Epithelial tissues are widely present across all organs and organisms, serving as barriers and boundaries between means. Yet, their 3D organization differs widely. Epithelia can organize in single or multiple layers, and cell shapes range from cuboidal to highly elon gated shapes, whose long axis runs either along the apico-basal axis (columnar) or parallel to the epithelial surface (squamous). Those different organizations must reflect differences in the biophysical cell parameters, but their type and magnitude have remained elusive. In this study, we leverage deep learning methods for the segmentation of 3D images of epithe lial samples acquired at the light sheet microscope. Then, we present EpiStats, a library for the analysis of morphological tissue features. By applying EpiStats to the collected epithelial samples, we provide a complete experimental characterization of the tissues, highlighting common traits and discrepancies between them. Finally, we employ Simu Cell3D, a 3D cell-based simulation framework, to perform an exploratory investigation of the mechanical and biophysical cell parameters. This preliminary research highlights some aspects of the current state of SimuCell3D that need to be improved in future it erations to provide more conclusive results about epithelial tissues’ mechanical properties.
I tessuti epiteliali sono ampiamente diffusi in ogni organo e organismo. Nonostante in ogni parte compiano la stessa funzione di barriera tra mezzi, la loro organizzazione tridi mensionale è estremamente varia. In particular, gli epiteli possono organizzarsi in uno o più strati, e le forme delle cellule vanno da cuboidali a forme maggiormente allun gate, come quelle colonnari, il cui asse lungo si estende lungo l’asse apico-basale e quelle squamose, disposte parallelamente alla superficie epiteliale. Le diverse conformazioni di tali tessuti epiteliali si riflettono in differenze nei parametri biofisici delle cellule. Tuttavia, la specificità e l’entità di tali discrepanze rimangono sfuggenti. In questo studio, utilizzi amo metodi di apprendimento profondo per la segmentazione di immagini 3D di campioni epiteliali acquisite al microscopio light sheet. Successivamente, presentiamo EpiStats, una libreria per l’analisi delle caratteristiche morfologiche dei tessuti. Applicando EpiStats ai campioni epiteliali raccolti, produciamo una caratterizzazione sperimentale dei vari tes suti, evidenziando tratti comuni e difference salienti. Infine, utilizziamo SimuCell3D, un software per la simulazione di modelli basati su singola cellular in 3D. Lo scopo è quello di condurre un’indagine esplorativa dei parametri meccanici e biofisici delle cellule. Questa ricerca preliminare mette in risalto alcuni aspetti dello stato attuale di SimuCell3D che richiedono miglioramenti per le future versioni dello studio, al fine di ottenere risultati più conclusivi sulle proprietà meccaniche dei tessuti epiteliali.
Deep learning-driven image analysis of epithelial tissues structure and organization
CARRARA, FEDERICO
2022/2023
Abstract
Epithelial tissues are widely present across all organs and organisms, serving as barriers and boundaries between means. Yet, their 3D organization differs widely. Epithelia can organize in single or multiple layers, and cell shapes range from cuboidal to highly elon gated shapes, whose long axis runs either along the apico-basal axis (columnar) or parallel to the epithelial surface (squamous). Those different organizations must reflect differences in the biophysical cell parameters, but their type and magnitude have remained elusive. In this study, we leverage deep learning methods for the segmentation of 3D images of epithe lial samples acquired at the light sheet microscope. Then, we present EpiStats, a library for the analysis of morphological tissue features. By applying EpiStats to the collected epithelial samples, we provide a complete experimental characterization of the tissues, highlighting common traits and discrepancies between them. Finally, we employ Simu Cell3D, a 3D cell-based simulation framework, to perform an exploratory investigation of the mechanical and biophysical cell parameters. This preliminary research highlights some aspects of the current state of SimuCell3D that need to be improved in future it erations to provide more conclusive results about epithelial tissues’ mechanical properties.File | Dimensione | Formato | |
---|---|---|---|
Federico_Carrara___Master_s_Thesis.pdf
accessibile in internet per tutti
Descrizione: Thesis Document
Dimensione
19.86 MB
Formato
Adobe PDF
|
19.86 MB | Adobe PDF | Visualizza/Apri |
Executive_Summary___Federico_Carrara.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Executive Summary
Dimensione
6.27 MB
Formato
Adobe PDF
|
6.27 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/10589/214017