Bayesian inference using hierarchical and spatial priors for intravoxel incoherent motion MR imaging in the brain: Analysis of cancer and acute stroke

IVIM-Morph: Motion-compensated quantitative Intra-voxel Incoherent Motion (IVIM) analysis for functional fetal lung maturity assessment from diffusion-weighted MRI data

Quantitative analysis of pseudo-diffusion in diffusion-weighted magnetic resonance imaging (DWI) data shows potential for assessing fetal lung maturation and generating valuable imaging biomarkers. Yet, the clinical utility of DWI data is hindered by unavoidable fetal motion during acquisition. We present IVIM-morph, a self-supervised deep neural network model for motion-corrected quantitative analysis of DWI data using the Intra-voxel Incoherent Motion (IVIM) model. IVIM-morph combines two sub-networks, a registration sub-network, and an IVIM model fitting sub-network, enabling simultaneous estimation of IVIM model parameters and motion. To promote physically plausible image registration, we introduce a biophysically informed loss function that effectively balances registration and model-fitting quality. We validated the efficacy of IVIM-morph by establishing a correlation between the predicted IVIM model parameters of the lung and gestational age (GA) using fetal DWI data of 39 subjects. Our approach was compared against six baseline methods: (1) no motion compensation, (2) affine registration of all DWI images to the initial image, (3) deformable registration of all DWI images to the initial image, (4) deformable registration of each DWI image to its preceding image in the sequence, (5) iterative deformable motion compensation combined with IVIM model parameter estimation, and (6) self-supervised deep-learning-based deformable registration. IVIM-morph exhibited a notably improved correlation with gestational age (GA) when performing in-vivo quantitative analysis of fetal lung DWI data during the canalicular phase. Specifically, over 2 test groups of cases, it achieved an Rf2 of 0.44 and 0.52, outperforming the values of 0.27 and 0.25, 0.25 and 0.00, 0.00 and 0.00, 0.38 and 0.00, and 0.07 and 0.14 obtained by other methods. IVIM-morph shows potential in developing valuable biomarkers for non-invasive assessment of fetal lung maturity with DWI data. Moreover, its adaptability opens the door to potential applications in other clinical contexts where motion compensation is essential for quantitative DWI analysis. The IVIM-morph code is readily available at: https://github.com/TechnionComputationalMRILab/qDWI-Morph.

Improved deep learning-based IVIM parameter estimation via the use of more "realistic" simulated brain data

BACKGROUND

Due to the low signal-to-noise ratio (SNR) and the limited number of b-values, precise parameter estimation of intravoxel incoherent motion (IVIM) imaging remains an open issue to date, especially for brain imaging where the relatively small difference between D and D* easily leads to outliers and obvious graininess in estimated results.

PURPOSE

To propose a synthetic data driven supervised learning method (SDD-IVIM) for improving precision and noise robustness in IVIM parameter estimation without relying on real-world data for neural network training.

METHODS

On account of the absence of standard IVIM parametric maps from real-world data, a novel model-based method for generating synthetic human brain IVIM data was introduced. Initially, the parameter values of synthetic IVIM parametric maps were sampled from the complex distributions composed of a series of simple and uniform distributions. Subsequently, these parametric maps were modulated with human brain texture to imitate brain tissue structure. Finally, they were used to generate synthetic human brain multi-b-value diffusion-weighted (DW) images based on the IVIM bi-exponential model. With the proposed data synthesis method, an ordinary U-Net with spatial smoothness was employed for IVIM parameter mapping within a supervised learning framework. The performance of SDD-IVIM was evaluated on both numerical phantom and 20 glioma patients. The estimated IVIM parametric maps were compared to those derived from five state-of-the-art methods.

RESULTS

In numerical phantom experiments, SDD-IVIM method produces IVIM parametric maps with lower mean absolute error, lower mean bias, and higher structural similarity compared to the other five methods, especially when the SNR of DW images is low. In glioma patient experiments, SDD-IVIM method offers lower coefficient of variation and more reasonable contrast-to-noise ratio between tumor and contralateral normal appearing white matter than the other five methods.

CONCLUSION

Our method owns superior performance in parametric map quality, parameter estimation precision, and lesion characterization in IVIM parameter estimation, with strong resistance to noise.

IVIM parameters mapping with artificial neural network based on mean deviation prior

BACKGROUND

The diffusion and perfusion parameters derived from intravoxel incoherent motion (IVIM) imaging provide promising biomarkers for noninvasively quantifying and managing various diseases. Nevertheless, due to the distribution gap between simulated and real datasets, the out-of-distribution (OOD) problem occurred in supervised learning-based methods degrades their performance and hinders their real applications.

PURPOSE

To address the OOD problem in supervised methods and to further improve the accuracy and stability of IVIM parameter estimation, this work proposes a novel learning framework called IterANN, based on mean deviation prior (MDP) between training and estimated IVIM parameters on the test set.

METHODS

Specifically, MDP indicates that the mean of the estimated IVIM parameters always locates between the mean of IVIM parameters in the test and train sets. In IterANN, we adopt a very simple artificial neural network (ANN) architecture of two hidden layers with 12 neurons per hidden layer, an input layer containing the signals acquired at multiple b-values and an output layer composed of three IVIM parameters ( D $D$ , F $F$ andD S t a r $DStar$ ). Inspired by MDP, the distribution of IVIM parameters in the training set (simulated data) is iteratively updated so that their mean gradually approaches the predicted values of the real data. This aims to achieve a strong correlation between the simulated data and the real data. To validate the effectiveness of IterANN, we compare it with several methods on both simulation and real acquisition datasets, including 21 healthy and 3 tumor subjects, in terms of residual errors of IVIM parameters or DW signals, the coefficients of variation (CV) of IVIM parameters, and the parameter contrast-to-noise ratio (PCNR) between normal and tumor tissues.

RESULTS

On two simulation datasets, the proposed IterANN achieves the lowest residual error in IVIM parameters, especially in the case of low signal-to-noise ratio (SNR = 10), the residual error of D $D$ , F $F$ andD S t a r $DStar$ is decreased by15.82 % / 14.92 % , 81.19 % / 74.04 % , 50.77 % / 1.549 % $15.82\%/14.92\%, 81.19\%/74.04\%, 50.77\%/1.549\%$ (Gaussian distribution /realistic distribution) respectively comparing to the suboptimal method. On real dataset, the IterANN achieves the highest PCNR when comparing the normal and tumor regions. Additionally, the proposed IterANN demonstrated better stability, with its CV being significantly lower than that of other methods in the vast majority of cases (p < 0.01 $p<0.01$ , paired-sample Student's t-test).

CONCLUSIONS

The superior performance of IterANN demonstrates that updating the distribution of the train set based on MDP can effectively solve the OOD problem, which allows us not only to improve the accuracy and stability of the estimated IVIM parameters, but also to increase the potential of IVIM in disease diagnosis.

Intravoxel incoherent motion imaging in stroke infarct core and penumbra is related to long-term clinical outcome

Intravoxel incoherent motion (IVIM) imaging, a contrast agent-free magnetic resonance imaging technique, enables the evaluation of microvascular perfusion abnormalities in acute stroke. Prior research reported reduced IVIM values within the infarct core in acute stroke. However, findings concerning IVIM characteristics in the penumbra have been mixed and the relationship between IVIM and clinical outcomes remains unknown. We employed a longitudinal multimodal imaging approach for ischemic stroke patients (n analyzed=24; pre-/post-treatment and 90-day post-stroke assessments) including IVIM, diffusion-weighted, and contrast-enhanced perfusion-weighted imaging. We evaluated IVIM in relevant stroke areas after endovascular treatment. Reduced post-treatment IVIM perfusion fraction in infarct core and recanalized penumbra was associated with poorer functional recovery at 90-days post-stroke (NIH Stroke Scale [NIHSS]; r=-0.64 and r=-0.69). Including IVIM perfusion fraction increased the explained variance of NIHSS from 42% up to 83% compared to well-known prognostic factors core volume and patient age. Additionally, IVIM perfusion fraction was reduced in the core and recanalized penumbra compared to contralateral healthy tissue, suggesting impaired microvascular reperfusion after endovascular treatment. In conclusion, IVIM characteristics of the infarct core and recanalized penumbra are strong prognostic factors for long-term outcome in stroke patients and IVIM shows promise for characterizing microvascular perfusion in relevant stroke areas.

Incorporating spatial information in deep learning parameter estimation with application to the intravoxel incoherent motion model in diffusion-weighted MRI

In medical image analysis, the utilization of biophysical models for signal analysis offers valuable insights into the underlying tissue types and microstructural processes. In diffusion-weighted magnetic resonance imaging (DWI), a major challenge lies in accurately estimating model parameters from the acquired data due to the inherently low signal-to-noise ratio (SNR) of the signal measurements and the complexity of solving the ill-posed inverse problem. Conventional model fitting approaches treat individual voxels as independent. However, the tissue microenvironment is typically homogeneous in a local environment, where neighboring voxels may contain correlated information. To harness the potential benefits of exploiting correlations among signals in adjacent voxels, this study introduces a novel approach to deep learning parameter estimation that effectively incorporates relevant spatial information. This is achieved by training neural networks on patches of synthetic data encompassing plausible combinations of direct correlations between neighboring voxels. We evaluated the approach on the intravoxel incoherent motion (IVIM) model in DWI. We explored the potential of several deep learning architectures to incorporate spatial information using self-supervised and supervised learning. We assessed performance quantitatively using novel fractal-noise-based synthetic data, which provide ground truths possessing spatial correlations. Additionally, we present results of the approach applied to in vivo DWI data consisting of twelve repetitions from a healthy volunteer. We demonstrate that supervised training on larger patch sizes using attention models leads to substantial performance improvements over both conventional voxelwise model fitting and convolution-based approaches.

Comprehensive assessment of postoperative recurrence and survival in patients with cervical cancer

BACKGROUND

The prediction of postoperative recurrence and survival in cervical cancer patients has been a major clinical challenge. The combination of clinical parameters, inflammatory markers, intravoxel incoherent motion diffusion-weighted imaging (IVIM-DWI), and MRI-derived radiomics is expected to support the prediction of recurrence-free survival (RFS), disease-free survival (DFS), tumor-specific survival (CSS), and overall survival (OS) of cervical cancer patients after surgery.

METHODS

A retrospective analysis of 181 cervical cancer patients with continuous follow-up was completed. The parameters of IVIM-DWI and radiomics were measured, analyzed, and screened. The LASSO regularization was used to calculate the radiomics score (Rad-score). Multivariate Cox regression analysis was used to construct nomogram models for predicting postoperative RFS, DFS, CSS, and OS in cervical cancer patients, with internal and external validation.

RESULTS

Clinical stage, parametrial infiltration, internal irradiation, D-value, and Rad-score were independent prognostic factors for RFS; Squamous cell carcinoma antigen, internal irradiation, D-value, f-value and Rad-score were independent prognostic factors for DFS; Maximum tumor diameter, lymph node metastasis, platelets, D-value and Rad-score were independent prognostic factors for CSS; Lymph node metastasis, systemic inflammation response index, D-value and Rad-score were independent prognostic factors for OS. The AUCs of each model predicting RFS, DFS, CSS, and OS at 1, 3, and 5 years were 0.985, 0.929, 0.910 and 0.833, 0.818, 0.816 and 0.832, 0.863, 0.891 and 0.804, 0.812, 0.870, respectively.

CONCLUSIONS

Nomograms based on clinical and imaging parameters showed high clinical value in predicting postoperative RFS, DFS, CSS, and OS of cervical cancer patients and can be used as prognostic markers.

Accelerated multi-b-value multi-shot diffusion-weighted imaging based on EPI with keyhole and a low-rank tensor constraint

PURPOSE

Multi-Shot (MS) Echo-Planar Imaging (EPI) may improve the in-plane resolution of multi-b-value DWI, yet it also considerably increases the scan time. Here we explored the combination of EPI with Keyhole (EPIK) and a calibrationless reconstruction algorithm for acceleration of multi-b-value MS-DWI.

METHODS

We firstly analyzed the impact of nonuniform phase accrual in EPIK on the reconstructed image. Based on insights gained from the analysis, we developed a calibrationless reconstruction algorithm based on a Space-Contrast-Coil Locally Low-Rank Tensor (SCC-LLRT) constraint for reconstruction of EPIK-acquired data. We compared the algorithm with a modified SPatial-Angular Locally Low-Rank (SPA-LLR) algorithm through simulations, phantoms, and in vivo study. We then compared EPIK with uniformly undersampled EPI for accelerating multi-b-value DWI in 6 healthy subjects.

RESULTS

Through theoretical derivations, we found that the reconstruction of EPIK with a SENSE-encoding-based algorithm, such as SPA-LLR, may cause additional aliasing artifacts due to the frequency-dependent distortion of the coil sensitivity. Results from simulations, phantoms, and in vivo study verified the theoretical finding by showing that the calibrationless SCC-LLRT algorithm reduced aliasing artifacts compared with SPA-LLR. Finally, EPIK with SCC-LLRT substantially reduced the ghosting artifacts compared with uniform undersampled multi-b-value DWI, decreasing the fitting errors in ADC (0.05 ± 0.01 vs 0.10 ± 0.01, P < 0.001) and IVIM mapping (0.026 ± 0.004 vs 0.06 ± 0.006, P < 0.001).

CONCLUSION

The SCC-LLRT algorithm reduced the aliasing artifacts of EPIK by using a calibrationless modeling of the multi-coil data. The dense sampling of k-space center offers EPIK a potential to improve image quality for acceleration of multi-b-value MS-DWI.

Comparison of methods for intravoxel incoherent motion parameter estimation in the brain from flow-compensated and non-flow-compensated diffusion-encoded data

PURPOSE

Joint analysis of flow-compensated (FC) and non-flow-compensated (NC) diffusion MRI (dMRI) data has been suggested for increased robustness of intravoxel incoherent motion (IVIM) parameter estimation. For this purpose, a set of methods commonly used or previously found useful for IVIM analysis of dMRI data obtained with conventional diffusion encoding were evaluated in healthy human brain.

METHODS

Five methods for joint IVIM analysis of FC and NC dMRI data were compared: (1) direct non-linear least squares fitting, (2) a segmented fitting algorithm with estimation of the diffusion coefficient from higher b-values of NC data, (3) a Bayesian algorithm with uniform prior distributions, (4) a Bayesian algorithm with spatial prior distributions, and (5) a deep learning-based algorithm. Methods were evaluated on brain dMRI data from healthy subjects and simulated data at multiple noise levels. Bipolar diffusion encoding gradients were used with b-values 0-200 s/mm2 and corresponding flow weighting factors 0-2.35 s/mm for NC data and by design 0 for FC data. Data were acquired twice for repeatability analysis.

RESULTS

Measurement repeatability as well as estimation bias and variability were at similar levels or better with the Bayesian algorithm with spatial prior distributions and the deep learning-based algorithm for IVIM parametersD $$ D $$ andf $$ f $$ , and for the Bayesian algorithm only forv d $$ {v}_d $$ , relative to the other methods.

CONCLUSION

A Bayesian algorithm with spatial prior distributions is preferable for joint IVIM analysis of FC and NC dMRI data in the healthy human brain, but deep learning-based algorithms appear promising.

Intravoxel incoherent motion magnetic resonance imaging to assess early tumor response to radiation therapy: Review and future directions

Early prediction of radiation response by imaging is a dynamic field of research and it can be obtained using a variety of noninvasive magnetic resonance imaging methods. Recently, intravoxel incoherent motion (IVIM) has gained interest in cancer imaging. IVIM carries both diffusion and perfusion information, making it a promising tool to assess tumor response. Here, we briefly introduced the basics of IVIM, reviewed existing studies of IVIM in various type of tumors during radiotherapy in order to show whether IVIM is a useful technique for an early assessment of radiation response. 31/40 studies reported an increase of IVIM parameters during radiotherapy compared to baseline. In 27 studies, this increase was higher in patients with good response to radiotherapy. Future directions including implementation of IVIM on MR-Linac and its limitation are discussed. Obtaining new radiologic biomarkers of radiotherapy response could open the way for a more personalized, biology-guided radiation therapy.

MetaCAE: Causal autoencoder with meta-knowledge transfer for brain effective connectivity estimation

Using machine learning methods to estimate brain effective connectivity networks from functional magnetic resonance imaging (fMRI) data has gradually become one of the hot subjects in the fields of neuroscience. In particular, the encoder-decoder based methods can effectively extract the connections in fMRI time series, which have achieved promising performance. However, these methods generally use Granger causality model, which may identify false directions due to the non-stationary characteristic of fMRI data. Additionally, fMRI datasets have limited sample sizes, which significantly constrains the development of these methods. In this paper, we propose a novel brain effective connectivity estimation method based on causal autoencoder with meta-knowledge transfer, called MetaCAE. The proposed approach employs a causal autoencoder (CAE) to extract causal dependencies from non-stationary fMRI time series, and leverages meta-knowledge transfer to improve the estimation accuracy on small-sample data. More specifically, MetaCAE first employs a temporal convolutional encoder to extract non-stationary temporal information from fMRI time series. Then it uses a structural equation model-based decoder to decode causal relationships between brain regions. Finally, it utilizes a model-agnostic meta-learning method to learn the meta-knowledge of the shared brain effective connectivity among different subjects, and transfers the meta-knowledge to the CAE to enhance its estimation ability on small-sample fMRI data. Comprehensive experiments on both simulated and real-world data demonstrate the efficacy of MetaCAE in estimating brain effective connectivity.

Clinical Interpretation of Intravoxel Incoherent Motion Perfusion Imaging in the Brain

Intravoxel incoherent motion (IVIM) perfusion imaging extracts information on blood motion in biological tissue from diffusion-weighted MR images. The method is attractive from a clinical stand point, because it measures in essence local quantitative perfusion, without intravenous contrast injection. Currently, the clinical interpretation of IVIM perfusion maps focuses on the IVIM perfusion fraction maps, but improvements in image quality of the IVIM pseudo-diffusion maps, using advanced postprocessing tools involving artificial intelligence, could lead to an increased interest in this parameters, as it could provide additional local perfusion information in the clinical setting, not otherwise available with other perfusion techniques.

Investigation of perfusion impairment in degenerative cervical myelopathy beyond the site of cord compression

The aim of this study was to determine tissue-specific blood perfusion impairment of the cervical cord above the compression site in patients with degenerative cervical myelopathy (DCM) using intravoxel incoherent motion (IVIM) imaging. A quantitative MRI protocol, including structural and IVIM imaging, was conducted in healthy controls and patients. In patients, T2-weighted scans were acquired to quantify intramedullary signal changes, the maximal canal compromise, and the maximal cord compression. T2*-weighted MRI and IVIM were applied in all participants in the cervical cord (covering C1-C3 levels) to determine white matter (WM) and grey matter (GM) cross-sectional areas (as a marker of atrophy), and tissue-specific perfusion indices, respectively. IVIM imaging resulted in microvascular volume fraction ([Formula: see text]), blood velocity ([Formula: see text]), and blood flow ([Formula: see text]) indices. DCM patients additionally underwent a standard neurological clinical assessment. Regression analysis assessed associations between perfusion parameters, clinical outcome measures, and remote spinal cord atrophy. Twenty-nine DCM patients and 30 healthy controls were enrolled in the study. At the level of stenosis, 11 patients showed focal radiological evidence of cervical myelopathy. Above the stenosis level, cord atrophy was observed in the WM (- 9.3%; p = 0.005) and GM (- 6.3%; p = 0.008) in patients compared to healthy controls. Blood velocity (BV) and blood flow (BF) indices were decreased in the ventral horns of the GM (BV: - 20.1%, p = 0.0009; BF: - 28.2%, p = 0.0008), in the ventral funiculi (BV: - 18.2%, p = 0.01; BF: - 21.5%, p = 0.04) and lateral funiculi (BV: - 8.5%, p = 0.03; BF: - 16.5%, p = 0.03) of the WM, across C1-C3 levels. A decrease in microvascular volume fraction was associated with GM atrophy (R = 0.46, p = 0.02). This study demonstrates tissue-specific cervical perfusion impairment rostral to the compression site in DCM patients. IVIM indices are sensitive to remote perfusion changes in the cervical cord in DCM and may serve as neuroimaging biomarkers of hemodynamic impairment in future studies. The association between perfusion impairment and cervical cord atrophy indicates that changes in hemodynamics caused by compression may contribute to the neurodegenerative processes in DCM.

Deep learning intravoxel incoherent motion modeling: Exploring the impact of training features and learning strategies

PURPOSE

The development of advanced estimators for intravoxel incoherent motion (IVIM) modeling is often motivated by a desire to produce smoother parameter maps than least squares (LSQ). Deep neural networks show promise to this end, yet performance may be conditional on a myriad of choices regarding the learning strategy. In this work, we have explored potential impacts of key training features in unsupervised and supervised learning for IVIM model fitting.

METHODS

Two synthetic data sets and one in-vivo data set from glioma patients were used in training of unsupervised and supervised networks for assessing generalizability. Network stability for different learning rates and network sizes was assessed in terms of loss convergence. Accuracy, precision, and bias were assessed by comparing estimations against ground truth after using different training data (synthetic and in vivo).

RESULTS

A high learning rate, small network size, and early stopping resulted in sub-optimal solutions and correlations in fitted IVIM parameters. Extending training beyond early stopping resolved these correlations and reduced parameter error. However, extensive training resulted in increased noise sensitivity, where unsupervised estimates displayed variability similar to LSQ. In contrast, supervised estimates demonstrated improved precision but were strongly biased toward the mean of the training distribution, resulting in relatively smooth, yet possibly deceptive parameter maps. Extensive training also reduced the impact of individual hyperparameters.

CONCLUSION

Voxel-wise deep learning for IVIM fitting demands sufficiently extensive training to minimize parameter correlation and bias for unsupervised learning, or demands a close correspondence between the training and test sets for supervised learning.