Investigation of IVIM diffusion weighted MRI in muscles of the lower limb at rest

Abstract

Aim

Intravoxel incoherent motion diffusion-weighted magnetic resonance imaging (IVIM DW-MRI) provides a means for non-invasive, simultaneous assessment of tissue water diffusion and perfusion characteristics without the need for contrast agent administration. In this study, we aim to analyse the differences in water perfusion and diffusion in the lower limb calf muscles of healthy human volunteers and clinically diagnosed patients with Peripheral Arterial Disease (PAD) with acquired IVIM MR images.

Methods

15 healthy subjects with no mobility issues and 9 clinically diagnosed PAD patients were included in this study. The axial MR images were acquired on the 3T Siemens Biograph mMR (Siemens Healthineers, Germany), with a 12-channel external phased-array coil covering both legs and positioned around the midpoint of the lower leg. Structural T1-weighted images were acquired before a set of baseline (BL) and high-quality (HQ) DWI images which were subsequently registered against each other. BL scan used 16 b-values, with a scan time of 2.5 minutes while HQ scan used 10 b-values with multiple averages to increase the signal-to-noise-ratio (SNR) for the images to be clinically acceptable with a scan time of 7.5 minutes. We utilised an open-source software, Medical Imaging Interaction Toolkit, MITK (version 2018.04.2), to segment the structural T1 images of the lower limb into five muscle groups; tibialis anterior (TA), peroneal muscles (PER), deep posterior calf muscles (DP), soleus muscle (SOL) and the gastrocnemius muscle (GM) regions. An in-house coded MATLAB script was then used to extract the Apparent Diffusion Coefficient (ADC) imaging parameter using the mono-exponential model, perfusion fraction (f), diffusion coefficient (D) and pseudo-diffusion (D*) imaging parameters from the 2-step, fixed D, Intravoxel Incoherent Motion (IVIM) model of the DWI images. Intraclass Correlation Coefficients (ICC) and Mann-Whitney U test were used to evaluate quantitative reproducibility of the paired BL and HQ images, and to determine significant differences between healthy subjects and PAD patients respectively.

Results

Our study found that the parameters f, D, and ADC demonstrated moderate to excellent reproducibility (Healthy ICC:0.694, 0.913, 0.623, PAD ICC:0.715, 0.991, 0.819 respectively), while D* showed weak reproducibility (Healthy ICC:0.344, PAD ICC:0.552) with the potential for future investigation. Imaging parameters ADC and f were found to be statistically significant (p-value < 0.05) when comparing across healthy subjects and PAD patients in BL and HQ scans, indicating these parameters' potential utility as imaging biomarkers to distinguish healthy subjects from clinically diagnosed PAD patients.

Conclusion

From the results of this pilot study, the quantitative parameters obtained by the BL and HQ scans were reproducible except for D*. The results also show that ADC and f are useful imaging parameters in distinguishing between healthy and PAD patients, supporting their potential use as non-invasive imaging biomarkers to characterise restricted diffusion and impaired microvascular perfusion in the lower limbs.