One third of cancer patients are affected by spinal metastases, which compromise bone integrity. However, the clinical criterion used to determine vertebral instability and fracture risk, the Spinal Instability Neoplastic Score (SINS), does not take bone biomechanics into account. This thesis is part of a project developed at the Bioengineering and Computational Laboratory at Rizzoli Orthopaedic Institute, which aims to complement this clinical score with a quantitative approach by evaluating vertebral instability through a validated subject-specific finite element model (SSFEM) derived from CT scans. This thesis aims to create computational models of intact and lesioned vertebrae and validate them against experimental results. Moreover, osteoporotic vertebrae with cortical defects in diverse locations have been evaluated and compared against healthy vertebrae to assess the validity of the protocol in pathological cases. Four vertebrae from three different donors have been studied, a T9 from #5594, a T10 from #5603 and a T8 and a T10 from #5598. The latter donor shows signs of osteoporosis, consequently its vertebrae are characterized by cortical defects that required careful handling. The geometry of each vertebra has been segmented from the CT scan of the donor’s spine through the cortical bone mapping (CBM) algorithm which estimates cortical thickness and density. For each vertebra, four models were created, the first one representing the intact condition, the others aimed to simulate the presence of metastatic lesions, by first performing a hole on the cranial endplate along the axial direction with an increasing diameter (10mm and 20mm), and then further drilling one hole transversally through the vertebral cortex, with a diameter of approximately 15mm, in the anterior (for both the T10) or lateral (for the T9 and T8) wall of the vertebra. To permit the identification of experimental constraints, holes position and measurement reference system, high resolution HR-pQCT scans of the vertebra with a hole were used. For each model, a 10-node tetrahedral mesh was created. Material properties were mapped from CT scan according to density to determine the longitudinal elastic modulus, and different transverse isotropy constants were applied to cortical and trabecular bone compartments. Mechanical simulations were performed in Ansys, reproducing the loading conditions of the experimental setup, which applies a homogeneous pressure on the caudal endplate of the vertebra, while the cranial endplate is submerged and constrained in a low melting metal. Superficial displacements and strains in the vertical direction of the antero-lateral aspects were acquired through a Digital Image Correlation (DIC) system, while the caudal endplate deflection was measured through an LVDT. DIC measurements correspond to point-wise displacement and strain values on the surface of the vertebra, spatially registered on the model afterwards. SSFEM results have been compared to experimental measurements. SSFEM predictions of caudal endplate displacement are in good agreement with experimental data and capture a proportional increase between endplate compliance and lesion size, with errors decreasing from 13.52% (20mm hole models) to 7.59% (cortical hole models), an overall correlation of R^2~0.96 and a slope of 0.83. SSFEMs and DIC longitudinal displacements are correlated with R^2~0.79 and a slope of 0.79, meaning that there is a slight underestimation of the computational models. This finding is confirmed by longitudinal strain errors, whose median relative percentage value, averaged among all different configurations was 14%. In conclusion, the protocol developed to create SSFEM of healthy thoracic vertebrae can be extended to osteoporotic vertebrae with cortical defects. However, to fully assess the quality of the model validation, a larger number of samples is required.
Un terzo dei pazienti oncologici è affetto da metastasi spinali, che compromettono l’integrità ossea. Tuttavia, il criterio clinico utilizzato per determinare l’instabilità vertebrale e il rischio di frattura, lo Spinal Instability Neoplastic Score (SINS), non tiene conto della biomeccanica ossea. Questa tesi fa parte di un progetto sviluppato presso il Laboratorio di Bioingegneria Computazionale dell’Istituto Ortopedico Rizzoli, che mira a integrare questo punteggio clinico con un approccio quantitativo valutando l’instabilità vertebrale attraverso un modello agli elementi finiti validato e soggetto-specifico (SSFEM) derivato da scansioni TC. Questa tesi ha lo scopo di creare modelli computazionali di vertebre intatte e lesionate e validarli rispetto a risultati sperimentali. Inoltre, vertebre osteoporotiche con difetti corticali in diverse posizioni sono state valutate e confrontate con vertebre sane per verificare la validità del protocollo in casi patologici. Sono state studiate quattro vertebre provenienti da tre diversi donatori: una T9 (#5594), una T10 (#5603), una T8 e una T10 (#5598). Quest’ultimo donatore presenta segni di osteoporosi, di conseguenza le sue vertebre sono caratterizzate da difetti corticali che hanno richiesto particolare attenzione. La geometria di ciascuna vertebra è stata segmentata dalla TC della colonna attraverso l’algoritmo di Cortical Bone Mapping (CBM), che stima spessore e densità corticale. Per ogni vertebra sono stati creati quattro modelli: uno intatto e tre con lesioni simulate, realizzando un foro sull’endplate craniale lungo la direzione assiale con diametro crescente (10 e 20 mm), e successivamente un foro trasversale di 15mm attraverso la corticale, nella parete anteriore (per le T10) o laterale (per T9 e T8). Per identificare vincoli sperimentali, posizione dei fori e sistema di riferimento, sono state utilizzate scansioni HR-pQCT ad alta risoluzione delle vertebre con foro. Per ciascun modello è stata creata una mesh tetraedrica a 10 nodi. Le proprietà dei materiali sono state mappate dalla scansione TC in funzione della densità per determinare il modulo elastico longitudinale, applicando diverse costanti di isotropia trasversale ai compartimenti ossei corticale e trabecolare. Le simulazioni meccaniche, eseguite in Ansys, riproducono il setup sperimentale: pressione omogenea sull’endplate caudale, mentre l’endplate craniale è immerso e vincolato in un metallo a basso fondente. Gli spostamenti superficiali e le deformazioni verticali degli aspetti antero-laterali sono stati acquisiti tramite Digital Image Correlation (DIC), mentre la deflessione dell’endplate caudale è stata misurata con LVDT. Le misure DIC sono valori puntuali di spostamento e deformazione sulla superficie, poi registrati sul modello. I risultati degli SSFEM sono stati confrontati con le misure sperimentali. Le previsioni dello spostamento dell’endplate caudale sono in buon accordo con i dati sperimentali e mostrano un aumento proporzionale tra compliance dell’endplate e dimensione della lesione, con errori dal 13.52% (foro 20 mm) al 7.59% (foro corticale), correlazione complessiva di ² ~ 0.96 e una pendenza di 0.83. Gli spostamenti longitudinali SSFEM e DIC sono correlati con ² ~ 0.79 e pendenza di 0.79, indicando una lieve sottostima da parte dei modelli computazionali. Ciò è confermato dagli errori nelle deformazioni longitudinali, il cui valore percentuale relativo mediano, mediato su tutte le diverse configurazioni, è pari al 14%. In conclusione, il protocollo per creare SSFEM di vertebre toraciche sane può essere esteso a vertebre osteoporotiche con difetti corticali. Tuttavia, per una validazione completa del modello è necessario un numero maggiore di campioni.
Validation of CT-Based Finite Element Models of Human Thoracic Vertebrae Using Digital Image Correlation Data: Effects of Simulated Lesions With and Without Cortical Bone Involvement
TOMIAZZO, CAMILLA
2025/2026
Abstract
One third of cancer patients are affected by spinal metastases, which compromise bone integrity. However, the clinical criterion used to determine vertebral instability and fracture risk, the Spinal Instability Neoplastic Score (SINS), does not take bone biomechanics into account. This thesis is part of a project developed at the Bioengineering and Computational Laboratory at Rizzoli Orthopaedic Institute, which aims to complement this clinical score with a quantitative approach by evaluating vertebral instability through a validated subject-specific finite element model (SSFEM) derived from CT scans. This thesis aims to create computational models of intact and lesioned vertebrae and validate them against experimental results. Moreover, osteoporotic vertebrae with cortical defects in diverse locations have been evaluated and compared against healthy vertebrae to assess the validity of the protocol in pathological cases. Four vertebrae from three different donors have been studied, a T9 from #5594, a T10 from #5603 and a T8 and a T10 from #5598. The latter donor shows signs of osteoporosis, consequently its vertebrae are characterized by cortical defects that required careful handling. The geometry of each vertebra has been segmented from the CT scan of the donor’s spine through the cortical bone mapping (CBM) algorithm which estimates cortical thickness and density. For each vertebra, four models were created, the first one representing the intact condition, the others aimed to simulate the presence of metastatic lesions, by first performing a hole on the cranial endplate along the axial direction with an increasing diameter (10mm and 20mm), and then further drilling one hole transversally through the vertebral cortex, with a diameter of approximately 15mm, in the anterior (for both the T10) or lateral (for the T9 and T8) wall of the vertebra. To permit the identification of experimental constraints, holes position and measurement reference system, high resolution HR-pQCT scans of the vertebra with a hole were used. For each model, a 10-node tetrahedral mesh was created. Material properties were mapped from CT scan according to density to determine the longitudinal elastic modulus, and different transverse isotropy constants were applied to cortical and trabecular bone compartments. Mechanical simulations were performed in Ansys, reproducing the loading conditions of the experimental setup, which applies a homogeneous pressure on the caudal endplate of the vertebra, while the cranial endplate is submerged and constrained in a low melting metal. Superficial displacements and strains in the vertical direction of the antero-lateral aspects were acquired through a Digital Image Correlation (DIC) system, while the caudal endplate deflection was measured through an LVDT. DIC measurements correspond to point-wise displacement and strain values on the surface of the vertebra, spatially registered on the model afterwards. SSFEM results have been compared to experimental measurements. SSFEM predictions of caudal endplate displacement are in good agreement with experimental data and capture a proportional increase between endplate compliance and lesion size, with errors decreasing from 13.52% (20mm hole models) to 7.59% (cortical hole models), an overall correlation of R^2~0.96 and a slope of 0.83. SSFEMs and DIC longitudinal displacements are correlated with R^2~0.79 and a slope of 0.79, meaning that there is a slight underestimation of the computational models. This finding is confirmed by longitudinal strain errors, whose median relative percentage value, averaged among all different configurations was 14%. In conclusion, the protocol developed to create SSFEM of healthy thoracic vertebrae can be extended to osteoporotic vertebrae with cortical defects. However, to fully assess the quality of the model validation, a larger number of samples is required.| File | Dimensione | Formato | |
|---|---|---|---|
|
Tomiazzo_Camilla.pdf
embargo fino al 13/04/2027
Dimensione
27.04 MB
Formato
Adobe PDF
|
27.04 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/106839