Twin-to-Twin Transfusion Syndrome (TTTS) affects 10-15% of monochorionic twin linebreak pregnancies and is associated with high perinatal mortality and morbidity. In this case, the vascular connections shared by the two fetuses, anastomoses, are not distributed equally leading to an imbalance in blood exchange. Causes and development of this syndrome are still unknown, however, treatment by photo-coagulation of anastomoses appears to be an effective method for the moment. The presence of few cases, the treatment of this pathology by only a few specialized centers, and the difficulties related to the uterine environment make the procedure very complex. A silicone phantom and a video-game style simulator appear to be the only training systems for medical personnel. However, such systems appear to be costly and/or unrealistic. To overcome these difficulties, this thesis proposes the development of a simulator that is more realistic than the current methods highlighted. For the development of such a simulator, the object modelling software Blender, the game engine Unity for the realization of the simulator itself and a neural network, specifically a Generative Adversarial Network, for the creation of the textures with the vascular map, were used. Particular focus for the realization of the elements was placed on amniotic fluid, particulate matter, placenta, laser and fetus. A questionnaire was submitted to a number of users for the evaluation of the simulator. The very realism of the environment was rated more highly by users than the video game style. The realistic nature of the vessels, the turbidity of the fluid, and and the setup of the fetoscope were highly valued. However, some additional features such as bleeding, more pronounced color differentiation, a greater effect in vessel ablation, and less fluent movements of the fetoscope were highlighted as features to be improved.
La Sindrome da Trasfusione Fetale da Gemello a Gemello (TTTS) colpisce il 10-15% delle gravidanze gemellari monocoriali ed è associata a un'elevata mortalità e morbilità perinatale. In questo caso, le connessioni vascolari condivise dai due feti, le anastomosi, non sono distribuite in modo equilibrato, determinando uno squilibrio negli scambi di sangue. Le cause e lo sviluppo di questa sindrome sono ancora sconosciute, tuttavia il trattamento mediante fotocoagulazione sembra essere per il momento il metodo più efficace. La presenza di pochi casi, il trattamento di questa patologia da parte di pochi centri specializzati e le difficoltà legate all'ambiente uterino, rendono la procedura molto complessa. Un fantoccio in silicone e un simulatore in stile videogioco, sembrano essere ad oggi gli unici sistemi di formazione per il personale medico. Tuttavia, tali sistemi sembrano essere costosi e/o poco realistici. Per superare le difficoltà elencate, questo lavoro di tesi propone lo sviluppo di un simulatore che sia più realistico degli attuali metodi evidenziati. Per lo sviluppo di tale simulatore sono stati utilizzati il software di modellazione di oggetti Blender, il motore di gioco Unity per la realizzazione del simulatore stesso e una rete neurale, nello specifico una Rete Generativa Adversaria, per la creazione delle texture con la mappa vascolare. Particolare attenzione per la realizzazione degli elementi è stata posta su liquido amniotico, particolato, placenta, laser e feto. Un questionario è stato sottoposto ad una serie di utenti per la valutazione del simulatore. Il realismo dell'ambiente è stato valutato dagli utenti in misura maggiore rispetto al simulatore in stile videogioco. In particolare, la natura realistica dei vasi, la torbidità del liquido e l'illuminazione sono stati molto apprezzati. Tuttavia, alcune caratteristiche aggiuntive come il sanguinamento, una differenziazione dei colori più marcata, un effetto maggiore nell'ablazione dei vasi e movimenti meno fluidi del fetoscopio sono stati evidenziati come caratteristiche da migliorare.
Efesto: a realistic fetoscopy simulator for better surgical training
Biagioli, Jessica
2021/2022
Abstract
Twin-to-Twin Transfusion Syndrome (TTTS) affects 10-15% of monochorionic twin linebreak pregnancies and is associated with high perinatal mortality and morbidity. In this case, the vascular connections shared by the two fetuses, anastomoses, are not distributed equally leading to an imbalance in blood exchange. Causes and development of this syndrome are still unknown, however, treatment by photo-coagulation of anastomoses appears to be an effective method for the moment. The presence of few cases, the treatment of this pathology by only a few specialized centers, and the difficulties related to the uterine environment make the procedure very complex. A silicone phantom and a video-game style simulator appear to be the only training systems for medical personnel. However, such systems appear to be costly and/or unrealistic. To overcome these difficulties, this thesis proposes the development of a simulator that is more realistic than the current methods highlighted. For the development of such a simulator, the object modelling software Blender, the game engine Unity for the realization of the simulator itself and a neural network, specifically a Generative Adversarial Network, for the creation of the textures with the vascular map, were used. Particular focus for the realization of the elements was placed on amniotic fluid, particulate matter, placenta, laser and fetus. A questionnaire was submitted to a number of users for the evaluation of the simulator. The very realism of the environment was rated more highly by users than the video game style. The realistic nature of the vessels, the turbidity of the fluid, and and the setup of the fetoscope were highly valued. However, some additional features such as bleeding, more pronounced color differentiation, a greater effect in vessel ablation, and less fluent movements of the fetoscope were highlighted as features to be improved.File | Dimensione | Formato | |
---|---|---|---|
2022_10_BIAGIOLI.pdf
accessibile in internet solo dagli utenti autorizzati
Dimensione
64.74 MB
Formato
Adobe PDF
|
64.74 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/10589/195733