Schedule

LECTURE 1: 12:00 - 12:30

Speaker: Erik Vrij

Subject: "Automated phenotypic screening and analysis platforms for stem cell-based embryo models"

Abstract: 
Recent advancements in stem cell technology have led to the development of sophisticated embryo models, providing a platform to study embryogenesis and implantation biology in vitro. In our lab, we have developed a partial mouse embryo model that replicates the co-developmental dynamics between the epiblast (precursor to the fetus) and the extraembryonic endoderm (precursor to the yolk sac) typical of the peri-implantation stage. This model recapitulates crucial developmental events such as cell sorting, cell polarization, exit from pluripotency, formation of the pro-amniotic cavity, and transcriptomic channeling, all coordinated by reciprocal inductions between these co-developing tissues.

To enhance observation and analysis, we employed a thin polymer film-based thermoformed microwell platform that enables automated fluorescence imaging-based feature detection. This setup, combined with machine learning, allows us to classify morphogenetic variations and assess the heterogeneity within the embryo models. Additionally, we conducted screenings with soluble compound libraries to delineate the molecular pathways involved in early embryonic progression and to evaluate the toxicological effects of known and potential toxins. 
Our findings highlight the potential of integrating advanced embryo models with advanced culture and analysis platforms, paving the way for large-scale screening assays that can impact embryology, embryo-toxicity assessments, drug development, and reproductive medicine.

LECTURE 2: 12:30 - 13:00

Speaker: Pierpaolo Fucile

Subject: "Development of an AI-based real-time control for advanced robotic 3D printing"

Abstract: 
Most of the current additive manufacturing (AM) technologies, also known as 3D printing, that are used for bio-fabrication and tissue engineering (TE) work in a layer-by-layer approach, limiting the architectural complexity of the constructs that can be produced. For instance, it is very difficult to obtain scaffolds with anisotropic fiber orientation in 3D through extrusion-based AM technologies. A motion system with higher degrees of freedom (DOF) could be used for volumetric extrusion-based 3D printing.

To fully exploit the potential offered by a multi-DOF 3D printing system it will be necessary to develop an ad hoc procedure. RAVEN (Robot-Assisted Volumetric ExtrusioN) is a new robotic AM platform for tissue engineering applications with the ability to do volumetric printing that we have developed within our group. 

When printing supportless filaments in air, precise control on fibers diameter and positioning is crucial. For this reason, we are developing an AI-based algorithm for automatic image recognition towards real-time monitoring and correction of fibers deposition in the 3D space. Our model is a pre-trained deep CNN (Convolutional Neural Network) based model, which allows for robust results even with small datasets. Our dataset is a collection of labelled pictures of 3D printed filaments. 

The trained model is the starting point of a Computer Vision (CV) algorithm that recognizes 3D printed filaments in real-time and evaluates their thickness. In case of over- or under-extrusion, real-time 3D printing commands are uploaded to compensate for that.

Further developments will allow for precise control and implementation of the algorithm in the robotic system control loop, thus achieving an advanced system for small-scale bioprinting.