Automated ischemic lesion detection in a neonatal model of hypoxic ischemic injury

Ghosh, Nirmalya; Recker, Rebecca; Shah, Amul; Bhanu, Bir; Ashwal, Stephen; Obenaus, André

doi:10.1002/jmri.22488

Cited by 34 publications

(55 citation statements)

References 62 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We have previously reported that HRS can rapidly and precisely define T2-map volumes of ischemic injury in a rat-pup model of global ischemia. 15 Refinement of our computational approach now provides strong evidence that: (1) HRS can detect and quantify the core-penumbra in focal and global models of ischemia using ADC/T2-map data that match IHC measures; (2) HRS from a single MR modality (T2WI) is as good as DPM in detecting core-penumbra; and (3) HRS works equally well in term newborns with AIS. We have purposefully varied injury severity (from mild to severe) to show successful applicability of HRS to a broad range of data.…”

Section: Discussionmentioning

confidence: 91%

“…Detection of core-penumbra builds on our published HRS algorithm 15 by further dichotomizing the extracted injury regions (i.e., lesion). The HRS extraction of the lesion encompasses: (1) skull stripping; (2) removing background noise; (3) rescaling MR values to reduce computational complexity; (4) modeling the histogram of the rescaled MR values as a bimodal distribution with two distinct and distant peaks; (5) splitting the MR image into two subimages using the valley between these two peaks in the histogram; (6) recursively resplitting the bimodal distributions to generate the HRS tree; (7) stopping the recursive splitting based on a set of uniformity criteria; (8) converting the rescaled values back to actual MR values; and (9) extracting the lesion volume based on 'a priori' approximate mean MR values (using a threshold meanTh; Table 1).…”

Section: Hierarchical Region Splitting Computational Assessment Of Comentioning

confidence: 99%

“…For T2 maps, we used a meanTh ¼ 75 ms to detect ischemic lesions because ischemia increases T2 relaxation time. 15 The DWI/ADC lesions evolve and in the first 1 to 2 days, the HII lesion is bright (hyperintense) with DWI and dark (hypointense) on ADC maps while after 2 days the appearance reverses. 23 For ADC maps within 1 to 2 days after injury, HRS detects lesion regions with mean ADComeanTh ¼ 60 while after 2 days, HRS detects lesions with mean ADC4meanTh ¼ 150.…”

Section: Hierarchical Region Splitting Computational Assessment Of Comentioning

confidence: 99%

See 2 more Smart Citations

Automated Core–Penumbra Quantification in Neonatal Ischemic Brain Injury

Ghosh

Yuan

Turenius

et al. 2012

J Cereb Blood Flow Metab

Self Cite

View full text Add to dashboard Cite

Neonatal hypoxic-ischemic brain injury (HII) and arterial ischemic stroke (AIS) result in irreversibly injured (core) and salvageable (penumbral) tissue regions. Identification and reliable quantification of salvageable tissue is pivotal to any effective and safe intervention. Magnetic resonance imaging (MRI) is the current standard to distinguish core from penumbra using diffusionperfusion mismatch (DPM). However, subtle MR signal variations between core-penumbral regions make their visual delineation difficult. We hypothesized that computational analysis of MRI data provides a more accurate assessment of core and penumbral tissue evolution in HII/AIS. We used two neonatal rat-pup models of HII/AIS (unilateral and global hypoxic-ischemia) and clinical data sets from neonates with AIS to test our noninvasive, automated computational approach, Hierarchical Region Splitting (HRS), to detect and quantify ischemic core-penumbra using only a single MRI modality (T2-or diffusion-weighted imaging, T2WI/DWI). We also validated our approach by comparing core-penumbral images (from HRS) to DPM with immunohistochemical validation of HII tissues. Our translational and clinical data results showed that HRS could accurately and reliably distinguish the ischemic core from penumbra and their spatiotemporal evolution, which may aid in the vetting and execution of effective therapeutic interventions as well as patient selection.

show abstract

Section: Discussionmentioning

confidence: 91%

Section: Hierarchical Region Splitting Computational Assessment Of Comentioning

confidence: 99%

Section: Hierarchical Region Splitting Computational Assessment Of Comentioning

confidence: 99%

See 1 more Smart Citation

Automated Core–Penumbra Quantification in Neonatal Ischemic Brain Injury

Ghosh

Yuan

Turenius

et al. 2012

J Cereb Blood Flow Metab

Self Cite

View full text Add to dashboard Cite

show abstract

“…These Gaussian distributions can be closely spaced, i.e., the mean MR signal intensity of different tissues are close, so that on occasion, one needs to consider fuzzy connectedness where class membership values are like weights in GMM but with "fuzzyness," in which the probabilities of a voxel being in tissue A and tissue B are assigned [ 31 ] . However, histogram fi tting to a normalized histogram, as adapted by Fan and colleagues, cannot be used when large lesions affect the shape of the histogram [ 32 ] .…”

Section: Gaussian Mixed Modelsmentioning

confidence: 99%

“…42.1 ). ) and overlap (sensitivity: 0.82, specifi city: 0.86, similarity: 1.47) to manually extracted results [ 32 ] . Although the methods and results of HRS were obtained from MR images, HRS is a generic method for any medical image (e.g., MRI, PET, CT) where contrast in the medical image is used to detect abnormalities.…”

Section: Hierarchical Region Splittingmentioning

confidence: 99%

Computational Analysis: A Bridge to Translational Stroke Treatment

Ghosh

Sun

Turenius

et al. 2012

Translational Stroke Research

Self Cite

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

Acute intranasal osteopontin treatment in male rats following TBI increases the number of activated microglia but does not alter lesion characteristics

Jullienne

Hamer

Haddad

et al. 2019

J of Neuroscience Research

Self Cite

View full text Add to dashboard Cite

Intranasal recombinant osteopontin (OPN) has been shown to be neuroprotective in different models of acquired brain injury but has never been tested after traumatic brain injury (TBI). We used a model of moderate-to-severe controlled cortical impact in male adult Sprague Dawley rats and tested our hypothesis that OPN treatment would improve neurological outcomes, lesion and brain tissue characteristics, neuroinflammation, and vascular characteristics at 1 day post-injury. Intranasal OPN administered 1 hr after the TBI did not improve neurological score, lesion volumes, blood-brain barrier, or vascular characteristics. When assessing neuroinflammation, we did not observe any effect of OPN on the astrocyte reactivity but discovered an increased number of activated microglia within the ipsilateral hemisphere. Moreover, we found a correlation between edema and heme oxygenase-1 (HO-1) expression which was decreased in OPN-treated animals, suggesting an effect of OPN on the HO-1 response to injury. Thus, OPN may increase or accelerate the microglial response after TBI, and early response of HO-1 in modulating edema formation may limit the secondary consequences of TBI at later time points. Additional experiments and at longer time points are needed to determine if intranasal OPN could potentially be used as a treatment after TBI where it might be beneficial by activating protective signaling pathways. K E Y W O R D S controlled cortical impact, edema, extracellular matrix protein, Heme oxygenase-1, microglial activation, T2 mapping S U PP O RTI N G I N FO R M ATI O N Additional supporting information may be found online in the Supporting Information section at the end of the article. Supplementary Table S1. Brain regions excluded from T2 values analysis Transparent Science Questionnaire for Authors. How to cite this article: JullienneA, HamerM, HaddadE, et al. Acute intranasal osteopontin treatment in male rats following TBI increases the number of activated microglia but does not alter lesion characteristics. J Neuro Res. 2020;98:141-154.

show abstract

Automated ischemic lesion detection in a neonatal model of hypoxic ischemic injury

Cited by 34 publications

References 62 publications

Automated Core–Penumbra Quantification in Neonatal Ischemic Brain Injury

Automated Core–Penumbra Quantification in Neonatal Ischemic Brain Injury

Computational Analysis: A Bridge to Translational Stroke Treatment

Acute intranasal osteopontin treatment in male rats following TBI increases the number of activated microglia but does not alter lesion characteristics

Contact Info

Product

Resources

About