Dual modality neutron and x-ray tomography for enhanced image analysis of the bone-implant interface


The quality of the bone tissue formed at the contact interface with a metallic implant is of utmost importance for the structural integrity of implant fixation. This quality can be assessed, e.g., based on its 3D microstructure, which is commonly visualised using x-ray tomography. However, due to large differences in interaction strength between the metal of the implant and the bone tissue, image artifacts appear at the bone-implant interface. This makes is challenging to visualise the bone structure. We have previously shown that neutron tomography is a promising technique for visualising the peri-implant bone, as no image artifacts are present. To highlight and explore the complementarity of neutron and x-ray tomography, rat tibiae with metallic implants were imaged with both modalities. Comparing the images, we saw that the peri-implant bone was indeed more clearly visible in the neutron images. In addition, soft skeletal tissues were captured in the neutron images whilst they were barely visible in the x-ray images. This could possibly be exploited to track changes in soft tissue distribution due to aging or resulting from different medical treatments.


To highlight and explore the complementarity of neutron (NT) and x-ray (XRT) tomography, dry rat tibiae (N = 11) with hollow metallic implants filled with a biomaterial (calcium sulphate/hydroxyapatite) were imaged with both modalities at the NeXT-Grenoble beamline at ILL, France. The rats were divided in a control group (n = 5) and a treated group (n = 6) where the biomaterial was mixed with bioactive molecules (zoledronic acid and recombinant human bone morphogenic protein-2) known to increase bone formation. The acquired NT and XRT images had a virtual isotropic voxel size of 18.6 µm3 and 24.4 µm3, respectively. A dedicated image registration algorithm employing the joint histogram was used to superimpose the images, which were then compared in terms of visualised structures and contrast-to-noise ratio. The complementarity was further investigated via analysis of the joint histograms, in which peaks with well-defined grey values corresponding to the different material phases observed in the specimens were identified and compared. The effect of the treatment was evaluated trough comparison of bone volume fraction in a region close to the implant.

Figure 1. a) Neutron (NT) and x-ray (XRT) tomographic images of a rat tibia with a hollow metallic implant, as well as a checkerboard comparison of both modalities. The colour bar shows relative attenuation. b) The volume fraction of peri-implant bone inside a region of interest (marked in black) was compared between treatment groups and modalities. c) Both modalities captured the differences in bone volume fraction between the treatment groups. No differences were seen between the modalities.

What´s next?

We now aim to further elucidate the possibilities of using neutron tomography for structural and mechanical analysis of bone-implant integration. In an ongoing study, we are looking at the effects of hydration state on image quality and bone mechanical properties as the hydrogen content in the wet tissue makes structural characterisation challenging. Using this dual modality approach shows promise for gaining additional and novel information about bone and soft tissue. Hence, further exploration and improved analysis is necessary.


This work was carried out by a team from Lund University (Elin Törnquist, Sophie Le Cann, Erika Tudisco, Johan Hektor, Deepak B. Raina, Magnus Tägil, Stephen A. Hall, and Hanna Isaksson) together with support from beamline scientists at Institut Laue-Langevin, Grenoble, France (Alessandro Tengattini), and researchers at University Grenoble Alpes (Edward Andò and Nicolas Lenoir). This research was funded by the Swedish Foundation for Strategic Research (SSF) within the Swedish national graduate school in neutron scattering (SwedNess, GSn15 - 0008).


Cite: Physics in Medicine and Biology 66 (2021) 135016

For the original articles, please visit:





PhD student, MSc. Elin Törnquist

Lund University