Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Automated image segmentation and implant generation for large defects of the mandible.
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_ | 1867613954538209280 |
|---|---|
| access_status_str | Open Access |
| author | Marx, Dorian |
| author2 | Van der Merwe, Johan |
| author_browse | Marx, Dorian Van der Merwe, Johan |
| author_facet | Van der Merwe, Johan Marx, Dorian |
| author_sort | Marx, Dorian |
| collection | Thesis |
| dc_rights_str_mv | Stellenbosch University |
| description | Thesis (MEng)--Stellenbosch University, 2023. |
| format | Thesis |
| id | oai:scholar.sun.ac.za:10019.1/129096 |
| institution | Stellenbosch University (South Africa) |
| language | en_ZA |
| last_indexed | 2026-06-10T12:44:21.236Z |
| 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:10019.1/129096 Automated image segmentation and implant generation for large defects of the mandible. Marx, Dorian Van der Merwe, Johan Stellenbosch University. Faculty of Engineering. Dept. of Mechanical and Mechatronic Engineering. Image segmentation -- Automation. Mandible -- Abnormalities Orthopedic implants Mandibular reconstruction Human anatomy -- Models Three-dimensional printing Thesis (MEng)--Stellenbosch University, 2023. ENGLISH ABSTRACT: Mandibular reconstruction is a complex surgical procedure performed for segmental pathological ablation and discontinuous traumatic fractures of the mandible. The success of mandibular reconstruction is paramount for overall positive prognosis by limitation of physiological, psychological and psychosocial impairment. Mandibular reconstruction conducted by open reduction internal fixation crucially depends on mandibular implants for mechanical stability, without which a positive recovery cannot be achieved. Patient specific mandibular implants demonstrate superior clinical performance but are currently constrained by labour-intensive workflows and are hindered by long lead times. Incentivised by these factors, this study proposes and implements an automated patient specific implant design algorithm. Implants generated by the algorithm outperform both standard implants and patient specific implants found in literature in terms of accuracy of fit and design time, both of which contribute to successful reconstruction and recovery of the patient. The automated patient specific implant design algorithm automates image segmentation, virtual geometric reconstruction and procedural implant generation by taking advantage of the structure of a statistical point distribution model for a fully integrated pipeline. The automated patient specific implant design algorithm was tested for simulated large defects of the mandible, which included defects containing displaced and non-displaced mandibular segments. Defects were simulated directly on otherwise unedited computed tomography medical image volumes to demonstrate full automation of the implant design process directly from medical images. Virtual mandibular reconstruction via automated image segmentation showed a root mean square error ranging from 0.91±0.12mmto 1.94±0.35mm for mandibles that were between 0% and 68% proportionally defective. The automatic generation of patient specific implants was tested for a central defect classification in which 32% of the mandible was defective in 10 unseen test patients. The implants automatically generated by the algorithm show a root mean square fit accuracy of 0.55±0.20mm, outperforming the best other implant accuracy found in literature of 1.19mm. The algorithm additionally demonstrated a decrease in total implant design time from observations in industry of almost three orders of magnitude (reduced to 473.61 ± 80.34 s from 469 800 s). The algorithm proposed in this study not only demonstrates a promising alternative to various labour-intensive manual design processes, it demonstrates that multidisciplinary knowledge can be integrated into an automated pipeline which greatly reduces design time and increases the accuracy of fit. AFRIKAANSE OPSOMMING: Mandibulêre herkonstruksie is ’n komplekse chirurgiese prosedure wat vir segmentele patalogiese ablasie of diskontinue frakture van die mandibel uitgevoer word. Suksesvolle mandibulêre herkonstruksie is uiters belangrik vir positiewe prognose deur fisiologiese, sielkundige en psigososiale waardedaling te beperk. Mandibulêre herkonstruksie, wat deur oop reduksie interne fiksasie plaasvind, is afhanklik van mandibulêre implantate vir meganiese stabiliteit, waarsonder ’n positiewe prognose nie bereikbaar is nie. Pasiëntspesifieke mandibulêre implantate toon voortreflike kliniese werkverrigting, maar word tans deur arbeidsintensiewe werkvloeie beperk en word deur lang aanlooptye belemmer. Aangespoor deur hierdie faktore, word ’n geoutomatiseerde pasiëntspesifieke implantaatontwerpalgoritme in hierdie studie voorgestel en geïmplementeer. Dit word bewys in hierdie studie dat implantate, wat deur hierdie algoritme gegenereer word, beter as beide standaard en pasiëntspesifieke implantate presteer. Dié prestasie is geëvalueer op die basis van ontwerptyd en akkuraatheid van passing, metrieke wat albei tot die sukses mandibulêre herkonstruksie bydra. Die outomatiese pasiëntspesifieke implantaatontwerpalgoritme outomatiseer beeldsegmentering, virtuele geometriese herkonstruksie en prosedurele implantaat generering deur gebruik te maak van die struktuur van ’n statistiese puntverspreidingsmodel vir ’n volledig geïntegreerde pyplyn. Die geoutomatiseerde pasiëntspesifieke implantaatontwerpalgoritme is getoets op gesimuleerde groot defekte van die mandibel, wat verplaasde en onverplaasde mandibulêre segmente bevat. Defekte is direk op andersins ongeredigeerde rekenaar-tomografie mediese beeldvolumes gesimuleer om volle outomatisering van die implantaatontwerpproses direk vanaf mediese beelde te demonstreer. Virtuele mandibulêre herkonstruksie deur outomatiese beeldsegmentering het ’n vierkantswortel van gemiddelde kwadraat afwyking tussen 0.91 ± 0.12mm en 1.94 ± 0.35mm getoon vir mandibels was tussen 0% en 68% proporsioneel defektief was. Die outomatiese generering van pasiëntspesifieke implantate was getoets vir ’n sentrale defek klassifikasie waarin 32% van die mandible defektief was in 10 ongesiende pasiënte. Die implantate, wat outomaties deur die algoritme gegenereer is, toon ’n vierkantswortel van gemiddelde kwadraat pasakkuraatheid van 0.55 ± 0.20mm. Dié pasakkuraatheid was beter as die beste pasiëntspesifieke implantaat pasakkuraatheid waarneming in literatuur van 1.19mm. Die algoritme het bykomend ’n afname in totale implantaatontwerptyd van amper drie grootte-ordes gedemonstreer (verminder tot 473.61 ± 80.34 s van 469 800 s). Die algoritme wat in hierdie studie voorgestel word demonstreer nie net ’n belowende alternatief vir verskeie arbeidsintensiewe handontwerpprosesse nie, dit demonstreer dat multidissiplinêre kennis in ’n geoutomatiseerde pyplyn geïntegreer kan word wat ontwerptyd aansienlik verminder en die akkuraatheid van passing verhoog. Masters 2023-11-22T17:08:27Z 2024-01-08T22:15:41Z 2023-11-22T17:08:27Z 2024-01-08T22:15:41Z 2023-12 Thesis https://scholar.sun.ac.za/handle/10019.1/129096 en_ZA Stellenbosch University xxiii, 132 pages : illustrations application/pdf Stellenbosch : Stellenbosch University |
| spellingShingle | Image segmentation -- Automation. Mandible -- Abnormalities Orthopedic implants Mandibular reconstruction Human anatomy -- Models Three-dimensional printing Marx, Dorian Automated image segmentation and implant generation for large defects of the mandible. |
| title | Automated image segmentation and implant generation for large defects of the mandible. |
| title_full | Automated image segmentation and implant generation for large defects of the mandible. |
| title_fullStr | Automated image segmentation and implant generation for large defects of the mandible. |
| title_full_unstemmed | Automated image segmentation and implant generation for large defects of the mandible. |
| title_short | Automated image segmentation and implant generation for large defects of the mandible. |
| title_sort | automated image segmentation and implant generation for large defects of the mandible |
| topic | Image segmentation -- Automation. Mandible -- Abnormalities Orthopedic implants Mandibular reconstruction Human anatomy -- Models Three-dimensional printing |
| url | https://scholar.sun.ac.za/handle/10019.1/129096 |
| work_keys_str_mv | AT marxdorian automatedimagesegmentationandimplantgenerationforlargedefectsofthemandible |