Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Reconstruction of 3D models of the femur from planar x-rays for surgical planning
Thesis (MEng)--Stellenbosch University, 2023.
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | en_ZA |
| Published: |
Stellenbosch : Stellenbosch University
2023
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867613773646266368 |
|---|---|
| access_status_str | Open Access |
| author | Oosthuizen, Jodri |
| author2 | Van der Merwe, Johan |
| author_browse | Oosthuizen, Jodri Van der Merwe, Johan |
| author_facet | Van der Merwe, Johan Oosthuizen, Jodri |
| author_sort | Oosthuizen, Jodri |
| collection | Thesis |
| dc_rights_str_mv | Stellenbosch University |
| description | Thesis (MEng)--Stellenbosch University, 2023. |
| format | Thesis |
| id | oai:scholar.sun.ac.za:10019.1/127372 |
| institution | Stellenbosch University (South Africa) |
| language | en_ZA |
| last_indexed | 2026-06-10T12:41:28.315Z |
| license_str | Other — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from SUNScholar — Stellenbosch University Repository |
| publishDate | 2023 |
| publishDateRange | 2023 |
| publishDateSort | 2023 |
| publisher | Stellenbosch : Stellenbosch University |
| publisherStr | Stellenbosch : Stellenbosch University |
| record_format | dspace |
| source_str | SUNScholar — Stellenbosch University Repository |
| spelling | oai:scholar.sun.ac.za:10019.1/127372 Reconstruction of 3D models of the femur from planar x-rays for surgical planning Oosthuizen, Jodri Van der Merwe, Johan Stellenbosch University. Faculty of Engineering. Dept. of Mechanical and Mechatronic Engineering. Three-dimensional imaging in medicine -- Statistical methods Bones -- Imaging -- Statistical methods Diagnostic imaging UCTD Thesis (MEng)--Stellenbosch University, 2023. ENGLISH SUMMARY: Slipped Capital Femoral Epiphysis (SCFE) is a common hip disorder of adolescents in which the femoral head is posteroinferiorly displaced from the neck of the femur. Several treatments have been described in the literature to treat the deformity, where an intertrochanteric osteotomy is usually performed to correct the slip of more severe cases. Pre-operative planning is a critical requirement for the success of any orthopaedic procedure. Traditionally the planning relies on 2D radiographs and although this method has proven to be sufficient for a subset of procedures, factors such as overlapping anatomical structures, lack of 3D information and variability of film quality could lead to inaccuracies of the preoperative estimations. Thus, to improve the accuracy and reliability in the treatment of SCFE, 3D models of bony structures have been used more extensively for pre-operative surgical planning. CT scans are the preferred 3D image modality but still have some limitations. These limitations include subjecting the patient to a high radiation dose, being expensive and time-consuming. 3D Reconstruction of a patient-specific model from 2D images has been considered to be a suitable alternative. Therefore this study aimed to develop a 2D/3D reconstruction algorithm that reconstructs a 3D patient-specific model of the proximal femur from a 2D x-ray image. The algorithm compensates for the lack of information provided by the x-ray images by using prior knowledge of the shape. The prior knowledge can be obtained through the use of a Statistical Shape Model (SSM) which aims to capture all possible variations within a population. Thus, as a part of the study, an SSM was constructed using a training dataset containing 27 proximal femurs. The constructed shape model along with a digitally reconstructed x-ray was used to reconstruct a 3D model from 2D images. The resulting patient-specific model was obtained by projecting the contour of the SSM to the corresponding x-ray and measuring the point-to-point distance between the projected contour and the contour extracted from the x-ray. The accuracy of the resulting model was determined by comparing the reconstructed model to a ground truth model segmented from image data. The models were generated with an average reconstruction error of 2.13 mm, a Hausdorff distance of 5.49 mm and a Dice coefficient of 0.88. An average difference of 3.05° was measured between the NSAs of the ground truth and reconstructed meshes. The average NSA difference indicated that the reconstruction algorithm could be adapted to include SCFE data. Although the results corresponded well with those of similar studies available in the literature, the accuracy of the model needs to be improved before being applied to clinical applications. AFRIKAANSE OPSOMMING: Geglyde Proksimale Femorale Epifises (GPFE) is a algemene heupversteuring van adolessente waar die femorale kop in ‘n postero-inferior wyse van die femorale nek verplaas word. Verskeie metodes word in literatuur beskryf vir die behandeling van die versteuring, waar ‘n intertrokanteriese osteotomie gewoontlik gebruik word om die gly van die femorale kop in meer ernstige gevalle te herstel. Vir enige ortopediese prosedure om ‘n sukses te wees, is die chirurgiese beplanning voor die prosedure ‘n noodsaaklike vereiste. Beplanning deur middel van 2D radiografiee word gesien as die konvensionele metode vir chirurgiese beplanning. Alhoewel hierdie metode bewys is as voldoende vir ‘n versameling van prosedures, kan faktore soos anatomiese strukture wat oorvleuel, tekort aan 3D informasie en die wisselvalligheid van filmkwaliteit die akuraatheid van preoperatiewe benaderings beinvloed. Dus, om die betroubaarheid en akkuraatheid in die behandeling van GPFE te verbeter word drie dimensionele been struktuur modelle meer gereeld gebruik vir die beplanning voor die prosedure. CT skandering is die 3D beeld modaliteit wat verkies word, maar dit het steeds beperkings. Die beperkings sluit ‘n hoe bestraling dosis, hoe kostes en tydrowendheid in. 3D rekonstruksie van ‘n patient-spesifieke model vanaf ‘n 2D beeld word as ‘n gepaste alternatief oorweeg. Daarom was die onwikkeling van ‘n 2D/3D rekonstruksie algoritme wat ‘n patient-spesifieke model van die proksimale femur vanaf ‘n 2D beeld rekonstrueer, ‘n mikpunt van die studie. So ‘n algoritme kompenseer vir die gebrek aan inligting wat deur die x-straal verskaf word deur vorige kennis van die vorm te gebruik. Die vorige kennis kan vekry word met die gebruik van ‘n Statistiese Vormmodel (SV) wat die moontlike variasies in ‘n populasie vas vang. Dus was daar, as deel van die studie, ‘n SV geskep vanuit ‘n opleidingsdatastel wat 27 proksimale femurs bevat. Die SV wat geskep is saam met ‘n digitale gerekonstrueerde x-straal was gebruik om ‘n 3D model van ‘n 2D beeld te rekonstrueer. Die resulterende patient-spesifieke model was verkry deur die kontoer van die SV te projekteer na die ooreenstemmende x-straal en die punt-tot-punt afstand tussen die geprojekteerde kontoer en die kontoer wat uit die x-straal onttrek is te meet. Die akkuraatheid van die resulterende model was bepaal deur die gerekonstrueerde model met die ware data, wat vanuit die CT data gesegmenteer is, te vergelyk. Die modelle was gegenereer met ‘n gemiddelde rekonstruksie fout van 2.13 mm, ‘n Hausdorff afstand van 5.49 mm en ‘n Dice se koeffisient van 0.88. ‘n Gemiddelde verskil van 3.05° was tussen die nek-skag hoek van die rekonstrueerde en die ware model gemeet. Hierdie verskil dui aan dat die algoritme angepas kan word om data van GPFE pasiente in te sluit. Alhoewel die resultate in die studie verkry ooreenstem met data wat in literatuur beskikbaar is, moet die akkuraatheid van die model verbeter word voordat dit oorweeg kan word vir kliniese toepassings. Masters 2023-03-02T10:08:57Z 2023-05-18T07:18:44Z 2023-03-02T10:08:57Z 2023-05-18T07:18:44Z 2023-03 Thesis http://hdl.handle.net/10019.1/127372 en_ZA Stellenbosch University xvi, 87 pages : illustrations application/pdf Stellenbosch : Stellenbosch University |
| spellingShingle | Three-dimensional imaging in medicine -- Statistical methods Bones -- Imaging -- Statistical methods Diagnostic imaging UCTD Oosthuizen, Jodri Reconstruction of 3D models of the femur from planar x-rays for surgical planning |
| title | Reconstruction of 3D models of the femur from planar x-rays for surgical planning |
| title_full | Reconstruction of 3D models of the femur from planar x-rays for surgical planning |
| title_fullStr | Reconstruction of 3D models of the femur from planar x-rays for surgical planning |
| title_full_unstemmed | Reconstruction of 3D models of the femur from planar x-rays for surgical planning |
| title_short | Reconstruction of 3D models of the femur from planar x-rays for surgical planning |
| title_sort | reconstruction of 3d models of the femur from planar x rays for surgical planning |
| topic | Three-dimensional imaging in medicine -- Statistical methods Bones -- Imaging -- Statistical methods Diagnostic imaging UCTD |
| url | http://hdl.handle.net/10019.1/127372 |
| work_keys_str_mv | AT oosthuizenjodri reconstructionof3dmodelsofthefemurfromplanarxraysforsurgicalplanning |