Rationale, design and methodology of the image analysis protocol for studies of patients with cerebral small vessel disease and mild stroke

Valdés-Hernández, Maria del C.; Armitage, Paul A.; Thrippleton, Michael J.; Chappell, Francesca M; Sandeman, Elaine M.; Maniega, Susana Muñoz; Shuler, Kirsten; Wardlaw, Joanna M.

doi:10.1002/brb3.415

Cited by 70 publications

(111 citation statements)

References 61 publications

Supporting

Mentioning

111

Contrasting

Order By: Relevance

“…We used data from 264 patients who gave written informed consents to participate in a study of lacunar stroke mechanisms [28].…”

Section: Subjects and Mri Protocolmentioning

confidence: 99%

“…The study that provided data for this manuscript [28] included patients with lacunar stroke and/or minor cortical strokes which were clinically evident, and did not consider diabetes, hypertension and other vascular risk factors as criteria for exclusion. However, it excluded patients with other nonvascular neurological disorders, major medical conditions including renal failure, contraindications to MRI, unable to give consent, and those who had haemorrhagic Brain MRI was conducted at baseline (i.e.…”

Section: Subjects and Mri Protocolmentioning

confidence: 99%

See 1 more Smart Citation

Reliability of an automatic classifier for brain enlarged perivascular spaces burden and comparison with human performance

et al. 2017

View full text Add to dashboard Cite

Correspondence: Víctor González-Castro (victor.gonzalez@ed.ac.uk) or María del C. Valdés Hernández (M. Valdes-Hernan@ed.ac.uk) In the brain, enlarged perivascular spaces (PVS) relate to cerebral small vessel disease (SVD), poor cognition, inflammation and hypertension. We propose a fully automatic scheme that uses a support vector machine (SVM) to classify the burden of PVS in the basal ganglia (BG) region as low or high. We assess the performance of three different types of descriptors extracted from the BG region in T2-weighted MRI images: (i) statistics obtained from Wavelet transform's coefficients, (ii) local binary patterns and (iii) bag of visual words (BoW) based descriptors characterizing local keypoints obtained from a dense grid with the scale-invariant feature transform (SIFT) characteristics. When the latter were used, the SVM classifier achieved the best accuracy (81.16%). The output from the classifier using the BoW descriptors was compared with visual ratings done by an experienced neuroradiologist (Observer 1) and by a trained image analyst (Observer 2). The agreement and cross-correlation between the classifier and Observer 2 (κ = 0.67 (0.58-0.76)) were slightly higher than between the classifier and Observer 1 (κ = 0.62 (0.53-0.72)) and comparable between both the observers (κ = 0.68 (0.61-0.75)). Finally, three logistic regression models using clinical variables as independent variable and each of the PVS ratings as dependent variable were built to assess how clinically meaningful were the predictions of the classifier. The goodness-of-fit of the model for the classifier was good (area under the curve (AUC) values: 0.93 (model 1), 0.90 (model 2) and 0.92 (model 3)) and slightly better (i.e. AUC values: 0.02 units higher) than that of the model for Observer 2. These results suggest that, although it can be improved, an automatic classifier to assess PVS burden from brain MRI can provide clinically meaningful results close to those from a trained observer.

show abstract

“…We used data from 264 patients who gave written informed consents to participate in a study of lacunar stroke mechanisms [28].…”

Section: Subjects and Mri Protocolmentioning

confidence: 99%

Section: Subjects and Mri Protocolmentioning

confidence: 99%

Reliability of an automatic classifier for brain enlarged perivascular spaces burden and comparison with human performance

et al. 2017

View full text Add to dashboard Cite

show abstract

“…(2015). To differentiate the lacunes and infarcts (new and old) from the surrounding brain on the FLAIR image, we considered a threshold equal to 3.5 SD in pixel intensity, respectively, below or above the mean of the fitted intensity distribution for the normal‐appearing white matter (Hernandez et al., 2010).…”

Section: Methodsmentioning

confidence: 99%

“…(2015). To report and analyze the probability density distribution (PD) per location we used two complementary approaches: (1) we identified the regions of maximum PD and noted the location guided by a standard atlas (Duvernoy, 1999) and (2) we developed a computational ROI template by combining the subcortical structures extracted using FSL FIRST together with 2 × 2 × 2 mm region of interest (ROI) placed on five main white matter pathways selected from the Johns Hopkins University DTI white matter atlas (anterior horns of the lateral ventricles, optic radiations, centrum semiovale bordering the corona radiata, external capsules, and retrolentiform part of the internal capsules; Oishi et al., 2008), and then determined the PD at each computationally determined ROI.…”

Section: Methodsmentioning

confidence: 99%

Interhemispheric characterization of small vessel disease imaging markers after subcortical infarct

et al. 2016

Self Cite

View full text Add to dashboard Cite

BackgroundIn structural Magnetic Resonance Imaging (MRI) of patients with a recent small subcortical infarct (RSSI) and small vessel disease (SVD) imaging markers coexist. However, their spatial distribution and prevalence with respect to the hemisphere of the RSSI remain unknown.Materials and MethodsFrom brain MRI in 187 patients with an acute lacunar ischemic stroke clinical syndrome and a relevant diffusion weighted imaging (DWI)‐positive lesion, we semiautomatically extracted the RSSI, microbleeds, lacunes, old cortical infarcts, and white matter hyperintensities (WMH) using optimized thresholding in the relevant sequences, and rated the load of perivascular spaces. We registered all images to an age‐relevant brain template and calculated the probability distribution of all SVD markers mentioned for patients who had the RSSI in each hemisphere separately. We used the Wilcoxon and chi‐squared tests to compare the volumes and frequencies of occurrence, respectively, of the SVD markers between hemispheres throughout the sample.ResultsFifty‐two percent patients (n = 97) had the RSSI in the left hemisphere, 42% (n = 78) in the right, 2.7% (n = 5) in both, and 3.7% (n = 7) in the cerebellum or brainstem. There was no significant difference in RSSI frequency between left and right hemispheres (p = .10) in the sample. The median volume of the RSSI (expressed as a percentage of the total intracranial volume) was 0.05% (IQR = 0.06). There was no difference in median percent volume of the right RSSIs versus left (p = .16). Neither was there a significant interhemispheric difference in the volume of any of the SVD markers regardless of the location of the RSSI and they were equally distributed in both hemispheres.ConclusionAssessment of SVD imaging markers in the contralateral hemisphere could be used as a proxy for the SVD load in the whole brain to avoid contamination by the RSSI of the measurements, especially of WMH.

show abstract

“…It took 152 and 392 min on average for the training process and 5 and 27 s on average for testing process for CEN and patch-CNN respectively. An image analyst can take from 15 to 60 min to segment WMH on a single dataset depending on the level of experience [40].…”

Section: Running Time Evaluationmentioning

confidence: 99%

Deep Learning vs. Conventional Machine Learning: Pilot Study of WMH Segmentation in Brain MRI with Absence or Mild Vascular Pathology

et al. 2017

View full text Add to dashboard Cite

In the wake of the use of deep learning algorithms in medical image analysis, we compared performance of deep learning algorithms, namely the deep Boltzmann machine (DBM), convolutional encoder network (CEN) and patch-wise convolutional neural network (patch-CNN), with two conventional machine learning schemes: Support vector machine (SVM) and random forest (RF), for white matter hyperintensities (WMH) segmentation on brain MRI with mild or no vascular pathology. We also compared all these approaches with a method in the Lesion Segmentation Tool public toolbox named lesion growth algorithm (LGA). We used a dataset comprised of 60 MRI data from 20 subjects in the Alzheimer's Disease Neuroimaging Initiative (ADNI) database, each scanned once every year during three consecutive years. Spatial agreement score, receiver operating characteristic and precision-recall performance curves, volume disagreement score, agreement with intra-/inter-observer reliability measurements and visual evaluation were used to find the best configuration of each learning algorithm for WMH segmentation. By using optimum threshold values for the probabilistic output from each algorithm to produce binary masks of WMH, we found that SVM and RF produced good results for medium to very large WMH burden but deep learning algorithms performed generally better than conventional ones in most evaluations.

show abstract

Rationale, design and methodology of the image analysis protocol for studies of patients with cerebral small vessel disease and mild stroke

Cited by 70 publications

References 61 publications

Reliability of an automatic classifier for brain enlarged perivascular spaces burden and comparison with human performance

Reliability of an automatic classifier for brain enlarged perivascular spaces burden and comparison with human performance

Interhemispheric characterization of small vessel disease imaging markers after subcortical infarct

Deep Learning vs. Conventional Machine Learning: Pilot Study of WMH Segmentation in Brain MRI with Absence or Mild Vascular Pathology

Contact Info

Product

Resources

About