The ambitious objective of the project is to develop a conceptually novel micro-rheometer for rapid and automatic characterization of biological fluids supported by a Machine/Deep Learning (ML/DL) approach. Rheological characterization of liquid properties usually requires tens of milliliters of sample, thus being unsuitable for rare or precious materials as those involved in biological studies. The quest for miniaturized, embedded, and reliable tools to characterize such properties is underway since at least a decade. The intelligent micro-rheometer we intend to develop will be based on various innovative tools, for the first time applied in the field of rheology: a new ElectroHydroDynamic (EHD) technology based on PyroelectricEffect (PE) in ferroelectric substrate able to probe the liquid through electric fields in non-contact mode; a double optical module combining computer vision operating in 2D with a digital holographic system for full 3D tracking of dynamic behaviors of liquids under the action of the EHD pressure; an intelligent strategy based on neural networks able to learn morphology and dynamic evolution of the liquids, and thus to classify/characterize them in terms of their rheological parameters.
The project aims at:
-Direct or indirect correlation of rheological parameters to body fluid alterations due to diseases; a compact and automatic micro-rheometer can constitute itself a diagnostic tool by screening human/animal body fluids (i.e. saliva, blood, urine, lacrimal, etc.);
-Fast measurement of liquid properties to design new tools relevant for the expected future development of biotechnologies basedon intensive use and manipulation of matter in liquid phase (i.e. biological inks for advanced inkjet printers, advanced microarray, appropriate handling/processing of analytes in biomedical instruments/devices, most appropriate coupling of analytes to biomedicalinstruments/sensors and wearables devices and sensors, lab on chip devices);
-ML/DL data-driven development of an integrated automatic platform for biofluid characterization/classification to achieve full control for manipulating liquids by means of PE pressure; ML/DL will support the experimentas in a “into-the-loop” version to guide the parameters’ choice.
The key-technologies are:
-Accurate classification/characterization of fluids avoiding classical and cumbersome measurements -Automatic system for classification/characterization of fluids -Ability to evaluate transport properties with very small liquid volumes -Non-contact mode to avoid any possible contamination -Flexibility to characterize fluids in wide ranges of parameters -Compactness and portability -Low cost -Label free modality
TIPER is based on a multi-disciplinary platform to meet all the above challenges thanks to the combination of model-based design of the PE multi-configurable technology with 2D-3D computer vision and fully integrated in the Artificial Intelligence paradigm.
Posts to the project