Date of Award

Fall 12-2021

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Electrical & Computer Engineering

Program/Concentration

Biomedical Engineering

Committee Director

Michel Audette

Committee Member

Jiang Li

Committee Member

Stacie Ringleb

Committee Member

Jérôme Schmid

Abstract

Scoliosis, an abnormal curvature of the spine, is traditionally corrected with bracing treatments or by a highly invasive posterior spinal fusion (PSF) operation. These correction strategies are constrained by current imaging modalities, which fail to elucidate the soft tissue anatomy that is known to play a critical role in spinal stiffness and overall structure. Osteoligamentous segmentations of the spinal column offer a foundation for downstream finite element (FE) studies seeking to optimize bracing treatments or determine ideal surgical approaches.

This thesis presents methods for automatically and semi-automatically segmenting vertebrae and surrounding soft tissues of the spinal column using X-ray computed tomography (CT) volumes of asymptomatic to severely symptomatic scoliotic patients. The automatic process further develops the osteoligamentous segmentations into FE meshes of individual anatomy, such as the ligaments and intervertebral discs. The FE meshes and an originally developed osteoligamentous lumbar spine FE mesh are analyzed in a FE solver. Additionally, a novel validation technique is presented, which uses synthetic CT volumes to confirm the surmised soft tissue positions of the patient-specific segmentations. Extremely competitive dice similarity coefficients and submillimeter average Hausdorff distances demonstrate vertebrae and intervertebral disc segmentation accuracy of the automatic methodology tested on CT and multimodal datasets. The osteoligamentous FE meshes display desirable characteristics within the FE solver environment and the FE lumbar spine is expectedly stiff compared to counterpart FE meshes. With the novel validation approach, the specificity of soft tissue localization is determined to be nearly perfect, and the presented osteoligamentous segmentation method is shown to outperform current studies when testing on severe cases.

The orthopedic surgical community will greatly benefit from the knowledge of soft tissue anatomy that has been localized in CT volumes. Patient-specific, osteoligamentous FE meshes can provide the basis for biomechanical simulations seeking to correct scoliosis through bracing, minimally invasive operations, or patient-specific, surgical procedures. The proposed validation method, which employs synthetic CT, is achieved fully in silico and may be generically applied to evaluate soft tissue anatomy segmented using future approaches or can be adapted to embody pathology for additional neural network training that ultimately improves neural network segmentation abilities.

Rights

In Copyright. URI: http://rightsstatements.org/vocab/InC/1.0/ This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s).

DOI

10.25777/3bfb-5e73

ISBN

9798762197397

ORCID

0000-0002-0043-6360

Download

Share

COinS