University of Twente Student Theses


Development of patient-specific finite element models for the simulation of strain adaptive tibial bone remodeling after total knee replacement

Langen, I.N. van (2022) Development of patient-specific finite element models for the simulation of strain adaptive tibial bone remodeling after total knee replacement.

[img] PDF
Abstract:An important factor in failure of total knee replacement (TKR) is periprosthetic bone loss, due to post-operative bone remodeling. An implant in the knee changes the stresses and strains within the knee. As bone adapts to changes of these mechanical loads, according to Wolff’s law, TKR will thus lead to bone remodeling. In this project, the bone remodeling after TKR was studied by finite element (FE) modeling. The first goal was to develop a workflow in which available CT patient data can be used to create patient-specific FE models of the pre- and post-operative tibia which can be simulated with an existing strain adaptive remodeling algorithm. The second goal was to compare the outcomes of these remodeling simulations with clinical dual energy X-ray absorptiometry (DEXA) data. For this project, 5 patient’s pre- and post-operative CT scans from a Japanese dataset were used. For the creation of FE models, an existing workflow was largely adapted. The implant position and orientation in the post-operative scans were extracted and used to resemble the clinical situation in the model. By using the image intensities in the pre-operative CT scans, bone material properties could be assigned to all elements of the models. The loads used to resemble daily activities were from the OrthoLoad dataset and scaled to bodyweight for each of the patients. The post-operative bone remodeling was simulated to resemble 5 years of clinical time after surgery. The resulting bone mineral densities at 0 weeks, 2 weeks, 6 months, 12 months, 24 months, 3 years and 5 years post-surgery were studied. Virtual DEXA scans were created out of the BMD data and DEXA values of three regions of interest (ROI) were obtained. Those virtual DEXA values were compared to clinical DEXA data at all 7 time points. In the clinical data the BMD in the medial ROI on average decreased from 0.97 to 0.57 g/cm2, while in simulations each of the models showed different behavior. Three of the simulations showed an increase in BMD over time and two of the simulations showed a decrease over time. The distal ROI on average had an almost constant BMD value in the clinical data around 0.75 g/cm2 but did increase from 0.53 to 0.72 g/cm2 in the simulation results. For the lateral ROI, the clinical data on average showed a decrease of 0.71 to 0.58 g/cm2 although each of the patients showed different behavior. In the simulations an average decrease from 0.35 to 0.20 g/cm2 was observed. The results of patient-specific models used in the strain adaptive remodeling algorithm do not compare with clinical remodeling data. The remodeling algorithm has thus not been proven to work for these models. This can be attributed to a number of uncertainties and assumptions in both the clinical data, like unknown patient activity and measurement errors in the BMD data and in the FE model, like the bone remodeling algorithm, the applied boundary conditions, used alignments and used implant orientations.
Item Type:Essay (Master)
Radboud UMC, Nijmegen
Faculty:ET: Engineering Technology
Subject:44 medicine, 52 mechanical engineering
Programme:Mechanical Engineering MSc (60439)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page