Avascular necrosis of the femoral head is a pathological condition caused by a reduction in blood flow to a portion of bone, leading to tissue degeneration and loss of functionality. The incidence of this condition is increasing, particularly among young adults. Among the emerging strategies for its treatment, the use of scaffolds represents a promising approach, as these structures are designed to mimic the properties of bone and support its regeneration. This study presents an innovative method for designing scaffolds with anisotropic, specifically orthotropic, properties to better replicate the characteristics of physiological bone tissue. To this end, a dataset of scaffolds was created based on Triply-Periodic Minimal Surface (TPMS) geometries, both in their classical form and in hybrid configurations, and subsequently characterised from both mechanical and morphological perspectives. To simulate necrotic tissue, a portion of healthy femoral head was extracted from a μ-CT scan. This segment was then divided into four sub-volumes, each of which was mechanically and morphologically characterised. The data obtained were used as target for selecting scaffolds to approximate the properties of the analysed bone sub-volumes. This selection was carried out using two approaches: one based on user-defined threshold settings and the other employing a generative adversarial network (GAN) model of Artificial Intelligence. The selected scaffolds were then integrated into complete devices through gradual transitions between geometric parameters, ensuring structural continuity. Two transition methods were tested: a linear variation and a sigmoidal variation. The results demonstrate the potential of these models for scaffolds customisation and optimisation according to specific clinical requirements. In particular, special attention is given to the innovation introduced by GANs models in Artificial Intelligence, which offer new possibilities in the design and optimisation of scaffolds, enhancing their structural and functional properties and opening new perspectives in Tissue Engineering and Regenerative Medicine.
La necrosi avascolare della testa femorale è una patologia causata dalla riduzione del flusso sanguigno in una porzione di osso, con conseguente degenerazione tessutale e perdita di funzionalità. L’incidenza di questa condizione è in aumento, soprattutto tra i giovani adulti. Tra le strategie emergenti per il suo trattamento, l’utilizzo di scaffolds rappresenta un’opzione promettente, poiché queste strutture sono progettate per mimare le proprietà dell’osso e supportarne la rigenerazione. In questo studio è stato sviluppato un metodo innovativo per la progettazione di scaffolds con caratteristiche anisotrope, in particolare ortotrope, per replicare al meglio le proprietà dell’osso fisiologico. A tal fine, è stato creato un dataset di scaffolds basati su geometrie Tripli-Periodiche a Minima Superficie (TPMS), sia nella loro forma classica sia in configurazioni ibride, e successivamente caratterizzati dal punto di vista meccanico e morfologico. Per simulare il tessuto necrotico, è stata considerata una porzione di tessuto osseo estratta da una μ-CT di testa femorale sana, suddivisa poi in quattro sotto-volumi, anch’essi caratterizzati meccanicamente e morfologicamente. I dati ottenuti sono stati utilizzati come riferimento per la selezione degli scaffolds al fine di approssimare le proprietà del tessuto analizzato. Tale selezione è avvenuta attraverso due approcci: uno basato sull’impostazione di soglie personalizzate dall’utente e l’altro mediante un modello generativo a reti avversarie (GAN) di Intelligenza Artificiale. Gli scaffolds selezionati sono stati sottoposti a un processo di unione per creare dispositivi completi, attraverso transizioni graduali tra i parametri geometrici, garantendo continuità strutturale. Due modalità di transizione sono state testate: una variazione lineare e una a sigmoide. I risultati ottenuti dimostrano il potenziale di questi modelli per la personalizzazione degli scaffolds e la loro ottimizzazione in base alle specifiche esigenze cliniche. In particolare, si guarda con attenzione all’innovazione introdotta dai modelli GAN di Intelligenza Artificiale, che offrono nuove possibilità nella progettazione e nell’ottimizzazione degli scaffolds, migliorandone le caratteristiche strutturali e funzionali, aprendo nuove prospettive nell’Ingegneria dei Tessuti e nella Medicina Rigenerativa.
Modelli di Intelligenza Artificiale per scaffolds personalizzati: una soluzione per il trattamento di necrosi avascolare dell'osso
Nesci, Giulia
2023/2024
Abstract
Avascular necrosis of the femoral head is a pathological condition caused by a reduction in blood flow to a portion of bone, leading to tissue degeneration and loss of functionality. The incidence of this condition is increasing, particularly among young adults. Among the emerging strategies for its treatment, the use of scaffolds represents a promising approach, as these structures are designed to mimic the properties of bone and support its regeneration. This study presents an innovative method for designing scaffolds with anisotropic, specifically orthotropic, properties to better replicate the characteristics of physiological bone tissue. To this end, a dataset of scaffolds was created based on Triply-Periodic Minimal Surface (TPMS) geometries, both in their classical form and in hybrid configurations, and subsequently characterised from both mechanical and morphological perspectives. To simulate necrotic tissue, a portion of healthy femoral head was extracted from a μ-CT scan. This segment was then divided into four sub-volumes, each of which was mechanically and morphologically characterised. The data obtained were used as target for selecting scaffolds to approximate the properties of the analysed bone sub-volumes. This selection was carried out using two approaches: one based on user-defined threshold settings and the other employing a generative adversarial network (GAN) model of Artificial Intelligence. The selected scaffolds were then integrated into complete devices through gradual transitions between geometric parameters, ensuring structural continuity. Two transition methods were tested: a linear variation and a sigmoidal variation. The results demonstrate the potential of these models for scaffolds customisation and optimisation according to specific clinical requirements. In particular, special attention is given to the innovation introduced by GANs models in Artificial Intelligence, which offer new possibilities in the design and optimisation of scaffolds, enhancing their structural and functional properties and opening new perspectives in Tissue Engineering and Regenerative Medicine.File | Dimensione | Formato | |
---|---|---|---|
Nesci_Thesis_Finale.pdf
accessibile in internet per tutti
Descrizione: Testo della Tesi di Laurea Magistrale
Dimensione
58.11 MB
Formato
Adobe PDF
|
58.11 MB | Adobe PDF | Visualizza/Apri |
Nesci_Executive_Summary_Finale.pdf
accessibile in internet per tutti
Descrizione: Testo dell'Executive Summary
Dimensione
4.29 MB
Formato
Adobe PDF
|
4.29 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/234516