Back to search

Schulich School of Engineering

Super-Resolution of Cone Beam Computed Tomography: Applications in Bone Analysis for Rheumatoid Arthritis

Abstract

dc:description.abstract

Rheumatoid arthritis (RA) is a chronic inflammatory disease characterized by bone erosions, which serve as critical indicators of disease progression and structural damage. High-resolution peripheral quantitative computed tomography (HR-pQCT) is the reference standard for imaging these erosions but is constrained by limited anatomical coverage and long acquisition times. In contrast, cone beam computed tomography (CBCT) offers faster imaging with broader coverage but insufficient resolution for detailed erosion assessment. To address this, a curriculum learning-based CycleGAN framework was developed to super-resolve CBCT images to HR-pQCT quality, enabling more accurate extraction of structural information. To support automated erosion detection and quantification, an atlas of healthy hand bones was constructed. This atlas served as an anatomical prior in a deep learning pipeline, AG-UNet (atlas-guided nnUNet). By providing spatial context from healthy anatomy, the atlas acted as an anatomical prior to improve segmentation accuracy. The super-resolution framework was trained in a staged manner, beginning with paired synthetic data and progressing to unpaired real CBCT and HR-pQCT scans. This approach yielded improved image quality, significantly enhanced similarity metrics, and reduced trabecular bone errors across stages (p < 0.001). A blinded rheumatologist classified 40% of super-resolved images as HR-pQCT, confirming perceptual quality and diagnostic relevance as all RA erosions remained manually identifiable in the enhanced images. AG-UNet was trained on manually annotated erosions in the 2nd and 3rd metacarpophalangeal joints. On HR-pQCT, the model achieved strong agreement with manual volumes (Pearson r = 0.946, MAE = 1.90 mm³) and 92.3% detection sensitivity. When applied to super-resolved CBCT, sensitivity was 71.4%, improving for larger erosions. Volume estimates from super-resolved CBCT were well correlated with HR-pQCT (r = 0.890), and longitudinal tracking of 21 erosions over one year showed moderate agreement in change scores (r = 0.613) with no significant bias. These findings demonstrate that super-resolution enhances the amount of meaningful structural information that can be extracted from lower-resolution CBCT scans, beyond what is possible at native resolution and expanding the utility of musculoskeletal CBCT.

Degree

thesis:*
Name thesis:degree_name
Master of Science (MSc)
Discipline thesis:degree_discipline
Engineering – Biomedical
Grantor
Schulich School of Engineering
Year dc:date.issued
2025

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Al-Khoury, Yousif Costantien
Advisors dc:contributor.advisor
  • Manske, Sarah L.
  • MacDonald, M. Ethan
Committee members dc:contributor.committeemember
  • Barber, Claire E. H.
  • Gobbi, David G.
  • Wilms, Matthias

Subjects

dc:subject × 9

Rights

dc:rights
Statement dc:rights
  • Unless otherwise indicated, this material is protected by copyright and has been made available with authorization from the copyright owner. You may use this material in any way that is permitted by the Copyright Act or through licensing that has been assigned to the document. For uses that are not allowable under copyright legislation or licensing, you are required to seek permission.
Language dc:language.iso
en

Identifiers

dc:identifier.*
Dc Identifier Citation
Al-Khoury, Y. (2025). Super-resolution of cone beam computed tomography: applications in bone analysis for rheumatoid arthritis (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca.

Chain of custody

source
Harvested from
University of Calgary
Base URL
ucalgary.scholaris.ca/server/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Al-Khoury, Yousif Costantien. Super-Resolution of Cone Beam Computed Tomography: Applications in Bone Analysis for Rheumatoid Arthritis. Schulich School of Engineering, 2025. https://hdl.handle.net/1880/122467