The accuracy of the analyses for studying the three dimensionaltrabecular bone microstructure rely on the quality of the segmentationbetween trabecular bone and bone marrow. Such segmentationis challenging for images from computed tomography modalities thatcan be used in vivo due to their low contrast and resolution. For thispurpose, we propose in this paper a granulometry-based segmentationmethod. In a first step, the trabecular thickness is estimated by usingthe granulometry in gray scale, which is generated by applying the openingmorphological operation with ball-shaped structuring elements ofdifferent diameters. This process mimics the traditional sphere-fittingmethod used for estimating trabecular thickness in segmented images.The residual obtained after computing the granulometry is comparedto the original gray scale value in order to obtain a measurement ofhow likely a voxel belongs to trabecular bone. A threshold is applied toobtain the final segmentation. Six histomorphometric parameters werecomputed on 14 segmented bone specimens imaged with cone-beam computedtomography (CBCT), considering micro-computed tomography(micro-CT) as the ground truth. Otsu’s thresholding and AutomatedRegion Growing (ARG) segmentation methods were used for comparison.For three parameters (Tb.N, Tb.Th and BV/TV), the proposedsegmentation algorithm yielded the highest correlations with micro-CT,while for the remaining three (Tb.Nd, Tb.Tm and Tb.Sp), its performancewas comparable to ARG. The method also yielded the strongestaverage correlation (0.89). When Tb.Th was computed directly fromthe gray scale images, the correlation was superior to the binary-basedmethods. The results suggest that the proposed algorithm can be usedfor studying trabecular bone in vivo through CBCT.