Technical Know-how in Image Processing

We support your Clinical Development and Publication

About us

The Human Morphology Service Center (HMSC) of the AO Development Institute (ADI) maintains a database of CT scans of cadaveric human specimens used in other projects in the AO institutes. Additionally, the database includes virtual bone models of the entire skeleton which are derived from the CT scans. This database is registered at the relevant regulatory authorities in Switzerland. A major task of the HMSC is to offer support in implant design and optimization to the AO Network and external partners, based on its database, know-how, and resources.

Services

Since its foundation in 2002 the HMSC has acquired more than a thousand bone data sets, know-how in medical image computing, and statistical shape analysis of 3-D bone models. Moreover, it has excellent software resources (Amira, Geomagic, Unigraphics, Matlab) to accomplish its support task. The service offerings include CT scans of cadaveric bones, image processing, segmentation, reverse engineering, statistical shape analysis, principal components analysis (PCA), rapid prototyping and CMF related image computing.

Example

An example of our services is the response to a request from the AO Hand Expert Group (HAEG) which was interested in the shape variability of the distal radius and shaft, with respect to the meta-diaphyseal volar distal radius plate. Together with a medical and a technical mentor, a sample of 15 radii from the database was selected and evaluated. To identify the region of interest (ROI) on the virtual specimens, we aligned a virtual template of a plate prototype on the radii, and placed landmarks on the bone surfaces through the first eight distal plate holes as illustrated in Fig 1. These landmarks were used to align the bones to each other by a generalized Procrustes fit, which minimises the sum of the distances between homologous landmarks.

Fig 1: The placement of homologous landmarks on the radius surface through the holes of a manually aligned template of a meta-diaphyseal volar distal radius plate prototype.

Then, the aligned surfaces were scan-converted and superposed, resulting in a 3-D image in which the grey values of the voxels represent the number of bone model intersections as illustrated in Fig 2. This 3-D image carries statistical information for the complete sample. It enables us to extract the median surface and to visualize local deviations from the median bone.

Fig 2: The different cuts through the superposed radii give insight into the shape variability with respect to the region of interest (the plate). The red color represents the median segmentation used for median surface generation.

Fig 3 shows three surfaces of the aligned and superposed volumes of the radii:

1. envelope of all radii

2. median intersection

3. common intersection

The median region of interest (ROI) is intended to be a template for potential preshaped plates. It can be observed that the shape variability is very small around the landmarks used for aligning the bones and that the variability increases with a growing distance from them. By carefully choosing only the ROI relevant for the implant, the aligning and averaging of the bone surface results in optimal templates for preshaped implants. The implant has to fit well only to the ROI. Variations outside are of no interest for the plate design. Such investigations can result in a basis for decision-making to engineers of implant manufacturers.

Fig 3: The surfaces of the envelope, the median, and the common intersection surfaces (from left to right) of the aligned and superposed volumes of the radii of the sample.

34-year-old male, with open fracture after motor vehicle accident.

Fig 1ab: Preoperative x-rays; primary stabilization with external fixator.

Fig 2ab: Eight months postoperative; full forearm rotation and 75% grip strength and wrist motion.

Fig 3ab: Full motion recovery.

Case provided by Jesse Jupiter, Boston