Comparative Sensitivity of RSI and ADC to Tumor Cellularity at Ultra-High b-Values
Abstract
Purpose
Apparent diffusion coefficient (ADC) is a widely used marker of tumor cellularity in diffusion MRI, but ADC values depend on b-value selection and are prone to reduced tumor conspicuity in the presence of edema. Restriction Spectrum Imaging (RSI) leverages multi-shell DWI to decompose the signal into a spectrum of diffusivities and isolate the restricted intracellular fraction associated with cellular packing. We investigate whether increasing the maximum diffusion weighting enhances RSI sensitivity to two important biomarkers in neuro-oncology, cell density (ρ) and nucleus-to-cytoplasm ratio (ν), and whether these gains translate into stronger discrimination than conventional ADC.
Methods
Monte Carlo simulations generated multi-shell DWI signals in synthetic 3D brain voxels with varied ρ and ν. Acquisition protocols included ultra-high b-values, up to 8000 s/mm2, and multiple echo times. RSI metrics were estimated by Tikhonov-regularized linear inversion of the simulated multi-shell dataset. ADC was obtained by mono-exponential fitting of the same DWI signals. Sensitivity to microstructural changes was quantified using Cohen’s d computed between the 15th percentile extremes of ρ and ν for each metric.
Results
RSI showed increasing sensitivity to ρ and ν as the maximum b-value went from b=4000 s/mm2 to b=8000 s/mm2. Cohen’s d comparisons confirmed that RSI consistently outperformed ADC. For example, at b=8000 s/mm2 the effect size for ρ discrimination was approximately double for RSI compared with ADC (d_RSI=21.1 vs d_ADC=11.3), with similar trends for ν. These gains held across all simulated echo times.
Conclusion
Simulations indicate that RSI provides markedly greater sensitivity to tumor microstructure relative to ADC under ultra-high b-value conditions. The superior effect sizes and tumor contrast suggest that including ultra-high b shells in clinical DWI protocols may substantially improve RSI-based detection of cellular tumor features by preferentially isolating truly restricted intracellular diffusion, thereby offering a more direct and specific measure of tumor cellularity.