Motivation

Whole-body MRI is increasingly being used to detect bone involvement in prostate cancer.¹ Diffusion-weighted imaging (DWI) and apparent diffusion coefficient (ADC) mapping are important components of the multiparametric approach recommended for whole-body metastasis screening.² Recent studies have shown that multicompartmental diffusion modeling outperforms these conventional DWI techniques for assessing cancer within the prostate,³ but such modeling has yet to be applied to whole-body imaging.

In this study, we optimized a multicompartmental model for describing diffusion signal throughout the body and applied it to examine the diffusion characteristics of metastatic bone lesions in vivo. The potential clinical utility of this model was examined by comparing lesion conspicuity on model-derived images against their conspicuity on conventional DWI.

Methods

This prospective study included 30 patients with prostate cancer who underwent an extended whole-body MRI examination in addition to routine clinical imaging. Standard-of-care evaluation identified 107 bone lesions in 25 of these patients.

Whole-body MRI acquisition
MR imaging was performed on a 3T clinical scanner (Discovery MR750; GE Healthcare). Five stations were imaged for each patient, corresponding roughly to the head, chest, abdomen, pelvis, and thighs. At each station, an axial volume of multi-shell diffusion data was acquired using 4 b-values: 0, 500, 1000, and 2000 s/mm², sampled at 1, 6, 6, and 12 unique diffusion-encoding gradient directions, respectively (default tensor, TR: 4750ms, TE: 75ms, matrix: 80×80 resampled to 128×128, FOV: 400mm, slices: 46, slice thickness: 6mm). For anatomical reference, a high resolution T2-weighted volume was also acquired at each station with scan-coverage identical to that of the multi-shell DWI volume (TR: 1350ms, TE: 113ms, matrix: 384×224 resampled to 512×512, FOV: 400mm, slices: 46, slice thickness: 6mm).

MRI data post-processing
Each multi-shell DWI volume was first corrected for distortions due to B0-inhomogeneity, gradient nonlinearity, and eddy currents.⁴ The signal intensity of each DWI volume was corrected to account for noise.⁵ Isotropic diffusion was assumed, so directional DWI volumes at each b-value were averaged. Conventional ADC maps were computed by fitting the DWI data to a monoexponential signal model.⁶

Regions of interest (ROIs) were defined on the DWI volumes. Whole-body ROIs (excluding the head) were defined for 10 patients. In 5 patients without metastases, tissue-specific ROIs were defined in the pelvic and thigh stations (to avoid respiratory motion artifacts), specifically over the bladder (including urine), prostate, testes, and subcutaneous fat. In 25 patients, ROIs were defined over each identified bone lesion.

Multicompartmental modeling and analysis
Restriction spectrum imaging (RSI) is a multicompartmental modeling framework that describes the DWI signal thusly: $$S(b)=\sum_{i=1}^{K}C_ie^{-bD_i}$$ where S(b) denotes the noise-corrected DWI signal at a particular b-value, K is the number of tissue compartments, C_i describes the contribution of a particular compartment to the overall signal, and D_i is the compartmental diffusion coefficient. A global fitting of this model to the DWI data within the 10 whole-body ROIs (~9 million voxels) was performed,³ with K ranging from 2 to 4. The relative Bayesian Information Criterion (ΔBIC⁷) and model-fitting residual of each model was recorded to evaluate how well it described the whole-body DWI data. Fitting residual was also examined at the voxel-level and ROI-level within specific tissues to examine how the fit of each model varied between anatomical regions.

For the optimal RSI model (the model with lowest ΔBIC), signal-contribution (C_i) maps were computed for each patient via nonnegative least-squares fitting of the model to the signal-vs-b-value curve from each voxel.³

Compartmental signal fractions for the optimal model were computed by normalizing the C value of each compartment by the sum of all C values: $$$C_i/\sum_{i=1}^{K}C_i$$$. Signal fractions were compared between bone lesions and other tissues using two-sample t-tests (α=0.05).

Lesion conspicuity
Bone-lesion conspicuity⁸ was defined as the mean signal within the lesion ROI divided by the mean signal within a control ROI defined by reflecting the lesion ROI laterally across the spine. Conspicuity was calculated for each lesion on the conventional DWI images, ADC map, and signal-contribution maps from the optimal RSI model. Paired t-tests (α=0.05) were used to determine if lesion conspicuity was significantly higher on RSI maps compared to conventional images.

Results

The lowest BIC and model-fitting residuals were observed from the 4-compartment model (Figures 1 and 2). Optimal D values for the 4-compartment model were 0, 1.2e-3, 2.9e-3, and >3.0e-2mm²/s. Signal-contribution (C_i) maps computed using this model are shown in Figure 3 for a patient with bone lesions, alongside conventional MR images. Figure 4 illustrates the significant increase in lesion conspicuity on RSI C₁ and C₂ maps compared to conventional ADC maps (C₁: P<0.001, C₂: P=0.02) or DWI images (C₁: P<0.001, C₂: P<0.04). Figure 5 shows that compartmental signal fraction was significantly higher in compartments 1 (P<0.04) and 2 (P<0.01) of lesions than in other tissues.

Discussion

An optimized 4-compartment RSI model provides a more comprehensive characterization of whole-body diffusion than conventional DWI methods. Compartmental signal-contributions revealed by this model may help to assess microstructural changes that accompany prostate-cancer bone involvement. Improved conspicuity of lesions on RSI signal-contribution maps may help to discriminate between cancerous and benign tissues during whole-body cancer screening.