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.

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.

Towards detection of early response in neoadjuvant chemotherapy of breast cancer using Bayesian intravoxel incoherent motion

Introduction

The early identification of good responders to neoadjuvant chemotherapy (NACT) holds a significant potential in the optimal treatment of breast cancer. A recent Bayesian approach has been postulated to improve the accuracy of the intravoxel incoherent motion (IVIM) model for clinical translation. This study examined the prediction and early sensitivity of Bayesian IVIM to NACT response.

Materials and methods

Seventeen female patients with breast cancer were scanned at baseline and 16 patients were scanned after Cycle 1. Tissue diffusion and perfusion from Bayesian IVIM were calculated at baseline with percentage change at Cycle 1 computed with reference to baseline. Cellular proliferative activity marker Ki-67 was obtained semi-quantitatively with percentage change at excision computed with reference to core biopsy.

Results

The perfusion fraction showed a significant difference (p = 0.042) in percentage change between responder groups at Cycle 1, with a decrease in good responders [-7.98% (-19.47-1.73), n = 7] and an increase in poor responders [10.04% (5.09-28.93), n = 9]. There was a significant correlation between percentage change in perfusion fraction and percentage change in Ki-67 (p = 0.042). Tissue diffusion and pseudodiffusion showed no significant difference in percentage change between groups at Cycle 1, nor was there a significant correlation against percentage change in Ki-67. Perfusion fraction, tissue diffusion, and pseudodiffusion showed no significant difference between groups at baseline, nor was there a significant correlation against Ki-67 from core biopsy.

Conclusion

The alteration in tumour perfusion fraction from the Bayesian IVIM model, in association with cellular proliferation, showed early sensitivity to good responders in NACT.

Clinical trial registration

https://clinicaltrials.gov/ct2/show/NCT03501394, identifier NCT03501394.

Estimation of intravoxel incoherent motion (IVIM) parameters in vertebral bone marrow: a comparative study of five algorithms

OBJECTIVES

To access the performances of different algorithms for quantification of Intravoxel incoherent motion (IVIM) parameters D, f, [Formula: see text] in Vertebral Bone Marrow (VBM).

MATERIALS AND METHODS

Five algorithms were studied: four deterministic algorithms (the One-Step and three segmented methods: Two-Step, Three-Step, and Fixed-[Formula: see text] algorithm) based on the least-squares (LSQ) method and a Bayesian probabilistic algorithm. Numerical simulations and quantification of IVIM parameters D, f, [Formula: see text] in vivo in vertebral bone marrow, were done on six healthy volunteers. The One-way repeated-measures analysis of variance (ANOVA) followed by Bonferroni's multiple comparison test (p value = 0.05) was applied.

RESULTS

In numerical simulations, the Bayesian algorithm provided the best estimation of D, f, [Formula: see text] compared to the deterministic algorithms. In vivo VBM-IVIM, the values of D and f estimated by the Bayesian algorithm were close to those of the One-Step method, in contrast to the three segmented methods.

DISCUSSION

The comparison of the five algorithms indicates that the Bayesian algorithm provides the best estimation of VBM-IVIM parameters, in both numerical simulations and in vivo data.

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.

Diffusion Breast MRI: Current Standard and Emerging Techniques

The role of diffusion weighted imaging (DWI) as a biomarker has been the subject of active investigation in the field of breast radiology. By quantifying the random motion of water within a voxel of tissue, DWI provides indirect metrics that reveal cellularity and architectural features. Studies show that data obtained from DWI may provide information related to the characterization, prognosis, and treatment response of breast cancer. The incorporation of DWI in breast imaging demonstrates its potential to serve as a non-invasive tool to help guide diagnosis and treatment. In this review, current technical literature of diffusion-weighted breast imaging will be discussed, in addition to clinical applications, advanced techniques, and emerging use in the field of radiomics.

Effects of echo time on IVIM quantifications of locally advanced breast cancer in clinical diffusion-weighted MRI at 3 T

PURPOSE

The purpose of this study was to investigate the effects of echo time dependence in IVIM quantification of the pseudo-diffusion fraction in breast cancer and whether correcting for the echo time dependence offers added clinical value.

MATERIALS AND METHODS

Fifteen patients with biopsy-proven breast cancer underwent a 3 T MRI examination with an extended DWI protocol at two different echo times (TE = 53 ms, b = 0, 50 s/mm² ; TE = 77 ms, b = 0, 50, 120, 200, 400, 700 s/mm² ). Volumes of interest were delineated around the tumors. In addition, simulated MRI data were generated for different levels of signal-to-noise ratio and two values for the blood T2 relaxation time (T2_p = 100 ms and 150 ms). The pseudo-diffusion signal fraction was estimated from the simulated and in vivo tumor data using both the standard IVIM model and an extended IVIM model that accounts for the echo time dependence arising from distinct transverse relaxation times.

RESULTS

Simulations showed that the standard IVIM model overestimated the pseudo-diffusion fraction by 25% (T2_p = 100 ms) and 60 % (T2_p = 150 ms) (p < 0.0001 at SNR = 50). In vivo, the estimated apparent T2 value at b = 50 s/mm² was around 8% lower than at b = 0 s/mm² (p = 0.01) demonstrating a removal of the signal contribution from blood with long T2 associated with pseudo-diffusion. Using two different fixed values for T2_p = 100, 150 ms, the pseudo-diffusion fraction was 15% and 46% higher in the standard model compared with the echo-time-corrected model (p < 0.01).

CONCLUSION

The standard IVIM model was found to overestimate the pseudo-diffusion fraction by 15% to 46% compared with the echo-time-corrected model in breast tumor DWI data acquired at 3 T. Our results suggest that a corrected model may give more accurate results in terms of signal fractions, but may not justify the added time needed to acquire the additional data in terms of clinical value.

Intravoxel incoherent motion diffusion-weighted imaging in the characterization of Alzheimer's disease

OBJECTIVES

Alzheimer's disease (AD) is the most common type of dementia, and characterizing brain changes in AD is important for clinical diagnosis and prognosis. This study was designed to evaluate the classification performance of intravoxel incoherent motion (IVIM) diffusion-weighted imaging in differentiating between AD patients and normal control (NC) subjects and to explore its potential effectiveness as a neuroimaging biomarker.

METHODS

Thirty-one patients with probable AD and twenty NC subjects were included in the prospective study. IVIM data were subjected to postprocessing, and parameters including the apparent diffusion coefficient (ADC), slow diffusion coefficient (D_s), fast diffusion coefficient (D_f), perfusion fraction (fp) and D_f*fp were calculated. The classification model was developed and confirmed with cross-validation (group A/B) using Support Vector Machine (SVM). Correlations between IVIM parameters and Mini-Mental State Examination (MMSE) scores in AD patients were investigated using partial correlation analysis.

RESULTS

Diffusion MRI revealed significant region-specific differences that aided in differentiating AD patients from controls. Among the analyzed regions and parameters, the D_f of the right precuneus (PreR) (ρ = 0.515; P = 0.006) and the left cerebellum (CL) (ρ = 0.429; P = 0.026) demonstrated significant associations with the cognitive function of AD patients. An area under the receiver operating characteristics curve (AUC) of 0.84 (95% CI: 0.66, 0.99) was calculated for the validation in dataset B after the prediction model was trained on dataset A. When the datasets were reversed, an AUC of 0.90 (95% CI: 0.75, 1.00) was calculated for the validation in dataset A, after the prediction model trained in dataset B.

CONCLUSION

IVIM imaging is a promising method for the classification of AD and NC subjects, and IVIM parameters of precuneus and cerebellum might be effective biomarker for the diagnosis of AD.

Understanding diffusion-weighted MRI analysis: Repeatability and performance of diffusion models in a benign breast lesion cohort

Diffusion-weighted MRI (DWI) is an important tool for oncology research, with great clinical potential for the classification and monitoring of breast lesions. The utility of parameters derived from DWI, however, is influenced by specific analysis choices. The purpose of this study was to critically evaluate repeatability and curve-fitting performance of common DWI signal representations, for a prospective cohort of patients with benign breast lesions. Twenty informed, consented patients with confirmed benign breast lesions underwent repeated DWI (3 T) using: sagittal single-shot spin-echo echo planar imaging, bipolar encoding, TR/TE: 11,600/86 ms, FOV: 180 x 180 mm, matrix: 90 x 90, slices: 60 x 2.5 mm, iPAT: GRAPPA 2, fat suppression, and 13 b-values: 0-700 s/mm² . A phase-reversed scan (b = 0 s/mm² ) was acquired for distortion correction. Voxel-wise repeat-measures coefficients of variation (CoVs) were derived for monoexponential (apparent diffusion coefficient [ADC]), biexponential (intravoxel incoherent motion: f, D, D*) and stretched exponential (α, DDC) across the parameter histograms for lesion regions of interest (ROIs). Goodness-of-fit for each representation was assessed by Bayesian information criterion. The volume of interest (VOI) definition was repeatable (CoV 13.9%). Within lesions, and across both visits and the cohort, there was no dominant best-fit model, with all representations giving the best fit for a fraction of the voxels. Diffusivity measures from the signal representations (ADC, D, DDC) all showed good repeatability (CoV < 10%), whereas parameters associated with pseudodiffusion (f, D*) performed poorly (CoV > 50%). The stretching exponent α was repeatable (CoV < 12%). This pattern of repeatability was consistent over the central part of the parameter percentiles. Assumptions often made in diffusion studies about analysis choices will influence the detectability of changes, potentially obscuring useful information. No single signal representation prevails within or across lesions, or across repeated visits; parameter robustness is therefore a critical consideration. Our results suggest that stretched exponential representation is more repeatable than biexponential, with pseudodiffusion parameters unlikely to provide clinically useful biomarkers.

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

The intravoxel incoherent motion (IVIM) model allows to map diffusion (D) and perfusion-related parameters (F and D*). Parameter estimation is, however, error-prone due to the non-linearity of the signal model, the limited signal-to-noise ratio (SNR) and the small volume fraction of perfusion in the in-vivo brain. In the present work, the performance of Bayesian inference was examined in the presence of brain pathologies characterized by hypo- and hyperperfusion. In particular, a hierarchical and a spatial prior were combined. Performance was compared relative to conventional segmented least squares regression, hierarchical prior only (non-segmented and segmented data likelihoods) and a deep learning approach. Realistic numerical brain IVIM simulations were conducted to assess errors relative to ground truth. In-vivo, data of 11 central nervous system cancer patients and 9 patients with acute stroke were acquired. The proposed method yielded reduced error in simulations for both the cancer and acute stroke scenarios compared to other methods across the whole investigated SNR range. The contrast-to-noise ratio of the proposed method was better or on par compared to the other techniques in-vivo. The proposed Bayesian approach hence improves IVIM parameter estimation in brain cancer and acute stroke.

Discrimination of Breast Cancer from Healthy Breast Tissue Using a Three-component Diffusion-weighted MRI Model

PURPOSE

Diffusion-weighted MRI (DW-MRI) is a contrast-free modality that has demonstrated ability to discriminate between predefined benign and malignant breast lesions. However, how well DW-MRI discriminates cancer from all other breast tissue voxels in a clinical setting is unknown. Here we explore the voxelwise ability to distinguish cancer from healthy breast tissue using signal contributions from the newly developed three-component multi-b-value DW-MRI model.

EXPERIMENTAL DESIGN

Patients with pathology-proven breast cancer from two datasets (n = 81 and n = 25) underwent multi-b-value DW-MRI. The three-component signal contributions C 1 and C 2 and their product, C 1 C 2, and signal fractions F 1, F 2, and F 1 F 2 were compared with the image defined on maximum b-value (DWI max), conventional apparent diffusion coefficient (ADC), and apparent diffusion kurtosis (K app). The ability to discriminate between cancer and healthy breast tissue was assessed by the false-positive rate given a sensitivity of 80% (FPR80) and ROC AUC.

RESULTS

Mean FPR80 for both datasets was 0.016 [95% confidence interval (CI), 0.008-0.024] for C 1 C 2, 0.136 (95% CI, 0.092-0.180) for C 1, 0.068 (95% CI, 0.049-0.087) for C 2, 0.462 (95% CI, 0.425-0.499) for F 1 F 2, 0.832 (95% CI, 0.797-0.868) for F 1, 0.176 (95% CI, 0.150-0.203) for F 2, 0.159 (95% CI, 0.114-0.204) for DWI max, 0.731 (95% CI, 0.692-0.770) for ADC, and 0.684 (95% CI, 0.660-0.709) for K app. Mean ROC AUC for C 1 C 2 was 0.984 (95% CI, 0.977-0.991).

CONCLUSIONS

The C 1 C 2 parameter of the three-component model yields a clinically useful discrimination between cancer and healthy breast tissue, superior to other DW-MRI methods and obliviating predefining lesions. This novel DW-MRI method may serve as noncontrast alternative to standard-of-care dynamic contrast-enhanced MRI.

Analysis of Renal Diffusion-Weighted Imaging (DWI) Using Apparent Diffusion Coefficient (ADC) and Intravoxel Incoherent Motion (IVIM) Models

Analysis of renal diffusion-weighted imaging (DWI) data to derive markers of tissue properties requires careful consideration of the type, extent, and limitations of the acquired data. Alongside data quality and general suitability for quantitative analysis, choice of diffusion model, fitting algorithm, and processing steps can have consequences for the precision, accuracy, and reliability of derived diffusion parameters. Here we introduce and discuss important steps for diffusion-weighted image processing, and in particular give example analysis protocols and pseudo-code for analysis using the apparent diffusion coefficient (ADC) and intravoxel incoherent motion (IVIM) models. Following an overview of general principles, we provide details of optional steps, and steps for validation of results. Illustrative examples are provided, together with extensive notes discussing wider context of individual steps, and notes on potential pitfalls.This publication is based upon work from the COST Action PARENCHIMA, a community-driven network funded by the European Cooperation in Science and Technology (COST) program of the European Union, which aims to improve the reproducibility and standardization of renal MRI biomarkers. This analysis protocol chapter is complemented by two separate chapters describing the basic concepts and experimental procedure.

Renal Diffusion-Weighted Imaging (DWI) for Apparent Diffusion Coefficient (ADC), Intravoxel Incoherent Motion (IVIM), and Diffusion Tensor Imaging (DTI): Basic Concepts

The specialized function of the kidney is reflected in its unique structure, characterized by juxtaposition of disorganized and ordered elements, including renal glomerula, capillaries, and tubules. The key role of the kidney in blood filtration, and changes in filtration rate and blood flow associated with pathological conditions, make it possible to investigate kidney function using the motion of water molecules in renal tissue. Diffusion-weighted imaging (DWI) is a versatile modality that sensitizes observable signal to water motion, and can inform on the complexity of the tissue microstructure. Several DWI acquisition strategies are available, as are different analysis strategies, and models that attempt to capture not only simple diffusion effects, but also perfusion, compartmentalization, and anisotropy. This chapter introduces the basic concepts of DWI alongside common acquisition schemes and models, and gives an overview of specific DWI applications for animal models of renal disease.This chapter is based upon work from the COST Action PARENCHIMA, a community-driven network funded by the European Cooperation in Science and Technology (COST) program of the European Union, which aims to improve the reproducibility and standardization of renal MRI biomarkers. This introduction chapter is complemented by two separate chapters describing the experimental procedure and data analysis.

A novel bayesian approach with conditional autoregressive specification for intravoxel incoherent motion diffusion-weighted MRI

The Intra-Voxel Incoherent Motion (IVIM) model is largely adopted to estimate slow and fast diffusion coefficients of water molecules in biological tissues, which are used in cancer applications. The most reported fitting approach is a voxel-wise segmented non-linear least square, whereas Bayesian approaches with a direct fit, also considering spatial regularization, were proposed too. In this work a novel segmented Bayesian method was proposed, also in combination with a spatial regularization through a Conditional Autoregressive (CAR) prior specification. The two segmented Bayesian approaches, with and without CAR specification, were compared with two standard least-square and a direct Bayesian fitting methods. All approaches were tested on simulated images and real data of patients with head-and-neck and rectal cancer. Estimation accuracy and maps noisiness were quantified on simulated images, whereas the coefficient of variation and the goodness of fit were evaluated for real data. Both versions of the segmented Bayesian approach outperformed the standard methods on simulated images for pseudo-diffusion (D^∗ ) and perfusion fraction (f), whilst the segmented least-square fitting remained the less biased for the diffusion coefficient (D). On real data, Bayesian approaches provided the less noisy maps, and the two Bayesian methods without CAR generally estimated lower values for f and D^∗ coefficients with respect to the other approaches. The proposed segmented Bayesian approaches were superior, in terms of estimation accuracy and maps quality, to the direct Bayesian model and the least-square fittings. The CAR method improved the estimation accuracy, especially for D^∗ .

Stromal Collagen Content in Breast Tumors Correlates With In Vivo Diffusion-Weighted Imaging: A Comparison of Multi b-Value DWI With Histologic Specimen From Benign and Malignant Breast Lesions

BACKGROUND

Increased deposition and reorientation of stromal collagen fibers are associated with breast cancer progression and invasiveness. Diffusion-weighted imaging (DWI) may be sensitive to the collagen fiber organization in the stroma and could provide important biomarkers for breast cancer characterization.

PURPOSE

To understand how collagen fibers influence water diffusion in vivo and evaluate the relationship between collagen content and the apparent diffusion coefficient (ADC) and the signal fractions of the biexponential model using a high b-value scheme.

STUDY TYPE

Prospective.

SUBJECTS/SPECIMENS

Forty-five patients with benign (n = 8), malignant (n = 36), and ductal carcinoma in situ (n = 1) breast tumors. Lesions and normal fibroglandular tissue (n = 9) were analyzed using sections of formalin-fixed, paraffin-embedded tissue stained with hematoxylin, erythrosine, and saffron.

FIELD STRENGTH/SEQUENCE

MRI (3T) protocols: Protocol I: Twice-refocused spin-echo echo-planar imaging with: echo time (TE) 85 msec; repetition time (TR) 9300/11600 msec; matrix 90 × 90 × 60; voxel size 2 × 2 × 2.5 mm³ ; b-values: 0 and 700 s/mm² . Protocol II: Stejskal-Tanner spin-echo echo-planar imaging with: TE: 88 msec; TR: 10600/11800 msec, matrix 90 × 90 × 60; voxel size 2 × 2 × 2.5 mm³ ; b-values [0, 200, 600, 1200, 1800, 2400, 3000] s/mm² .

ASSESSMENT

Area fractions of cellular and collagen content in histologic sections were quantified using whole-slide image analysis and compared with the corresponding DWI parameters.

STATISTICAL TESTS

Correlations were assessed using Pearson's r. Univariate analysis of group median values was done using the Mann-Whitney U-test.

RESULTS

Collagen content correlated with the fast signal fraction (r = 0.67, P < 0.001) and ADC (r = 0.58, P < 0.001) and was lower (P < 0.05) in malignant lesions than benign and normal tissues. Cellular content correlated inversely with the fast signal fraction (r = -0.67, P < 0.001) and ADC (r = -0.61, P < 0.001) and was different (P < 0.05) between malignant, benign, and normal tissues.

DATA CONCLUSION

Our findings suggest stromal collagen content increases diffusivity observed by MRI and is associated with higher ADC and fast signal fraction of the biexponential model.

LEVEL OF EVIDENCE

3 Technical Efficacy Stage: 3 J. Magn. Reson. Imaging 2020;51:1868-1878.