Machine learning based white matter models with permeability: An experimental study in cuprizone treated in-vivo mouse model of axonal demyelination

Hill, Ioana; Palombo, Marco; Santin, Mathieu; Branzoli, Francesca; Philippe, Anne-Charlotte; Wassermann, Demián; Aigrot, Marie‐Stéphane; Stankoff, Bruno; Evercooren, A. Baron‐Van; Felfli, Mehdi; Langui, Dominique; Zhang, Hui; Lehericy, Stéphane; Petiet, Alexandra; Alexander, Daniel C.; Ciccarelli, Olga; Drobnjak, Ivana

doi:10.1016/j.neuroimage.2020.117425

Cited by 13 publications

(19 citation statements)

References 66 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Notably, the unchanged diffusivity of the intra‐axonal metabolite tNAA in CPZ versus control mice suggested that the intra‐axonal compartments were still intact after 6 weeks of CPZ diet and under our experimental conditions, while glial reactivity was the dominant ongoing process in response to CPZ intoxication at this time point. This result was consistent with the presence of partial demyelination associated with still intact axonal diameters quantified using EM in the same mouse group and reported recently in another study 25 …”

Section: Discussionsupporting

confidence: 93%

“…After the imaging session, half of them were sacrificed for electron microscopy (EM) assessment and the other half for IHC assessment. The MRI markers were correlated with EM markers, and the results were the object of an independent article 25 . Here, we present the results obtained from DW‐MRS experiments, which were correlated with astrocyte‐ and microglia‐specific IHC markers.…”

Section: Methodsmentioning

confidence: 84%

“…This result was consistent with the presence of partial demyelination associated with still intact axonal diameters quantified using EM in the same mouse group and reported recently in another study. 25 The ADC of tCr was also not significantly different between CPZ and healthy mice. Because tCr is located in both glia and neurons, this negative finding may reflect the concomitant presence of competing mechanisms that would independently lead to higher or lower tCr ADC values.…”

Section: Mins and Tcho Diffusion Probe Ongoing Inflammation Processmentioning

confidence: 86%

“…The MRI markers were correlated with EM markers, and the results were the object of an independent article. 25 Here, we present the results obtained from DW-MRS experiments, which were correlated with astrocyte-and microglia-specific IHC markers.…”

Section: Animal Experimentsmentioning

confidence: 95%

“…were evaluated from the spectra measured at b 0 , and scaled to the water reference peak acquired in the same conditions. The percentage of cerebral spinal fluid in the VOI was estimated from the diffusion MRI data acquired in the same protocol and reported elsewhere, 25 by thresholding the mean diffusivity maps obtained from the diffusion tensor imaging analysis at values of 2 × 10 −3 mm 2 /s or more. The cerebral spinal fluid content did not vary significantly between groups (10% ± 5% in CPZ versus 13% ± 2% in control mice, p = 0.100).…”

Section: Methodsmentioning

confidence: 99%

See 4 more Smart Citations

Inflammation‐driven glial alterations in the cuprizone mouse model probed with diffusion‐weighted magnetic resonance spectroscopy at 11.7 T

et al. 2021

Self Cite

View full text Add to dashboard Cite

Inflammation of brain tissue is a complex response of the immune system to the presence of toxic compounds or to cell injury, leading to a cascade of pathological processes that include glial cell activation. Noninvasive MRI markers of glial reactivity would be very useful for in vivo detection and monitoring of inflammation processes in the brain, as well as for evaluating the efficacy of personalized treatments. Due to their specific location in glial cells, myo‐inositol (mIns) and choline compounds (tCho) seem to be the best candidates for probing glial‐specific intra‐cellular compartments. However, their concentrations quantified using conventional proton MRS are not specific for inflammation. In contrast, it has been recently suggested that mIns intra‐cellular diffusion, measured using diffusion‐weighted MRS (DW‐MRS) in a mouse model of reactive astrocytes, could be a specific marker of astrocytic hypertrophy. In order to evaluate the specificity of both mIns and tCho diffusion to inflammation‐driven glial alterations, we performed DW‐MRS in a volume of interest containing the corpus callosum and surrounding tissue of cuprizone‐fed mice after 6 weeks of intoxication, and evaluated the extent of astrocytic and microglial alterations using immunohistochemistry. Both mIns and tCho apparent diffusion coefficients were significantly elevated in cuprizone‐fed mice compared with control mice, and histologic evaluation confirmed the presence of severe inflammation. Additionally, mIns and tCho diffusion showed, respectively, strong and moderate correlations with histological measures of astrocytic and microglial area fractions, confirming DW‐MRS as a promising tool for specific detection of glial changes under pathological conditions.

show abstract

Section: Discussionsupporting

confidence: 93%

Section: Methodsmentioning

confidence: 84%

Section: Mins and Tcho Diffusion Probe Ongoing Inflammation Processmentioning

confidence: 86%

Section: Animal Experimentsmentioning

confidence: 95%

Section: Methodsmentioning

confidence: 99%

See 3 more Smart Citations

Inflammation‐driven glial alterations in the cuprizone mouse model probed with diffusion‐weighted magnetic resonance spectroscopy at 11.7 T

et al. 2021

Self Cite

View full text Add to dashboard Cite

show abstract

Training data distribution significantly impacts the estimation of tissue microstructure with machine learning

Gyori

Palombo

Clark

et al. 2021

Magnetic Resonance in Med

Self Cite

View full text Add to dashboard Cite

Purpose Supervised machine learning (ML) provides a compelling alternative to traditional model fitting for parameter mapping in quantitative MRI. The aim of this work is to demonstrate and quantify the effect of different training data distributions on the accuracy and precision of parameter estimates when supervised ML is used for fitting. Methods We fit a two‐ and three‐compartment biophysical model to diffusion measurements from in‐vivo human brain, as well as simulated diffusion data, using both traditional model fitting and supervised ML. For supervised ML, we train several artificial neural networks, as well as random forest regressors, on different distributions of ground truth parameters. We compare the accuracy and precision of parameter estimates obtained from the different estimation approaches using synthetic test data. Results When the distribution of parameter combinations in the training set matches those observed in healthy human data sets, we observe high precision, but inaccurate estimates for atypical parameter combinations. In contrast, when training data is sampled uniformly from the entire plausible parameter space, estimates tend to be more accurate for atypical parameter combinations but may have lower precision for typical parameter combinations. Conclusion This work highlights that estimation of model parameters using supervised ML depends strongly on the training‐set distribution. We show that high precision obtained using ML may mask strong bias, and visual assessment of the parameter maps is not sufficient for evaluating the quality of the estimates.

show abstract

IMPULSED model based cytological feature estimation with U‐Net: Application to human brain tumor at 3T

Kang

Lan

et al. 2022

Magnetic Resonance in Med

View full text Add to dashboard Cite

This work introduces and validates a deep-learning-based fitting method, which can rapidly provide accurate and robust estimation of cytological features of brain tumor based on the IMPULSED (imaging microstructural parameters using limited spectrally edited diffusion) model fitting with diffusion-weighted MRI data. Methods: The U-Net was applied to rapidly quantify extracellular diffusion coefficient (D ex ), cell size (d), and intracellular volume fraction (v in ) of brain tumor. At the training stage, the image-based training data, synthesized by randomizing quantifiable microstructural parameters within specific ranges, was used to train U-Net. At the test stage, the pre-trained U-Net was applied to estimate the microstructural parameters from simulated data and the in vivo data acquired on patients at 3T. The U-Net was compared with conventional non-linear least-squares (NLLS) fitting in simulations in terms of estimation accuracy and precision.Results: Our results confirm that the proposed method yields better fidelity in simulations and is more robust to noise than the NLLS fitting. For in vivo data, the U-Net yields obvious quality improvement in parameter maps, and the estimations of all parameters are in good agreement with the NLLS fitting. Moreover, our method is several orders of magnitude faster than the NLLS fitting (from about 5 min to <1 s). Conclusion:The image-based training scheme proposed herein helps to improve the quality of the estimated parameters. Our deep-learning-based fitting method can estimate the cell microstructural parameters fast and accurately.

show abstract

Machine learning based white matter models with permeability: An experimental study in cuprizone treated in-vivo mouse model of axonal demyelination

Cited by 13 publications

References 66 publications

Inflammation‐driven glial alterations in the cuprizone mouse model probed with diffusion‐weighted magnetic resonance spectroscopy at 11.7 T

Inflammation‐driven glial alterations in the cuprizone mouse model probed with diffusion‐weighted magnetic resonance spectroscopy at 11.7 T

Training data distribution significantly impacts the estimation of tissue microstructure with machine learning

IMPULSED model based cytological feature estimation with U‐Net: Application to human brain tumor at 3T

Contact Info

Product

Resources

About