An Object-Based Approach for Detecting Small Brain Lesions: Application to Virchow-Robin Spaces

Descombes, Xavier; Kruggel, Frithjof; Wollny, Gert; Gertz, Hermann‐Josef

doi:10.1109/tmi.2003.823061

Cited by 51 publications

(54 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…So far, only a few studies have focused on automatic PVS segmentation from MR images in an unsupervised manner. For example, Descombes et al [18] enhanced the PVSs with filters and used a region-growing approach to get initial segmentation, followed by a geometry prior constraint for further improving the segmentation accuracy. Wuerfel et al [4] segmented the PVSs with a semi-automatic software by adjusting intensity threshold.…”

Section: Introductionmentioning

confidence: 99%

Structured Learning for 3-D Perivascular Space Segmentation Using Vascular Features

Zhang

Gao

Park

et al. 2017

IEEE Trans. Biomed. Eng.

View full text Add to dashboard Cite

Objective The goal of this paper is to automatically segment perivascular spaces (PVSs) in brain from high-resolution 7T MR images. Methods We propose a structured-learning-based segmentation framework to extract the PVSs from high-resolution 7T MR images. Specifically, we integrate three types of vascular filter responses into a structured random forest for classifying voxels into two categories, i.e., PVS and background. In addition, we propose a novel entropy-based sampling strategy to extract informative samples in the background for training an explicit classification model. Since the vascular filters can extract various vascular features, even thin and low-contrast structures can be effectively extracted from noisy backgrounds. Moreover, continuous and smooth segmentation results can be obtained by utilizing patch-based structured labels. Results The performance of our proposed method is evaluated on 19 subjects with 7T MR images, with the Dice similarity coefficient reaching 66%. Conclusion The joint use of entropy-based sampling strategy, vascular features, and structured learning can improve the segmentation accuracy. Significance Instead of manual annotation, our method provides an automatic way for PVS segmentation. Moreover, our method can be potentially used for other vascular structure segmentation because of its data-driven property.

show abstract

Section: Introductionmentioning

confidence: 99%

Structured Learning for 3-D Perivascular Space Segmentation Using Vascular Features

Zhang

Gao

Park

et al. 2017

IEEE Trans. Biomed. Eng.

View full text Add to dashboard Cite

show abstract

“…Wuerfel et al (Wuerfel, Haertle et al 2008) segmented the PVSs by using a semi-automatic software, which can adjust intensity threshold (Makale, Solomon et al 2002). Descombes et al (Descombes, Kruggel et al 2004) constructed a model defined by the pre-defined PVS filters and geometric properties, and then optimized it by the Markov chain Monte Carlo method. Uchiyama et al (Uchiyama, Kunieda et al 2008) enhanced the intensities of tubular structures using white top-hat transformation, and then extracted them by intensity thresholding.…”

Section: Introductionmentioning

confidence: 99%

Segmentation of perivascular spaces in 7 T MR image using auto-context model with orientation-normalized features

et al. 2016

View full text Add to dashboard Cite

Quantitative study of perivascular spaces (PVSs) in brain magnetic resonance (MR) images is important for understanding the brain lymphatic system and its relationship with neurological diseases. One of major challenges is the accurate extraction of PVSs that have very thin tubular structures with various directions in three-dimensional (3D) MR images. In this paper, we propose a learning-based PVS segmentation method to address this challenge. Specifically, we first determine a region of interest (ROI) by using the anatomical brain structure and the vesselness information derived from eigenvalues of image derivatives. Then, in the ROI, we extract a number of randomized Haar features which are normalized with respect to the principal directions of the underlying image derivatives. The classifier is trained by the random forest model that can effectively learn both discriminative features and classifier parameters to maximize the information gain. Finally, a sequential learning strategy is used to further enforce various contextual patterns around the thin tubular structures into the classifier. For evaluation, we apply our proposed method to the 7T brain MR images scanned from 17 healthy subjects aged from 25 to 37. The performance is measured by voxel-wise segmentation accuracy, cluster- wise classification accuracy, and similarity of geometric properties, such as volume, length, and diameter distributions between the predicted and the true PVSs. Moreover, the accuracies are also evaluated on the simulation images with motion artifacts and lacunes to demonstrate the potential of our method in segmenting PVSs from elderly and patient populations. The experimental results show that our proposed method outperforms all existing PVS segmentation methods.

show abstract

“…Towards computational segmentation of enlarged PVS in brain, there are a few other methods being investigated (Hernandez Mdel, Piper et al 2013), including a fully automatic method (Descombes, Kruggel et al 2004) based on complex shape models and the marked point process framework. Comparing the performance between our quantitation method and the previous published ones would be valuable, albeit, it is beyond the focus of this study on testing the feasibility of quantitative PVS MRI at 7T.…”

Section: Discussionmentioning

confidence: 99%

The feasibility of quantitative MRI of perivascular spaces at 7 T

Cai

Tain

Das

et al. 2015

Journal of Neuroscience Methods

View full text Add to dashboard Cite

Background Dilated brain perivascular spaces (PVSs) are found to be associated with many conditions, including aging, dementia, and Alzheimer's disease (AD). Conventionally, PVS assessment is mainly based on subjective observations of the number, size and shape of PVSs in MR images collected at clinical field strengths (≤ 3T). This study tests the feasibility of imaging and quantifying brain PVS with an ultrahigh 7T whole-body MRI scanner. New Method 3D high resolution T2-weighted brain images from healthy subjects (n=3) and AD patients (n=5) were acquired on a 7T whole-body MRI scanner. To automatically segment the small hyperintensive fluid-filling PVS structures, we also developed a quantitative program based on algorithms for spatial gradient, component connectivity, edge-detection, k-means clustering, etc, producing quantitative results of white matter PVS volume densities. Results The 3D maps of automatically segmented PVS show an apparent increase in PVS density in AD patients compared to age-matched healthy controls due to the PVS dilation (8.0 ± 2.1 v/v% in AD vs. 4.9 ± 1.3% in controls, p<0.05). Comparison with Existing Method We demonstrated that 7T provides sufficient SNR and resolution for quantitatively measuring PVSs in deep white matter that is challenging with clinical MRI systems (≤ 3T). Compared to the conventional visual counting and rating for the PVS assessment, the quantitation method we developed is automatic and objective. Conclusions Quantitative PVS MRI at 7T may serve as a non-invasive and endogenous imaging biomarker for diseases with PVS dilation.

show abstract

An Object-Based Approach for Detecting Small Brain Lesions: Application to Virchow-Robin Spaces

Cited by 51 publications

References 27 publications

Structured Learning for 3-D Perivascular Space Segmentation Using Vascular Features

Structured Learning for 3-D Perivascular Space Segmentation Using Vascular Features

Segmentation of perivascular spaces in 7 T MR image using auto-context model with orientation-normalized features

The feasibility of quantitative MRI of perivascular spaces at 7 T

Contact Info

Product

Resources

About