Cerebral oxygen metabolism during and after therapeutic hypothermia in neonatal hypoxic–ischemic encephalopathy: a feasibility study using magnetic resonance imaging

Shetty, Anil N.; Lucke, Ashley M.; Liu, Peiying; Cortés, Magdalena Sanz; Hagan, Joseph; Chu, Zili; Hunter, Jill V.; Lu, Hanzhang; Lee, Wesley; Kaiser, Jeffrey R.

doi:10.1007/s00247-018-4283-9

Cited by 24 publications

(45 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The learning rate is another important hyperparameter we chose 0.0001. This architecture has 4 convolutional blocks, comprising a 2D convolutional layer with a 2D kernel (kernel size: 3×3), a 2D batch-normalization layer, a dropout rate (0.2), a rectified linear unit (ReLU) as an activation function and a max-pooling layer pool size= (2,2). The number of feature channels for each block is 32, 64, 128, 256 respectively.…”

Section: A 2d Cnn Modelmentioning

confidence: 99%

“…Figure 9 shows a 3D-CNN architecture used in the proposed neurodevelopmental age estimation methodology. As similar to 2D CNN, this architecture has 4 convolutional blocks consisting of a 3D convolutional layer (kernel size: 3×3×3), a 3D batch-normalization layer, dropout rate (0.2), ReLU activation layer and a max-pooling layer (pool_size=(1, 2, 2), strides= (1,2,2)). The number of feature channels is 32, 64, 128, and 256 for each block, respectively.…”

Section: A 2d Cnn Modelmentioning

confidence: 99%

See 1 more Smart Citation

Classification of Neurodevelopmental Age in Normal Infants Using 3D-CNN based on Brain MRI

Shabanian

Eckstein

Chen

et al. 2019

2019 IEEE International Conference on Bioinformatics and Biomedicine (BIBM)

View full text Add to dashboard Cite

Human brain development is rapid during infancy and early childhood. Many disease processes impair this development. Therefore, brain developmental age estimation (BDAE) is essential for all diseases affecting cognitive development. Brain magnetic resonance imaging (MRI) of infants shows brain growth and morphologic patterns during childhood. Therefore, we can estimate the developmental age from brain images. However, MRI analysis is time consuming because each scan contains millions of data points (voxels). We investigated the three-dimensional convolutional neural network (3D CNN), a deep learning algorithm, to rapidly classify neurodevelopmental age with high accuracy based on MRIs. MRIs from normal newborns were obtained from the National Institute of Mental Health (NIMH) Data Archive. Age categories of pediatric MRIs were 3 wks ± 1 wk, 1 yr ± 2 wks, and 3 yrs ± 4 wks. We trained a BDAE method using T1, T2, and proton density (PD) images from MRI scans of 112 individuals using 3D CNN. Compared with the known age, our method has a sensitivity of 99% and specificity of 98.3%. Moreover, our 3D CNN model has better performance in neurodevelopmental age estimation than does 2D CNN.

show abstract

Section: A 2d Cnn Modelmentioning

confidence: 99%

Section: A 2d Cnn Modelmentioning

confidence: 99%

Classification of Neurodevelopmental Age in Normal Infants Using 3D-CNN based on Brain MRI

Shabanian

Eckstein

Chen

et al. 2019

2019 IEEE International Conference on Bioinformatics and Biomedicine (BIBM)

View full text Add to dashboard Cite

show abstract

“…We examined two sequences relevant to the brain's oxygen homeostasis, T 2 ‐relaxation‐under‐spin‐tagging (TRUST) MRI for the assessment of OEF and phase‐contrast (PC) MRI for CBF measurement. The study focused on these two methods because they are among the most commonly used sequences for oxygenation and blood flow measurements, respectively, and both have shown some promises in neonatal applications . For both sequences, we characterized the dependence of the results on imaging resolution, given the considerably smaller size of the neonatal brain.…”

Section: Introductionmentioning

confidence: 99%

“…It is pragmatically challenging to conduct technical studies in neonates. To date, most neonatal studies have been performed on neonatal patients who had clinical indications for MRI, 7,8,10,11,15,16 and the research sequences were conducted as add-ons to clinical sequences. This is because it is difficult to recruit normal neonates who would otherwise not undergo MRI to participate in MRI research studies.…”

Section: Introductionmentioning

confidence: 99%

“…The study focused on these two methods because they are among the most commonly used sequences for oxygenation and blood flow measurements, respectively, and both have shown some promises in neonatal applications. 8,11,15 For both sequences, we characterized the dependence of the results on imaging resolution, given the considerably smaller size of the neonatal brain. Additionally, we investigated sequence-specific characteristics, ie consistency in oxygenation between venous vessels using TRUST MRI and consistency in blood flow between gated and non-gated PC MRI.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Characterization of MRI techniques to assess neonatal brain oxygenation and blood flow

et al. 2019

Self Cite

View full text Add to dashboard Cite

There is increasing interest in applying physiological MRI in neonates, based on the premise that physiological parameters may provide an early biomarker of neonatal brain health and injury. Two commonly used techniques are oxygen extraction fraction (OEF) measurement using T2‐relaxation‐under‐spin‐tagging (TRUST) MRI and cerebral blood flow measurement using phase‐contrast (PC) quantitative flow MRI, which collectively provide an assessment of the brain's oxygen consumption. However, prior research has only demonstrated proof of principle of these methods in neonates, without characterization or benchmarking of the techniques. This is because available time is limited in neonatal subjects, especially when scans are performed as add‐ons to clinical scans (typically less than 5 min). The work presented aims to examine the TRUST and PC MRI sequences systematically in normal neonates, through research‐dedicated scan sessions. A series of characterization and optimization studies were conducted in a total of 26 radiographically normal neonates on 3 T systems. Our results show that TRUST MRI at the superior sagittal sinus (SSS) provides an OEF measurement equivalent to that at the internal jugular vein (r = 0.80, n = 10), yet with shorter scan time. Lower resolution provided better precision in the TRUST measurement (p = 0.001, n = 9). Therefore, the preferred OEF measurement is to apply TRUST MRI at the SSS using a spatial resolution of 2.5 mm. For PC MRI, our results showed that non‐gated PC MRI yielded blood flow measurements comparable to those from the more time‐consuming gated approach in neonates (r = 0.89, n = 7). It was also found that blood flow could be overestimated by 18% when imaging resolution is larger than 0.3 mm (n = 7). Therefore, non‐gated PC MRI with a spatial resolution of 0.3 mm is recommended for neonatal applications. In conclusion, this study verifies consistency of neonatal brain oxygenation and flow measurements across acquisition schemes and points to optimal strategies in parameter selection when using these sequences.

show abstract

Fractional anisotropy from diffusion tensor imaging correlates with acute astrocyte and myelin swelling in neonatal swine models of excitotoxic and hypoxic‐ischemic brain injury

et al. 2021

Self Cite

View full text Add to dashboard Cite

The specific cytopathology that causes abnormal fractional anisotropy (FA) and mean diffusivity (MD) from diffusion tensor imaging (DTI) after neonatal hypoxia‐ischemia (HI) is not completely understood. The panoply of cell types in the brain might contribute differentially to changes in DTI metrics. Because glia are the predominant cell type in brain, we hypothesized that changes in FA and MD would signify perturbations in glial microstructure. Using a 3‐Tesla clinical scanner, we conducted in vivo DTI MRI in nine neonatal piglets at 20–96 h after excitotoxic brain injury from striatal quinolinic acid injection or global HI. FA and MD from putamen, caudate, and internal capsule in toto were correlated with astrocyte swelling, neuronal excitotoxicity, and white matter injury. Low FA correlated with more swollen astrocytes immunophenotyped by aquaporin‐4 (AQP4), glial fibrillary acidic protein (GFAP), and glutamate transporter‐1 (GLT‐1). Low FA was also related to the loss of neurons with perineuronal GLT‐1+ astrocyte decorations, large myelin swellings, lower myelin density, and oligodendrocyte cell death identified by 2′,3′‐cyclic nucleotide 3′‐phosphodiesterase, bridging integrator‐1, and nuclear morphology. MD correlated with degenerating oligodendrocytes and depletion of normal GFAP+ astrocytes but not with astrocyte or myelin swelling. We conclude that FA is associated with cytotoxic edema in astrocytes and oligodendrocyte processes as well as myelin injury at the cellular level. MD can detect glial cell death and loss, but it may not discern subtle pathology in swollen astrocytes, oligodendrocytes, or myelin. This study provides a cytopathologic basis for interpreting DTI in the neonatal brain after HI.

show abstract

Cerebral oxygen metabolism during and after therapeutic hypothermia in neonatal hypoxic–ischemic encephalopathy: a feasibility study using magnetic resonance imaging

Cited by 24 publications

References 23 publications

Classification of Neurodevelopmental Age in Normal Infants Using 3D-CNN based on Brain MRI

Classification of Neurodevelopmental Age in Normal Infants Using 3D-CNN based on Brain MRI

Characterization of MRI techniques to assess neonatal brain oxygenation and blood flow

Fractional anisotropy from diffusion tensor imaging correlates with acute astrocyte and myelin swelling in neonatal swine models of excitotoxic and hypoxic‐ischemic brain injury

Contact Info

Product

Resources

About