Volumetric bone marrow cellularity (VBMC) assessment from routinely processed trephines using three-dimensional x-ray histology and Gaussian peak modelling

OBJECTIVE: Bone marrow cellularity is routinely estimated from a small number of 2D histology sections, making assessment sensitive to section representativeness, processing artefacts and observer interpretation. Three-dimensional X-ray histology (XRH), using computed microtomography (µCT), enables non-destructive whole-block imaging of trephine biopsies. This study evaluated whether XRH combined with Gaussian-peak modelling could provide a pragmatic whole-block volumetric bone marrow cellularity (VBMC) estimate from formalin-fixed paraffin-embedded (FFPE) trephine biopsy blocks. APPROACH: Six routinely processed FFPE bone marrow trephine blocks were imaged using µCT-based XRH at ~15 µm resolution. VBMC was defined as the red-marrow (RM) fraction of the marrow soft-tissue compartment, RM/(RM+intra-biopsy wax), with wax serving as the volumetric proxy for adipocyte/yellow marrow space. Whole-volume greyscale histograms were modelled using a three-peak Gaussian approach representing intra-biopsy wax, red marrow and demineralised trabecular matrix. Peak-height and area-under-the-curve metrics were compared with whole-volume 3D segmentation and clinical 2D cellularity estimates. MAIN RESULTS: Gaussian peak modelling successfully approximated the segmented tissue-phase distributions. The peak-height-derived VBMC metric showed the closest agreement with whole-volume 3D segmentation, with an average absolute percentage difference of 9.3%, compared with 18.6% for clinical expert 2D cellularity estimates. The area-under-the-curve metric followed similar trends but consistently overestimated VBMC. Clinical 2D cellularity broadly followed whole-biopsy trends but showed one discordant case not explained by slice-position sampling alone. XRH also enabled unrestricted virtual reslicing and visualisation of sectioning-associated artefacts prior to further microtomy. SIGNIFICANCE: Pre-sectioning XRH combined with Gaussian peak modelling provides a rapid, segmentation-free route to volumetric cellularity estimation from intact clinical FFPE trephine blocks. The approach supports objective whole-biopsy assessment while remaining compatible with routine histopathology workflows, reflecting the expected limitations of section-based visual estimation despite its role as the current clinical standard. In the near term, it could provide a non-disruptive adjunct to conventional 2D cellularity reporting, pending larger validation studies.