Neural networks improve repeatability of intravoxel incoherent motion (IVIM) parameter estimation in pancreatic diffusion-weighted MRI.

Comparison of algorithms for IVIM modelling of DW-MRI with a clinical set of b-values for head and neck cancer: In silico and in vivo analysis.

BACKGROUND Intravoxel incoherent motion (IVIM) modelling of diffusion-weighted magnetic resonance imaging (DW-MRI) quantifies tissue diffusion (Dt), capillary microcirculation (Dp) and perfusion fraction (Fp). However, accurate IVIM modelling using the few b-values acquired in clinical settings is challenging. This study mainly investigated the performance of four algorithms for accurate IVIM parameter modelling with few b-values before exploring the IVIM parameters' biomarker potential for head and neck cancer (HNC) radiotherapy response. METHODS Synthetic IVIM signals with 4, 5 and 11 b-values were generated. DW-MRI was acquired with 11 b-values for 10 HNC patients. Self-supervised (DNNSSL) and supervised (DNNSL) deep neural networks, and least square (LSQ) and segmented (SEG) fitting algorithms estimated IVIM parameters for both synthetic and patient imaging data using 4, 5 and 11 b-values. The algorithm accuracy was evaluated in silico and in vivo by calculating the error between the estimated and ground truth parameter values. A Kaplan-Meier analysis evaluated the associations between progression free survival (PFS) of 20 HNC patients and longitudinal changes in IVIM parameters estimated with 4 b-values. RESULTS LSQ and DNNSL were the least and most accurate IVIM modelling algorithms, respectively. DNNSL, however, was biased towards the mean which limits the parameters' biomarker potential. None of the algorithms, however, estimated Dp accurately with few b-values. Preliminary survival analysis suggested that an increase in Fp during radiotherapy reduced the PFS of HNC patients. CONCLUSIONS SEG and DNNSSL were the preferred algorithms for IVIM modelling few b-values, although reliable estimation was limited to the Dt and Fp parameters.

Evaluation of acquisition strategies in IVIM-DKI brain imaging: Effects on image quality, parameter stability, and comparison with DSC perfusion.

BACKGROUND Intravoxelincoherent motion (IVIM) and diffusion kurtosis imaging (DKI) provide complementary information on cerebral microstructure and microvascular perfusion without contrast agents. However, clinical translation of IVIM-DKI imaging is limited by variability in b-value selection and signal averaging, leading to inconsistent parameter estimates, and by limited validation against contrast-enhanced perfusion MRI. This study aimed to identify a robust IVIM-DKI acquisition strategy and evaluate its relationship with DSC (Dynamic Susceptibility Contrast)-derived perfusion metrics. METHODS This prospective study included 29 adults undergoing brain MRI on a 3T scanner (uMR 780). IVIM-DKI was acquired using four protocols (A1-A4) with varying b-value density and averaging strategies, including high b-value sampling (A1), optimized low b-value-weighted acquisition with targeted averaging (A2), and reduced sampling schemes (A3-A4). DSC perfusion imaging was subsequently performed. Parametric maps were generated using vendor-provided tools. Standardized gray matter (GM) and white matter (WM) ROIs were placed by two blinded radiologists. Image quality (SNR, CNR), stability (CV), and GM-WM contrast were compared using Friedman tests. Spearman correlation and regression-based analysis were used to assess IVIM-DSC relationships. RESULTS A2 demonstrated higher SNR and CNR and lower CV for most IVIM-derived perfusion parameters (p < 0.05), while certain diffusion and kurtosis metrics showed superior performance with alternative acquisitions (MK with A1 and MD with A3). A2 also achieved the highest GM-WM ratios for f (2.25), rBF (2.89), and D* (1.53), indicating improved tissue contrast. IVIM perfusion parameters showed moderate correlations with DSC-derived perfusion metrics, with f demonstrating the strongest association with rCBF (ρ = 0.67) and D* showing the highest correlation with rCBV (ρ = 0.51). Equation-based analysis demonstrated a strong relationship between rCBV and c_CBV (r = 0.77, p < 0.001), whereas no meaningful association was observed between MTT and c_MTT (r = 0.03, p = 0.84). CONCLUSION An optimized low b-value-weighted acquisition with targeted averaging (A2) provided the strongest overall balance between image quality, parameter stability, tissue contrast, and correspondence with DSC-derived perfusion measures. IVIM perfusion metrics demonstrate physiologically meaningful but non-equivalent relationships with DSC, supporting their complementary role in perfusion assessment. PLAIN LANGUAGE SUMMARY Brain MRI scans can provide information about brain tissue structure and blood flow, but finding the best scan settings remains a challenge. This study compared several acquistion strategies for IVIM-DKI MRI in adults to identify which approach produced the most reliable image measurements and how these measurements related to a standard blood flow imaging technique. This study found that one optimized scanning method provided the best balance of image quality, measurement stability, tissue contrast, and agreement with blood flow measurements. This matters because more reliable brain imaging may help clinicians better assess brain health without relying solely on contrast agents.

Parameter Estimation for the Intravoxel Incoherent Motion-diffusional Kurtosis Imaging Model Using Synthetic Q-space Learning Toward Breast Tumor Diagnosis.

PURPOSE Synthetic Q-space learning (synQSL), which uses only synthetic data in the training of regressors, has demonstrated promising results in parameter estimation for brain diffusion MRI (dMRI). This study aimed to evaluate the performance of synQSL in breast dMRI for intravoxel incoherent motion-diffusional kurtosis imaging (IVIM-DKI) parameter estimation, with comparing several types of regressors and conventional fitting by nonlinear least squares fitting (NL-LSF). METHODS In synthesizing data for synQSL, IVIM-DKI parameters were sampled from uniform distributions and substituted into the signal model along with b-values to generate diffusion-weighted imaging (DWI) signals. In addition, Rician noise was mixed to the signals. We prepared datasets of 105 and 106 samples, and trained multi-layer perceptron (MLP), Kolmogorov-Arnold networks (KAN), and random forest (RF) regressors for synQSL. The performance of the parameter estimation methods including NL-LSF was evaluated using a digital phantom including various value combinations of IVIM-DKI parameters and clinical data of 67 cases (13 benign and 54 malignant lesions) through quantitative analysis and visual assessment. RESULTS In the digital phantom, the regressors of synQSL achieved significantly lower root mean square error (RMSE) for f, D∗, and D than NL-LSF (P < 0.05) with less variation among the parameter sets. In clinical datasets, synQSL not only improved the visual quality of parameter maps but also showed significant differences between benign and malignant lesions in parameters while the NL-LSF failed. Furthermore, the estimation times for all synQSL regressors were substantially shorter than that of NL-LSF. Among the synQSL regressors, MLP showed superior properties including computational cost. CONCLUSION synQSL demonstrated superior parameter estimation performance compared to NL-LSF in breast IVIM-DKI analysis. In this study, MLP was considered the most suitable regressor for synQSL among those we examined, based on the balance between estimation accuracy and computational costs.

UvA-DARE (Digital Academic Repository) MRI of pancreatic cancer for radiotherapy

Intravoxel incoherent motion imaging with improved image quality by deep learning reconstruction for prediction of lymph-vascular space invasion in patients with cervical cancer.

The Combination of IVIM-DWI and BOLD-fMRI to Evaluate the Early Efficacy of Photodynamic Microenvironment-Modulating Therapy in Nasopharyngeal Carcinoma.

BACKGROUND Photodynamic microenvironment-modulating therapy (PMMT) represents a promising strategy for nasopharyngeal carcinoma (NPC). However, early assessment of therapeutic response remains challenging. PURPOSE To evaluate whether intravoxel incoherent motion diffusion-weighted imaging (IVIM-DWI) and blood oxygen level-dependent functional MRI (BOLD-fMRI) can noninvasively detect early responses to PMMT in an NPC xenograft model. STUDY TYPE Experimental, animal model. ANIMAL MODEL Subcutaneous 5-8F NPC xenograft model (110 female BALB/c nude mice). FIELD STRENGTH/SEQUENCE 3.0 T, T2WI FSE sequence, IVIM-DWI single-shot echo-planar imaging sequence, and BOLD-fMRI SPGR sequence. ASSESSMENT Following tumor establishment, mice were randomly assigned to five treatment groups: PBS, NH2-MIL-101(Fe), PPa@NH2-MIL-101(Fe) + near-infrared (P@MIL+NIR), Doxy@NH2-MIL-101(Fe) (D@MIL), and PPa + Doxy@NH2-MIL-101(Fe) + NIR (PD@MIL+NIR). IVIM-DWI (D and f) and BOLD-fMRI (R2*) were acquired before and after treatment to assess changes in tumor microstructure, perfusion, and oxygenation. The proliferation, apoptosis, angiogenesis, and hypoxia of NPC tumor were evaluated by Ki-67 immunofluorescence staining, TUNEL immunofluorescence staining, VEGF immunohistochemical staining, and HIF-1α immunohistochemical staining. STATISTICAL TESTS One-way ANOVA, the intraclass correlation coefficient (ICC), Pearson's correlation analysis, and the Least Significant Difference (LSD) test for post hoc pairwise comparisons, p value < 0.05 was considered significant. RESULTS Good and excellent agreements between two evaluators can be seen (The range of ICCs was from 0.864 to 0.985). In P@MIL+NIR, D@MIL, and PD@MIL+NIR groups, D and f values initially decreased and then increased, while R2* values peaked at 0.5 h (63.20 ± 2.57, 61.62 ± 0.98, and 67.38 ± 1.37) and gradually declined. D, f, and R2* values were significantly correlated with the histological staining results. DATA CONCLUSION The combination of IVIM-DWI and BOLD-fMRI could be used in monitoring the early therapeutic responses to PD@MIL-based PMMT in NPC. EVIDENCE LEVEL 1. TECHNICAL EFFICACY Stage 1.