Periodic Reporting for period 1 - REVEAL (Neuronal microscopy for cell behavioural examination and manipulation)
Okres sprawozdawczy: 2021-01-01 do 2022-06-30
The REVEAL project is developing an innovative and disruptive photonic tool to meet the objectives of the European H2020 call: develop the next generation of disruptive biophotonics methods and devices as research tools to understand the cellular origin of diseases
The expected impact of REVEAL will contribute to major progress for health and society. In particular, the project will identify the mechanisms driving disease initiation will allow to design prevention strategies for minimizing occurrence. The bio-teams of the project, Policlinico Milano, ENS-Lyon and University of Munich, are focusing on HepatoCellular Carcinoma (HCC) – the more predominant form of liver cancer. The causes of this disease are varied, but also include viral infections, diet, and alcohol abuse. Obesity and fatty liver disease often progress to cirrhosis and cancer of the liver as well.
To date, spatial omics, the combination of cell imaging and omics measurements allows mapping the protein landscape and/or gene expression within a human tissue. This is becoming the most precise methods to decipher the hallmarks of cancer, tuning its aggressivity and resistance to treatment. However, this approach does not describe the dynamic scenario involved in the origin of cancer diseases. In the Reveal project, we propose to couple live cell microscopy and transcriptomics analysis in order to obtain evidence of the dynamic processes at the origin of cancer diseases. To this aim, the REVEAL project introduces the concept of ‘neuronal microscopy’, a microscopy technique relying heavily on neural networks able to perceive, interpret, conjecture, infer, anticipate and act. In a proof of principle, a ‘cell picking neuronal microscope’ will detect and let collect different subtype of liver cells. In particular cells that are due to become tumoral will be collected and transcriptomics analyses will be conducted at the University of Munich. Hence the analysis is intended to be ahead of tumour manifestation to understand the cellular origin of the disease.
The expected impact of REVEAL will contribute to major progress for health and society. In particular, the project will identify the mechanisms driving disease initiation will allow to design prevention strategies for minimizing occurrence. The bio-teams of the project, Policlinico Milano, ENS-Lyon and University of Munich, are focusing on HepatoCellular Carcinoma (HCC) – the more predominant form of liver cancer. The causes of this disease are varied, but also include viral infections, diet, and alcohol abuse. Obesity and fatty liver disease often progress to cirrhosis and cancer of the liver as well.
To date, spatial omics, the combination of cell imaging and omics measurements allows mapping the protein landscape and/or gene expression within a human tissue. This is becoming the most precise methods to decipher the hallmarks of cancer, tuning its aggressivity and resistance to treatment. However, this approach does not describe the dynamic scenario involved in the origin of cancer diseases. In the Reveal project, we propose to couple live cell microscopy and transcriptomics analysis in order to obtain evidence of the dynamic processes at the origin of cancer diseases. To this aim, the REVEAL project introduces the concept of ‘neuronal microscopy’, a microscopy technique relying heavily on neural networks able to perceive, interpret, conjecture, infer, anticipate and act. In a proof of principle, a ‘cell picking neuronal microscope’ will detect and let collect different subtype of liver cells. In particular cells that are due to become tumoral will be collected and transcriptomics analyses will be conducted at the University of Munich. Hence the analysis is intended to be ahead of tumour manifestation to understand the cellular origin of the disease.
Our work is structured on three important objectives: 1) develop a neuronal microscope, (2) develop a neuronal cell picking microscope and (3) discover biological signatures at the origin of liver cancer disease.
(1) The development of neuronal microscopy is part of the ongoing evolutions of cell microscopy. In the first year of the Reveal projects we have contributed to three important evolutionary leaps:
a- computational microscopy, i.e. optics are replaced by algorithms
b- neuronal microscopy, i.e.– algorithms are replaced by neural networks
c- an all-neural framework, i.e. physical and biological models can be built upon neural networks, from 3D sample reconstruction to the of study cell-cell interactions within a large population.
One important achievement at M18 is the development of two versions of the ‘silico visual cortex’, being the interconnection of several neural networks used for e.g. image formation, cell analysis, cell fate prediction and decision making. Overall, we have demonstrated that several functionalities are now at reach with convolutional neural networks, i.e. image formation, cell quantification and cell tracking. Importantly the framework of deep learning allows (i) fast computation and (ii) continuous performance improvements. The ability to conduct live imaging and analysis in quasi real-time analysis (>10.000 cells in few tens of seconds, see above), let us envision the microscope able of decision and action that will come next.
(2) Develop neuronal cell picking microscopy
A concept of a new instrument, called a neuronal cell picking microscope, has been designed and is now under fabrication. It combines a bi-modal microscopy setup with an automated cell picking system. It has been developed in a close collaboration between the different partners of the project, i.e. ALS, CEA, Iprasense and ENS-Lyon. To achieve single cell picking, the live-imaging acquisition and the cell picking module must match to within one micrometer, therefore a linear path measurement system, has been developed. The integration of the ‘silico visual cortex’ to this new automation will deliver the microscope able of cell fate prediction and decision making that we envision.
(3) Discover biological signatures at the origin of a specific disease
Hepatocellular carcinoma (HCC) is the most common liver cancer and the second most frequent cause of cancer-related deaths worldwide. The most recent clinical research has pointed to extensive biological heterogeneity at the level of cell morphologies, genetic fingerprints and responses to drug treatment in liver cancer patients. Our stated objective in project REVEAL is to validate the capacity of neuronal microscopy to detect and identify the biological heterogeneity that take root in patients very early on- during early stages of dysmetabolism in the liver – far before patients experience steatohepatitis, fibrosis, liver adenomas or cancer.
Under the aegis of this project, we have now characterized many different types of patient derived cells that grow on a 2D matrix in laboratory culture as well as patient derived normal or tumoral 3D organoids- mini livers- that reflect either the health or the cancerous state in how they propogate in laboratory conditions. Moreover, we have developed state of the art murine 2D cellular and 3D mini livers that reflect nearly every stage of liver disease- from early stage metabolic disorders to late stage cancer. As a first step, these laboratory models have been used to develop both the 2D and the 3D neuronal microscope, and immediately we have realized the very first stated goal of the project- that our laboratory models do indeed capture the heterogeneity seen in the clinical patient-derived samples. We are actively interrogating the molecular signatures of these samples with the aim to provide a state of the art “omics” signature to the images acquired by the neuronal microscope. In the coming months, this critical step will be achieved.
(1) The development of neuronal microscopy is part of the ongoing evolutions of cell microscopy. In the first year of the Reveal projects we have contributed to three important evolutionary leaps:
a- computational microscopy, i.e. optics are replaced by algorithms
b- neuronal microscopy, i.e.– algorithms are replaced by neural networks
c- an all-neural framework, i.e. physical and biological models can be built upon neural networks, from 3D sample reconstruction to the of study cell-cell interactions within a large population.
One important achievement at M18 is the development of two versions of the ‘silico visual cortex’, being the interconnection of several neural networks used for e.g. image formation, cell analysis, cell fate prediction and decision making. Overall, we have demonstrated that several functionalities are now at reach with convolutional neural networks, i.e. image formation, cell quantification and cell tracking. Importantly the framework of deep learning allows (i) fast computation and (ii) continuous performance improvements. The ability to conduct live imaging and analysis in quasi real-time analysis (>10.000 cells in few tens of seconds, see above), let us envision the microscope able of decision and action that will come next.
(2) Develop neuronal cell picking microscopy
A concept of a new instrument, called a neuronal cell picking microscope, has been designed and is now under fabrication. It combines a bi-modal microscopy setup with an automated cell picking system. It has been developed in a close collaboration between the different partners of the project, i.e. ALS, CEA, Iprasense and ENS-Lyon. To achieve single cell picking, the live-imaging acquisition and the cell picking module must match to within one micrometer, therefore a linear path measurement system, has been developed. The integration of the ‘silico visual cortex’ to this new automation will deliver the microscope able of cell fate prediction and decision making that we envision.
(3) Discover biological signatures at the origin of a specific disease
Hepatocellular carcinoma (HCC) is the most common liver cancer and the second most frequent cause of cancer-related deaths worldwide. The most recent clinical research has pointed to extensive biological heterogeneity at the level of cell morphologies, genetic fingerprints and responses to drug treatment in liver cancer patients. Our stated objective in project REVEAL is to validate the capacity of neuronal microscopy to detect and identify the biological heterogeneity that take root in patients very early on- during early stages of dysmetabolism in the liver – far before patients experience steatohepatitis, fibrosis, liver adenomas or cancer.
Under the aegis of this project, we have now characterized many different types of patient derived cells that grow on a 2D matrix in laboratory culture as well as patient derived normal or tumoral 3D organoids- mini livers- that reflect either the health or the cancerous state in how they propogate in laboratory conditions. Moreover, we have developed state of the art murine 2D cellular and 3D mini livers that reflect nearly every stage of liver disease- from early stage metabolic disorders to late stage cancer. As a first step, these laboratory models have been used to develop both the 2D and the 3D neuronal microscope, and immediately we have realized the very first stated goal of the project- that our laboratory models do indeed capture the heterogeneity seen in the clinical patient-derived samples. We are actively interrogating the molecular signatures of these samples with the aim to provide a state of the art “omics” signature to the images acquired by the neuronal microscope. In the coming months, this critical step will be achieved.
At M18, one achievement goes beyond state of the art. It is the development of a multi-slice light scattering model (beam propagation method) inside a deep-learning framework (diffractive deep neural network, DDNN) to reconstruct the 3D objects. In other words, it is now demonstrated that a neural network allows reconstructing 3D biological objects, hence making the image itself and not only its analysis. This new framework will be assessed in the ongoing months, to determine whether quantitative cell measurements are at reach, a work led by Politechnika Warszawska (WUT). This framework will be further developed, to test whether continuous training of the networks can improve image rendering and integrate cell analysis in the pipeline.
Overall, the scope of this project is to confirm that a new methodology is at reach: spatio-temporal omics. This goes beyond the state-of-the art of spatial-omics and will let us study in a near future the dynamic processes at the origin of diseases.
Overall, the scope of this project is to confirm that a new methodology is at reach: spatio-temporal omics. This goes beyond the state-of-the art of spatial-omics and will let us study in a near future the dynamic processes at the origin of diseases.