University of Twente Student Theses

Login

Personalized prediction of knee ligaments mechanical properties from MR images

Conte, Riccardo (2021) Personalized prediction of knee ligaments mechanical properties from MR images.

[img] PDF
922kB
Abstract:The health status assessment of the knee ligaments is currently performed via qualitative observation of MR images. Knowing quantitatively to which extent the ligaments can be mechanically stressed before an injury occurs is only possible after invasively collecting a sample of the ligament’s tissue. Computational models for personalized knee joint surgery pre-planning are a possible alternative and they rely on the study and observation of the main tibiofemoral ligaments. Objectives - Magnetic resonance imaging (MRI) of the knee is the standard-of-care imaging modality to evaluate knee disorders, and more musculoskeletal (MSK) MRI examinations are performed on the knee than on any other region of the body and for this reason this study focuses mainly on researching into finding a model capable of predicting knee ligaments mechanical properties from MR images. A recent study conducted by Naghibi et al. [1] explored the potential role of quantitative MRI and dimensional properties, in characterizing the mechanical properties of the main tibiofemoral ligaments. After MR scanning of cadaveric legs, all the main tibiofemoral bone-ligaments-bone specimens were tested in vitro to measure their stiffness and rupture force. The study revealed the potentials of using quantitative MR parameters combined with specimen volume to estimate the essential mechanical properties of all main tibiofemoral ligaments required for subject-specific computational modelling of the human knee joint. This study aims to continue the investigation conducted exploring the chances of creating a model that proves correlation between MR images data and mechanical properties. Methodology – Previous studies observed promising results while deriving average values (like cross sectional area and qualitative MR image parameters) of the whole ligament to produce a correlative model between MR images and mechanical properties. In this study we explore the chance of creating a model that correlate regional characteristics to mechanical properties. The assumption is that the knee ligaments, like most biological tissues, have different characteristics regionally. For this reason, the MR Images composing the dataset are segmented and processed specifically to partition each ligament into smaller portions. By doing so there is the added value of a increasing the size of the dataset. From each of these sub-volumes the following parameters are extracted: volume, cross-sectional area and average MR value (T_1ρ). These parameters are subsequently used together with the mechanical parameters measured during the tensile test by Naghibi et al. [1] to train a linear regression model. To evaluate the results, the output of the created model is compared with the output of the model created by Naghibi et al. Results - The results collected in this study display a little correlation on the training set and no correlation on the test set. The inclusion of the ligament type in the model produced marginally better results. Discussion - From the results of this study, it can be inferred that, in agreement with the literature, the volume (without partition) is the parameter with the highest influence on the correlation between MR images and mechanical properties. Moreover, it has been proven that, in agreement with the literature, the distinction between ligament type improves the correlation. It is not possible yet to conclude whether there is a correlation between MR images data and mechanical properties.
Item Type:Essay (Master)
Faculty:ET: Engineering Technology
Programme:Biomedical Engineering MSc (66226)
Link to this item:https://purl.utwente.nl/essays/86916
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page