Feasibility of multi‐energy CT and fat fraction MRI for characterizing active bone marrow response to pelvic external beam radiation therapy

Abstract Background Pelvic external beam radiation therapy (EBRT), frequently exposes a substantial fraction of active bone marrow (ABM), leading to radiation‐induced marrow suppression and associated hematologic toxicity. Accurate delineation and quantitative assessment of ABM are therefore essential to enable marrow‐sparing radiotherapy strategies. Purpose To identify and quantify changes in ABM after EBRT using multi‐energy CT (MECT) and fat fraction MRI (FFMRI). Methods Patients with gynecological malignancies underwent MECT and FFMRI pre‐ and post‐radiation therapy. ABM was delineated on both imaging modalities using thresholding where ABM was defined as voxels with CT numbers above the mean in MECT‐derived virtual non‐calcium (VNCa) datasets and voxels with percent fat fraction (%FF) below the mean on the FFMRI. ABM volumes and mean HU/%FF values were evaluated before and after treatment. A Wilcoxon‐signed rank test was computed to compare paired MECT versus FFMRI ABM volume percent changes. Changes in ABM as a function of dose were also evaluated. A paired t‐test was computed for each dose gradient (significant p < 0.05). Results The average Dice Similarity Coefficient (DSC) between the MECT and FFMRI pre‐treatment ABM volumes was 0.57 (range: 0.50–0.68). The mean volumetric change in ABM volume between pre‐ and post‐treatment scans was −37%15% (range −57% to −12%) and −66%27% (range 8% to −94%) for MECT and FFMRI, respectively. All paired patients demonstrated greater ABM volume reduction with FFMRI than with MECT, and this difference was statistically significant ( p = 0.0078). ABM changes were observed as a function of dose: A decrease in mean CT numbers and an increase in %FF corresponds to an increase in dose. CT number decreased significantly at doses ≥10 Gy ( p ≤ 0.005), while %FF increased significantly across all dose gradients ( p ≤ 0.0017), with both metrics showing strong dose‑response behavior at higher doses ( p < 10 − 6 ). Conclusions This feasibility study demonstrates the utility of using MECT and FFMRI for delineating ABM in patients receiving pelvic EBRT. These imaging modalities showed good agreement in characterizing ABM volumes changes and radiation‐induced marrow suppression. These findings support further investigation of ABM segmentation into the RT‐planning workflow to facilitate marrow‐sparing strategies.