In the frame of the EU project Restore, Reykjavik University is developing the 1st European database of chondral lesions morphometric and associated 3D models. Thanks to its partnership with the University Hospital Landspitali, 47 knees have been scanned since March 2019. Three different types of subjects have been recruited : 25 degenerative (D), 14 traumatic (T), and 8 healthy acting as control (C). This database will contain information regarding the different type of chondral lesions and the behaviour of the cartilage, based on Computed Tomography (CT) and Magnetic Resonance (MRI) images. Exhaustive measurements, regarding the thickness, the grading of the cartilage, as well as the presence of medical pathologies such as cysts, bone attrition, were performed on those medical images, for each patients, based on a robust radiological approach. It provides a complete overview of the patient‘s condition from those 2D images.
On the 3D aspect, using high performing medical imaging software (Materialise, MIMICS), these images were processed in order to segment and dissociate the different part of the knee : femur, tibia and patella. CT and MRI acquisitions were taken positioning the patient knee in the same way therefore an accurate registration between the datasets was possible, in order to have in the same dataset bone and cartilage. The geometry is studied from this processed images.
From this final dataset, we extract information about density and volume : the mean value of Hounsfield Unit (HU), a scale to quantify the radiodensity, is computed for each bone and cartilage part. This enables us to conduct preliminary analyses that will be displayed in the database. Figure 1 shows the first results of the cartilage density analysis.
Those first results show that the D group globally have the less dense cartilage, which is something that can be expected. The C group have a subsequent higher density in the tibia, and the T group a higher density infor the patella and the femur. The T group is globally a young population that suffered an accident, so is having a cartilage that is healthy before the accident. The medical images were taken a few days after the accident, meaning that the impact on the cartilage is not clearly visible yet. It is then not surprising from those preliminary results to observe that the D group is having the lowest density, which is to be expected from our hypothesis.
In the frame of this study, a novel approach is developed to assess the cartilage condition studying the thickness. Using a CAO software (3matic, Materialise), a wall thickness analysis (figure 2) is performed for each cartilage. This gives us the voxel distribution over the thickness (figure 3). From this distribution, we aim to create a thickness index, associated to presence of holes, to assess and grade the cartilage status.
