One of the main causes of childhood morbidity and infant mortality in developed countries, within the first year of life, are congenital malformations and they actually have an overall global prevalence of 2-3% among all births and can be involved in more or less severe disability conditions. Congenital heart defects (CHD) are the most common congenital malformations in prenatal maturation, approximately represents one third of serious infant congenital diseases. In Europe, CHD prevalence is around 8/1000 births and CHD defines generic structural heart defects and malformations that appear at birth and that could compromise the organ walls, the valves and the blood vessels, arteries and veins close to the heart. There are various elements and factors that influence heart’s correct maturation and that cause CHD. Symptoms are very difficult to be recognized, as they can occur either in the first years of life or in adulthood. A prompt prenatal diagnosis of CHD allows to detect a preventive treatment with superior benefits, addressing precautions to follow and to respect before birth in order to control fetus cardiac underdevelopment adequately. An accurate diagnosis can reduce infant morbidity and mortality drastically and it becomes clear that cardiac development knowledge and monitoring of cardiac growth phases can be a fundamental aid to investigate heart malformations morphogenesis, which is a pivotal requirement to obtain optimal medical care. Monitoring cardiac growth progress is feasible through live imaging at the cellular level thanks to an innovative improvement of optical microscopy technology and the discovery of a wide spectrum of fluorescent proteins suitable as cells markers, which allow to observe cell movements and variations of their expression in time. The importance to track and analyze cell motion is focused on their dynamic behavior in order to obtain valid parameters to characterize physiological or pathological maturation processes. The point of this work is to suggest a motion estimation algorithm to reproduce trajectories of a culture of mouse’s embryonal cells from information collected through a sequence of 3D microscopy images. This issue consists of a temporal elastic registration model implementation and its following validation of its pre-clinic applicability. The technique of elastic registration is considered more suitable for this sequence of images and above all according to the movement of the cell culture during its growth. The proposed algorithm is based on a parametric deformation model which uses B-spline features to maintain controlled smoothness. As these are heavy images, a multiresolution optimization strategy was chosen to increase processing\computation speed and robustness of the method. The first step in validation consisted in creating a set of artificial images realized to reproduce the cardiac cell motion and the characteristics of the real microscopy images in a simpler way. The method was validated on this synthetic sequence to detect its accuracy. The final endorsement was provided by testing the algorithm on the sequence of real images. The image registration has been applied following an accurate pre-processing and the final tracking results were compared to those achieved from the manual cell tracking by an expert biologist. Further control was obtained with the aid of an open source software for automatic cells tracking. The obtained results validate the improved algorithm defining it a solid tool for cell tracking and for object motion estimation generally, in 3D image sequences. In particular in pre-processed 3D fluorescence microscopy images, this approach has shown good results in cell tracking.
DINAMICA DELLO SVILUPPO DEL CUORE MEDIANTE IMMAGINI 3D DA MICORSCOPIO A FLUORESCENZA. Le patologie congenite (le malformazioni) rappresentano una delle principali cause di mortalità infantile in tutto il mondo, mostrando una prevalenza complessiva del 2-3% sulle nascite globali. L’alta varietà di malformazioni e l’eventuale coinvolgimento di qualsiasi tipo di organo, comporta la possibilità di ricadere in condizioni più o meno gravi di invalidità, fino alla compromissione delle funzioni vitali del corpo. Le cardiopatie congenite sono le malformazioni congenite più comuni durante il periodo prenatale e rappresentano circa un terzo delle malattie congenite infantili gravi. In Europa, la probabilità di cardiopatie congenite è di circa 8/1000 nascite e definisce tutti i difetti cardiaci strutturali che appaiono alla nascita e che potrebbero compromettere le pareti cardiache, le valvole cardiache e i vasi sanguigni presenti in prossimità del cuore. Sono molteplici i fattori che influenzano la corretta crescita di questo organo e che possano perciò causare cardiopatie congenite. I sintomi sono estremamente difficili da riconoscere ed una pronta diagnosi prenatale consente di determinare un trattamento preventivo con benefici superiori, può ridurre drasticamente la mortalità infantile e la conoscenza dello sviluppo cardiaco e il monitoraggio delle fasi di crescita possono essere un aiuto fondamentale per indagare la morfogenesi delle malformazioni, che è un requisito fondamentale per ottenere cure mediche ottimali. Il monitoraggio della crescita cardiaca è possibile attraverso l’acquisizione di immagini dal vivo a livello cellulare grazie a un innovativo miglioramento della tecnologia di microscopia ottica e grazie alla scoperta di un ampio spettro di proteine fluorescenti utilizzate come marker cellulari, che consentono di osservare i movimenti cellulari e le variazioni della loro espressione nel tempo. L'importanza di tracciare e analizzare il movimento cellulare è focalizzata sul loro comportamento dinamico al fine di ottenere parametri validi a caratterizzare i processi di maturazione fisiologici o patologici. Lo scopo di questo lavoro è di suggerire un algoritmo di stima del movimento cellulare per riprodurre le traiettorie percorse da una coltura di cellule embrionali di topo da informazioni raccolte attraverso una sequenza di immagini 3D di microscopia. Questo studio presenta l'implementazione di un modello di registrazione elastico temporale e la successiva convalida della sua applicabilità pre-clinica. Per questa sequenza di immagini è stata ritenuta più adatta una tecnica di registrazione elastica, che permettesse di seguire al meglio il movimento della coltura cellulare durante la sua crescita. L'algoritmo proposto si basa su un modello di deformazione parametrico che utilizza le caratteristiche B-splines per mantenere l’uniformità dell’immagine controllata. Poiché si tratta di dati pesanti, è stata scelta una strategia di ottimizzazione a multirisoluzione che permettesse l’aumento della velocità di calcolo e la robustezza del metodo. Il primo passo di convalida ha previsto la creazione di un insieme di immagini artificiali che riproducessero il movimento delle cellule cardiache e le caratteristiche delle immagini reali di microscopia in modo semplice. La verifica finale è stata fornita testando l'algoritmo sulla sequenza di immagini reali. La registrazione è stata applicata seguendo un'accurata pre-elaborazione e i risultati finali del tracciamento sono stati confrontati con quelli ottenuti dal tracking manuale fornito da un biologo esperto. I risultati ottenuti convalidano l'algoritmo adattato definendolo uno strumento solido per il tracciamento cellulare e di oggetti in generale, nelle sequenze di immagini 3D. In particolare, sono stati mostrati buoni risultati nel tracciamento cellulare, ottenuto con immagini 3D da microscopia a fluorescenza.
Dynamics of heart development using 3D fluorescence microscope images.
DAL GAL, ELISABETTA
2016/2017
Abstract
One of the main causes of childhood morbidity and infant mortality in developed countries, within the first year of life, are congenital malformations and they actually have an overall global prevalence of 2-3% among all births and can be involved in more or less severe disability conditions. Congenital heart defects (CHD) are the most common congenital malformations in prenatal maturation, approximately represents one third of serious infant congenital diseases. In Europe, CHD prevalence is around 8/1000 births and CHD defines generic structural heart defects and malformations that appear at birth and that could compromise the organ walls, the valves and the blood vessels, arteries and veins close to the heart. There are various elements and factors that influence heart’s correct maturation and that cause CHD. Symptoms are very difficult to be recognized, as they can occur either in the first years of life or in adulthood. A prompt prenatal diagnosis of CHD allows to detect a preventive treatment with superior benefits, addressing precautions to follow and to respect before birth in order to control fetus cardiac underdevelopment adequately. An accurate diagnosis can reduce infant morbidity and mortality drastically and it becomes clear that cardiac development knowledge and monitoring of cardiac growth phases can be a fundamental aid to investigate heart malformations morphogenesis, which is a pivotal requirement to obtain optimal medical care. Monitoring cardiac growth progress is feasible through live imaging at the cellular level thanks to an innovative improvement of optical microscopy technology and the discovery of a wide spectrum of fluorescent proteins suitable as cells markers, which allow to observe cell movements and variations of their expression in time. The importance to track and analyze cell motion is focused on their dynamic behavior in order to obtain valid parameters to characterize physiological or pathological maturation processes. The point of this work is to suggest a motion estimation algorithm to reproduce trajectories of a culture of mouse’s embryonal cells from information collected through a sequence of 3D microscopy images. This issue consists of a temporal elastic registration model implementation and its following validation of its pre-clinic applicability. The technique of elastic registration is considered more suitable for this sequence of images and above all according to the movement of the cell culture during its growth. The proposed algorithm is based on a parametric deformation model which uses B-spline features to maintain controlled smoothness. As these are heavy images, a multiresolution optimization strategy was chosen to increase processing\computation speed and robustness of the method. The first step in validation consisted in creating a set of artificial images realized to reproduce the cardiac cell motion and the characteristics of the real microscopy images in a simpler way. The method was validated on this synthetic sequence to detect its accuracy. The final endorsement was provided by testing the algorithm on the sequence of real images. The image registration has been applied following an accurate pre-processing and the final tracking results were compared to those achieved from the manual cell tracking by an expert biologist. Further control was obtained with the aid of an open source software for automatic cells tracking. The obtained results validate the improved algorithm defining it a solid tool for cell tracking and for object motion estimation generally, in 3D image sequences. In particular in pre-processed 3D fluorescence microscopy images, this approach has shown good results in cell tracking.È consentito all'utente scaricare e condividere i documenti disponibili a testo pieno in UNITESI UNIPV nel rispetto della licenza Creative Commons del tipo CC BY NC ND.
Per maggiori informazioni e per verifiche sull'eventuale disponibilità del file scrivere a: unitesi@unipv.it.
https://hdl.handle.net/20.500.14239/19378