variants of Thirion's demons algorithm. This analysis predicts a theoretical advantage for the symmetric forces variant of the demons algorithm. We show on controlled experiments that this advantage is confirmed in practice and yields a faster convergence. In the second part of this paper, we adapt the optimization procedure underlying the demons algorithm to a space of diffeomorphic transformations. In contrast to many diffeomorphic registration algorithms, our solution is computationally efficient since in practice it only replaces an addition of displacement fields by a few compositions. Our experiments show that in addition to being diffeomorphic, our algorithm provides results that are similar to the ones from the demons algorithm but with transformations that are much smoother and closer to the gold standard, available in controlled experiments, in terms of Jacobians.

show abstract

Section: Experiments: Practical Advantage Of the Symmetric Forcesmentioning

confidence: 99%

Diffeomorphic demons: Efficient non-parametric image registration

et al. 2009

View full text Add to dashboard Cite

show abstract

“…This extra information leads to better final shape and appearance models. Yeo et al [13] also register intensity and label images to build probabistic atlases.…”

Section: Computing Correspondencementioning

confidence: 99%

3D Brain Segmentation Using Active Appearance Models and Local Regressors

Babalola

Cootes

Twining

et al. 2008

Medical Image Computing and Computer-Assisted Intervention – MICCAI 2008

View full text Add to dashboard Cite

Abstract. We describe an efficient and accurate method for segmenting sets of subcortical structures in 3D MR images of the brain. We first find the approximate position of all the structures using a global Active Appearance Model (AAM). We then refine the shape and position of each structure using a set of individual AAMs trained for each. Finally we produce a detailed segmentation by computing the probability that each voxel belongs to the structure, using regression functions trained for each individual voxel. The models are trained using a large set of labelled images, using a novel variant of 'groupwise' registration to obtain the necessary image correspondences. We evaluate the method on a large dataset, and demonstrate that it achieves results comparable with some of the best published.

show abstract

“…In this study, the Gaussian fluid-like and diffusion-like smoothing parameters were chosen empirically; however, similar to (Yeo et al 2008), a more extensive study is still required to optimise and to analyse the effect of the smoothing parameters on the estimated displacement field and on the quantitative accuracy of the reconstructed activity image. As illustrated in figure 4, noise on the emission data propagates into the MLRR attenuation map as uncertainties on the position of the contours in the deformed attenuation image.…”

Section: Discussionmentioning

confidence: 99%

Simultaneous reconstruction of the activity image and registration of the CT image in TOF-PET

Rezaei

Michel²,

Casey³

et al. 2016

Phys. Med. Biol.

View full text Add to dashboard Cite

Abstract. Previously, maximum-likelihood methods have been proposed to jointly estimate the activity image and the attenuation image or the attenuation sinogram from time-of-flight (TOF) positron emission tomography (PET) data. In this contribution, we propose a method that addresses the possible alignment problem of the TOF-PET emission data and the computed tomography (CT) attenuation data, by combining reconstruction and registration. The method, called MLRR, iteratively reconstructs the activity image while registering the available CT-based attenuation image, so that the pair of activity and attenuation images maximise the likelihood of the TOF emission sinogram. The algorithm is slow to converge, but some acceleration could be achieved by using Nesterov's momentum method and by applying a multiresolution scheme for the non-rigid displacement estimation. The latter also helps to avoid local optima, although convergence to the global optimum cannot be guaranteed. The results are evaluated on 2D and 3D simulations as well as a respiratory gated clinical scan. Our experiments indicate that the proposed method is able to correct for possible misalignment of the CT-based attenuation image, and is therefore a very promising approach to suppressing attenuation artefacts in clinical PET/CT. When applied to respiratory gated data of a patient scan, it produced deformations that are compatible with breathing motion and which reduced the well known attenuation artefact near the dome of the liver. Since the method makes use of the energy-converted CT attenuation image, the scale problem of joint reconstruction is automatically solved.

show abstract

Effects of Registration Regularization and Atlas Sharpness on Segmentation Accuracy

Cited by 37 publications

References 19 publications

Diffeomorphic demons: Efficient non-parametric image registration

Diffeomorphic demons: Efficient non-parametric image registration

3D Brain Segmentation Using Active Appearance Models and Local Regressors

Simultaneous reconstruction of the activity image and registration of the CT image in TOF-PET

Contact Info

Product

Resources

About