Multimodal MRI segmentation of ischemic stroke lesions

Kabir, Yacine; Dojat, Michel; Scherrer, Benoît; Garbay, Catherine; Forbes, Florence

doi:10.1109/iembs.2007.4352610

Cited by 71 publications

(59 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Atlas prior probabilities and MRFs have been used to discard partial volume effects and reduce outliers in patients with multiple sclerosis lesions [ 33 ] . Kabir and colleagues have used multimodal MR data (T2, fl uid attenuated inversion recovery (FLAIR), diffusion sequences) and longitudinal data (6 h, 5 and 30 days) that were coregistered to form a multimodal MRF [ 35 ] . Using this novel approach, they were able to demonstrate that in adults with stroke, the lesion distribution follows the vascular territory of an occluded blood vessel.…”

Section: Markov Random Fieldsmentioning

confidence: 99%

“…Finally, compared to other neurological disorders, there has been little research dedicated to automated detection and quantifi cation of ischemic injury [ 35 ] .…”

Section: Recently Emerging Approaches For Automated Lesion Detectionmentioning

confidence: 99%

“…One should be able to determine whether an anatomical region is recovering faster than another, whether the lesion is shifting in the anatomical space, how the MR statistics (e.g., mean, standard deviation) of the lesion changes across anatomy and over time, etc. Despite research using automated methods to identify brain lesions, very limited research has considered the importance of the spatiotemporal evolution of the lesion [ 35 ] . Kabir and colleagues have qualitatively shown that adult stroke shifts from one anatomical region to other regions without further quantitative evaluation.…”

Section: Monitoring Spatiotemporal Interactions Between Stroke Tissuementioning

confidence: 99%

See 2 more Smart Citations

Computational Analysis: A Bridge to Translational Stroke Treatment

Ghosh

Sun

Turenius

et al. 2012

Translational Stroke Research

View full text Add to dashboard Cite

Objective rapid quantifi cation of injury using computational methods can improve the assessment of the degree of stroke injury, aid in the selection of patients for early or specifi c treatments, and monitor the evolution of injury and recovery. In this chapter, we use neonatal ischemia as a case-study of the application of several computational methods that in fact are generic and applicable across the age and disease spectrum. We provide a summary of current computational approaches used for injury detection, including Gaussian mixture models (GMM), Markov random fi elds (MRFs), normalized graph cut, and K-means clustering. We also describe more recent automated approaches to segment the region(s) of ischemic injury including hierarchical region splitting, support vector machine, a brain symmetry/asymmetry integrated model, and a watershed method that are robust at different developmental stages. We conclude with our assessment of probable future research directions in the fi eld of computational noninvasive stroke analysis such as automated detection of the ischemic core and penumbra, monitoring of implanted neuronal stem cells in the ischemic brain, injury localization specifi c to different brain anatomical regions, and quantifi cation of stroke evolution, recovery and spatiotemporal interactions between injury volume/severity and treatment. Computational analysis is expected to open a new horizon in current clinical and translational stroke research by exploratory data mining that is not detectable using the standard "methods" of visual assessment of imaging data.

show abstract

Section: Markov Random Fieldsmentioning

confidence: 99%

“…Finally, compared to other neurological disorders, there has been little research dedicated to automated detection and quantifi cation of ischemic injury [ 35 ] .…”

Section: Recently Emerging Approaches For Automated Lesion Detectionmentioning

confidence: 99%

Section: Monitoring Spatiotemporal Interactions Between Stroke Tissuementioning

confidence: 99%

See 1 more Smart Citation

Computational Analysis: A Bridge to Translational Stroke Treatment

Ghosh

Sun

Turenius

et al. 2012

Translational Stroke Research

View full text Add to dashboard Cite

show abstract

“…The lesion volume extracted by the automated method showed high correlation with either the T2W or DWI volume across different time-points. Kabir et al [9] suggested an automatic segmentation method based on MRF from multimodal MRI images comprising of DWI, T2W and FLAIR. The primary objective of the method was to categorize the stroke sub-types at the acute, sub-acute and later stages by registering with the blood supply territories atlas, therefore no quantitative results were presented related to the accuracy of the stroke detection.…”

Section: Introductionmentioning

confidence: 99%

Classification Forests and Markov Random Field to Segment Chronic Ischemic Infarcts from Multimodal MRI

Mitra

Bourgeat

Fripp

et al. 2013

Multimodal Brain Image Analysis

View full text Add to dashboard Cite

Abstract. Accurate identification of ischemic lesions and brain atrophy is critical in the management of stroke patients and may serve as an important biomarker in studying post-stroke depression. In this paper we present an automated method to identify chronic ischemic infarcts in gray matter and gray/white matter partial volume areas that may be used to measure the amount of tissue loss due to atrophy in the area. The measure of tissue loss may then be used as a potential biomarker in analyzing the relation between stroke and post-stroke depression. The automated segmentation method relies on Markov random field (MRF) and random forest based classifications. The MRF classification identifies the possible lesion areas from the fluid attenuated inversion recovery (FLAIR) magnetic resonance (MR) images. Thereafter, the multimodal (T1-, T2-weighted, FLAIR, and apparent diffusion coefficient (ADC)) MR images of the possible lesion areas are fed into the classification forests along with other context-aware features and probability maps of the gray and white matter regions. The results of classification from the MRF and the classification forests are finally combined using connected component analysis to identify the final lesion area. The accuracy of the method in identifying infarcted regions from multimodal MR images has been validated on 17 patient datasets with a mean accuracy of 99%, a mean positive predictive value (PPV) of 75% and a mean negative predictive value (NPV) of 99% and a volume correlation of r = 0.98.

show abstract

“…Moreover, creating and evaluating automatic methods for segmenting ischemic lesions is difficult because the ground truth is often operator or rater dependent and stroke lesions are widely varied. Accurate segmentation of stroke lesions also requires anatomical knowledge [6].…”

Section: Introductionmentioning

confidence: 99%

Automatic processing of diffusion-weighted ischemic stroke images based on divergence measures: slice and hemisphere identification, and stroke region segmentation

Prakash

Gupta

et al. 2008

Int J CARS

View full text Add to dashboard Cite

Objective Fast, accurate and automatic segmentation of acute ischemic stroke lesions is important for clinical trials and has potential for efficient stroke management. To identify stroke slices, stroke hemisphere, and segment stroke regions in diffusion-weighted magnetic resonance imaging (DWI), divergence based algorithms are proposed. Materials and methodsThe study used 57 DWI volumes with inter 0.94-2.42 mm and intra 5-14 mm plane resolutions. We used ratio of intensity probability density functions (pdf) as a divergence measure. For slice identification, this measure is the ratio of pdfs of slice and the volume; for hemisphere and infarct segmentation, it is the ratio of the difference of the pdfs of the left and right hemispheres to the sum of their pdfs. The median and cross over points are thresholds for slice and region segmentation while hemisphere identification is threshold free. Descriptive statistics were determined and ROC analysis was performed. Results The median sensitivity, specificity, and Dice statistical index for segmentation are 86.34%, 99.83%, 0.72, respectively. For slice and hemisphere identification sensitivity and specificity are (90.05%; 68.78%) and (94.74%; 94.74%), respectively. The algorithm implemented in VC++ takes 3-5 s per volume. Conclusion This automatic, accurate and fast method is potentially useful in clinical setting and clinical trials to localize and quantify the stroke regions, eliminate inter-and

show abstract

Multimodal MRI segmentation of ischemic stroke lesions

Cited by 71 publications

References 17 publications

Computational Analysis: A Bridge to Translational Stroke Treatment

Computational Analysis: A Bridge to Translational Stroke Treatment

Classification Forests and Markov Random Field to Segment Chronic Ischemic Infarcts from Multimodal MRI

Automatic processing of diffusion-weighted ischemic stroke images based on divergence measures: slice and hemisphere identification, and stroke region segmentation

Contact Info

Product

Resources

About