University of Twente Student Theses


Evaluation of options to improve accurate execution of osteotomies of the lower limb

Rook, Anne (2022) Evaluation of options to improve accurate execution of osteotomies of the lower limb.

[img] PDF
Abstract:Accuracy and precision are important when executing a corrective osteotomy around the knee. However, it can be difficult to achieve a high accuracy without additional tools. Therefore, the aim of this thesis was twofold. The first aim was to map the corrective osteotomy procedure and to identify surgical challenges and needs of an orthopedic surgeon to overcome these challenges in clinical practice. The second aim was to research the feasibility of using ultrasound (US) to estimate the bone pose during surgery. A corrective osteotomy around the knee was used as a case study here. First, interviews among orthopedic surgeons were held to gain insight into the difficulties and challenges they experience when executing a corrective osteotomy around the knee or when using additional tools to perform surgery more accurately. These interviews showed that variation exist among surgeons regarding the leg stance that is aimed for, the execution procedure of a correction osteotomy around the knee and the importance of a surgical plan within this process. Hence, it was decided to focus on improving the registration procedure (which is the alignment of a surgical plan to the anatomy of a patient) to make a preoperative plan more easily available and usable within the operation room. A trade-off was made between different registration techniques. It was decided to further investigate the possibility of using US as a registration technique in a surgical setting. Therefore, first computer simulations that mimicked the registration procedure were performed to evaluate whether either A-mode or a combination of A- and B-mode US sensors should be used. Different Monte-Carlo simulations using a sensor model to mimic realistic US measurements were performed to determine the number of sensors that were needed to obtain a desired accuracy of 1 mm and 1 degree, and to determine the effect of the sensor positioning and measurement inaccuracy on the registration accuracy. Second, an experiment was conducted to determine how accurate US could detect the bone surface. Therefore, US images of a cadaver were made inside a CT scanner to acquire simultaneous imaging of both US and CT (ground truth). In total, the bone localization error between CT and US was determined in 11 static measurements made on a human femoral bone cadaver. The simulation findings suggested that at least 22 A-modes or 6 A-modes and 7 B-modes were needed to obtain a registration accuracy of 1 mm and 1 degree. A combination of A- and B-mode sensors was more robust to measurement inaccuracy compared to using only A-mode sensors. Still, all US measurements depended heavily on the amount of measurement inaccuracy present in the data. Verification of the bone localization error via an experiment revealed that the mean absolute difference in bone surface depth between CT and US was -0.47 +- 0.90 mm. This accuracy was within the voxel range of the CT scan. When repeating the Monte Carlo simulation with this B-mode inaccuracy level, the registration accuracy worsened: using 6 A-modes and 7 B-modes resulted in a registration accuracy of around 4 mm and 4 degrees. This thesis shows that theoretically, it is possible to use US as a bone localization tool to align a surgical plan with the anatomy of the patient, although the number of sensors that are needed to obtain the set accuracy of 1 mm and 1 degree is not clinically feasible yet. This thesis provides the next step towards an accurate execution of a surgical plan within orthopedics, providing more predictable outcomes in the future. For the application of US as a registration technique in clinical practice for corrective osteotomies, the usability of this system as a whole should be tested in a surgical setting. Indeed, the interviews showed good usability is one of the most important factors in the adoption of this technique in clinical practice. Before that time, first, some technical aspects of this system should be improved to increase its accuracy. Therefore, more research regarding a more sufficient registration algorithm, accurate, real-time, automatic bone detection in US images, and US sensor positioning and localization should be performed. With this information, a predictive sensor model can be made more realistic by identifying the magnitude of the different sources of inaccuracy more accurately, and by evaluating the effect of combining multiple sources of errors.
Item Type:Essay (Master)
Radboudumc, Nijmegen, The Netherlands
Faculty:TNW: Science and Technology
Subject:44 medicine, 50 technical science in general
Programme:Technical Medicine MSc (60033)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page