Finite Element (FE) models allow the simulation of the biomechanical behavior of the foot, predicting internal stresses and strain. Models proposed in literature exploited the standard Magnetic Resonance (MR) and Computed Tomography (CT) technology, but in last decades more advanced acquisition imaging techniques were developed. The image quality of the 3 T MR is improved than the one of the 1.5 T MR, allowing the detection of smaller details, but it still misses the weight-bearing condition that leads to the correct and real morphology of the foot. This limitation is overcome by the Cone-Beam CT (CBCT) technology. This thesis aims at evaluating the extent to which how more advanced medical imaging data, whose segmentation is carried out with the same software, may enhance an established FE model of the human foot developed previously and proposed in the literature. The foot of a healthy subject was acquired with 3T MR and CBCT technology, analyzing both the load and the unloaded condition with the latter, respectively in single leg up-right posture, and in supine. FE models are created with the 3D objects, i.e. bones, cartilage and soft tissues, obtained from the segmentation of each of these acquisitions, to which boundary conditions taken from the gait analysis are applied. Two simulations of the models obtained from each of the three image acquisitions are run, one for the initial contact phase of the gait cycle and one for the midstance. Plantar pressure map is then compared with corresponding data from gait analysis to validate the model. The results of the pressure maps in the initial contact show no substantial differences among the three and the corresponding experimental data by visual inspection. Although, the former with the respect to the latter underestimate both the peak and the mean pressure, providing a slightly larger contact surface area. Greater differences occur in the midstance since pressure maps of foot-to-floor contacts from the simulations not concurring with the corresponding map from gait analysis measurements. In the under load CBCT simulation the midfoot results in contact with the plate, not the case of corresponding experimental data and of the model predictions from unloading CBCT data. This could be justified by the plantar surface being already deformed during the acquisition. Both the predicted peak and the mean pressure are underestimated from the simulations with respect to the corresponding value obtained experimentally, leading to a larger contact surface. Overall, the best experiment-to-model match was observed when using anatomical 3D objects from unloading CBCT. Basically, in the initial contact the three FE models behave somehow as observed experimentally, while in the midstance the mismatch with respect to the experimental data is justified by the different foot condition during the three acquisitions. In the 3T MR scan the foot is constrained by a support to maintain the same ankle position, while the subject keeps a self-controlled neutral position of the ankle in the CBCT supine acquisition and a weight-bearing posture in the under load one. The underestimation of the predicted pressure data with respect to the experimental values probably can be accounted on the application of the boundary conditions obtained in dynamic to 3D objects from images acquired in static condition. The introduction of muscle in the model could solve this issue. Moreover, the static physiological load of the subject in the CBCT acquisition in up-right posture probably might not reproduce the load obtained from the Ground Reaction Force in dynamic. In addition, a standard lateral support of the foot could be used for the acquisitions with different devices in supine position to keep the foot constrained, so fixing the ankle angle with no constraint on the plantar surface. Finally, future investigations could enlarge the cohort of study in evaluating these advanced imaging techniques in foot FE model.
I modelli ad elementi finiti (FE) simulano il comportamento biomeccanico del piede, consentendo la valutazione delle deformazioni e degli stress interni. Quelli proposti in letteratura si basano su immagini mediche ottenute da risonanza magnetica (MR) o tomografia computerizzata (CT), ma negli ultimi decenni sono state sviluppate tecniche di acquisizione più avanzate. La risonanza magnetica 3 T, rispetto a quella 1.5 T, permette l’individuazione di dettagli anatomici più piccoli ma non è ancora possibile valutare il soggetto nella sua condizione di carico fisiologica. Tale limite è superato con la Cone-Beam Computed Tomography (CBCT). Lo scopo di questa tesi è valutare quanto le immagini mediche ottenute da tecniche di acquisizione più recenti e avanzate possano migliorare un modello FE precedentemente sviluppato e proposto in letteratura. Il piede di un soggetto sano è stato acquisito con la MR 3T e la CBCT, dove con quest’ultima sono state considerate entrambe le condizioni di carico e scarico, rispettivamente in posizione eretta di appoggio mono-podalico e supina. I modelli FE sono stati elaborati da oggetti 3D, i.e. le ossa, la cartilagine e i tessuti molli, ottenuti dalla segmentazione delle immagini di ciascuna delle tre acquisizioni, ai quali sono stati imposte le condizioni al contorno ricavate dall’analisi del cammino. Per ogni modello sono state lanciate due simulazioni, nella fase di contatto iniziale del ciclo del passo e in quella intermedia (midstance). Per la validazione del modello, è stata confrontata la mappa delle pressioni plantari con i corrispondenti dati ottenuti durante l’analisi del cammino. Da un punto di vista visivo, i risultati delle simulazioni nella fase di contatto iniziale non mostrano sostanziali differenze tra le tre simulazioni e i dati sperimentali. Nonostante ciò, le prime sottostimano il picco e la pressione media rispetto ai valori sperimentali, con il risultato di una superficie di contatto tra il piede e la pedana leggermente maggiore. Le differenze più evidenti sono presenti nella midstance. Infatti, nella simulazione dalla CBCT in carico, il mesopiede risulta in contatto con la pedana, diversamente da quanto accade sperimentalmente e dalla predizione del modello dalla CBCT acquisita in supino. Ciò è giustificabile dal fatto che la superficie plantare già durante l’acquisizione è deformata dal carico fisiologico. Inoltre, i valori predittivi del picco e della pressione media sono sottostimati dalle simulazioni e la superficie di contatto piede-pedana è più estesa. Nel complesso, si può dedurre che l’oggetto 3D che meglio simula i dati sperimentali è ottenuto dall’acquisizione in posizione supina con la CBCT. In sostanza, i modelli FE si avvicinano a quanto osservato sperimentalmente nella fase di contatto iniziale, mentre nella midstance il discostamento dei dati simulati rispetto a quelli sperimentali è dovuto alla diversa condizione del piede durante le tre acquisizioni. La sottostima dei valori simulati rispetto a quelli sperimentali può essere dovuta all’applicazione di dati ottenuti in dinamica, quali le condizioni al contorno, ad oggetti 3D ricavati da immagini acquisite in statica. L’introduzione nel modello dei muscoli potrebbe risolvere tale problema. Inoltre, il carico fisiologico del soggetto imposto mantenendo la posizione eretta nell’acquisizione statica con la CBCT probabilmente non riproduce il carico ricavato dalla forza di reazione vincolare nella condizione dinamica del cammino. Si potrebbero quindi misurare le pressioni plantari durante l’acquisizione in carico della CBCT. In aggiunta, un supporto standard laterale potrebbe essere utilizzato in futuro per le acquisizioni in posizione supina con diverse tecniche. Futuri studi potrebbero poi estendere la valutazione di queste recenti tecniche applicate a modelli FE del piede a una coorte di studio più numerosa.
Assessment of finite element models of the human foot from most recent advanced medical imaging: weight-bearing CT and 3T magnetic resonance
BONATTI, GIULIA
2021/2022
Abstract
Finite Element (FE) models allow the simulation of the biomechanical behavior of the foot, predicting internal stresses and strain. Models proposed in literature exploited the standard Magnetic Resonance (MR) and Computed Tomography (CT) technology, but in last decades more advanced acquisition imaging techniques were developed. The image quality of the 3 T MR is improved than the one of the 1.5 T MR, allowing the detection of smaller details, but it still misses the weight-bearing condition that leads to the correct and real morphology of the foot. This limitation is overcome by the Cone-Beam CT (CBCT) technology. This thesis aims at evaluating the extent to which how more advanced medical imaging data, whose segmentation is carried out with the same software, may enhance an established FE model of the human foot developed previously and proposed in the literature. The foot of a healthy subject was acquired with 3T MR and CBCT technology, analyzing both the load and the unloaded condition with the latter, respectively in single leg up-right posture, and in supine. FE models are created with the 3D objects, i.e. bones, cartilage and soft tissues, obtained from the segmentation of each of these acquisitions, to which boundary conditions taken from the gait analysis are applied. Two simulations of the models obtained from each of the three image acquisitions are run, one for the initial contact phase of the gait cycle and one for the midstance. Plantar pressure map is then compared with corresponding data from gait analysis to validate the model. The results of the pressure maps in the initial contact show no substantial differences among the three and the corresponding experimental data by visual inspection. Although, the former with the respect to the latter underestimate both the peak and the mean pressure, providing a slightly larger contact surface area. Greater differences occur in the midstance since pressure maps of foot-to-floor contacts from the simulations not concurring with the corresponding map from gait analysis measurements. In the under load CBCT simulation the midfoot results in contact with the plate, not the case of corresponding experimental data and of the model predictions from unloading CBCT data. This could be justified by the plantar surface being already deformed during the acquisition. Both the predicted peak and the mean pressure are underestimated from the simulations with respect to the corresponding value obtained experimentally, leading to a larger contact surface. Overall, the best experiment-to-model match was observed when using anatomical 3D objects from unloading CBCT. Basically, in the initial contact the three FE models behave somehow as observed experimentally, while in the midstance the mismatch with respect to the experimental data is justified by the different foot condition during the three acquisitions. In the 3T MR scan the foot is constrained by a support to maintain the same ankle position, while the subject keeps a self-controlled neutral position of the ankle in the CBCT supine acquisition and a weight-bearing posture in the under load one. The underestimation of the predicted pressure data with respect to the experimental values probably can be accounted on the application of the boundary conditions obtained in dynamic to 3D objects from images acquired in static condition. The introduction of muscle in the model could solve this issue. Moreover, the static physiological load of the subject in the CBCT acquisition in up-right posture probably might not reproduce the load obtained from the Ground Reaction Force in dynamic. In addition, a standard lateral support of the foot could be used for the acquisitions with different devices in supine position to keep the foot constrained, so fixing the ankle angle with no constraint on the plantar surface. Finally, future investigations could enlarge the cohort of study in evaluating these advanced imaging techniques in foot FE model.File | Dimensione | Formato | |
---|---|---|---|
Bonatti_Giulia.pdf
embargo fino al 10/04/2025
Dimensione
8.24 MB
Formato
Adobe PDF
|
8.24 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/20.500.12608/29069