Purpose: Spatial heterogeneity of hematological disease is suspected to affect treatment response. In this study, we developed a method to quantitatively assess heterogeneity in bone marrow response to treatment via a voxel‐to‐voxel analysis for intra‐and inter‐regional marrow assessment. Methods: The method begins with articulated registration — registration of whole body CTs by auto‐segmenting individual bones then rigidly aligning them — of pre‐and post‐treatment CTs. Resulting segmentations and transformations are applied to the respective PET data allowing for voxel‐to‐voxel comparisons. Intra‐and inter‐regional heterogeneity of PET uptake and response is assessed through four distinct analyses: (1) Intra‐regional uptake heterogeneity, (2) Intra‐regional response heterogeneity, (3) Interregional uptake heterogeneity, and (4) Inter‐regional response heterogeneity. Mean SUV (SUVmean) and SUV coefficient of variation (SUVCoV)) are calculated within each region at each time point. Response (R) is calculated as a change in SUVmean). The method was applied to two patients with acute myeloid leukemia having received standard ‘7+3’ induction chemotherapy, but experiencing different clinical responses. Results: In the two patients, our method revealed wide variation in uptake and response within skeletal regions. Intra‐regionally we observed high uptake heterogeneity in all regions (SUVCoV</sub≫ 0.8) as well as high response heterogeneity; this was evidenced by standard deviations of Rmean) spanning 0.7xRmean) to 1.3×Rmean). Despite clear heterogeneity within each region, inter‐regional uptake heterogeneity was low (CoV<0.05 for both patients pre‐treatment). Interestingly, the method revealed significantly different inter‐regional response heterogeneity between the patients, with the clinically responding patient having a more homogeneous response (variation of Rmean) <0.06), than the non‐responding patient (variation of Rmean) <0.14). Conclusion: The newly developed heterogeneity assessment method allows for a quantitative assessment of bone marrow response. The localization of treatment failure could assist in clinical decision making by discriminating between responding and non‐responding patients.