Periodic Reporting for period 2 - OrganVision (OrganVision: Technology for real-time visualizing and modelling of fundamental process in living organoids towards new insights into organ-specific health, disease, and recovery)
Reporting period: 2022-07-01 to 2023-12-31
Problem statement: OrganVision envisions to shift the paradigm for organoids from disease or drug-screening models to observable tissue micro-bio-environment for unravelling key physiological and pathological processes in humans. We aspire to enable the exploitation of the most important feature of organoids – life, health, disease, and death unfolding in real-time at sub-cellular and inter-cellular scales in cells and tissues, respectively. Our ambition is to (a) overcome the central obstacle that prevents realization of the vision, namely lack of real-time high-resolution label-free imaging technology suitable for organoids, and (b) create new opportunities for organoid research and exploit them.
Importance to society: Cardiovascular disease (CVD) causes over 1.8 million deaths in EU each year (~37% of all deaths in the EU) and is estimated to cost the EU economy €210 billion a year . It is the leading cause of human deaths globally, and a high priority in the UN’s sustainable development goal no. 3 (good health and well-being). Understanding the mechanisms of CVDs and regenerative and repair potential definitely has a direct impact on the global cardiovascular health, potentially leading to better prevention, treatment, and cure of CVD. The technology platform will also be tested on cancer spheroids, therefore readying it for a big impact in the future. The technologies developed will serve the society and the globe by advancing understanding and resolving other diseases as well. Lastly, with better exploitation of organoids, the number of primary cells and tissues harvested from animals for biological studies will reduce. This will contribute to animal welfare.
The broad objectives of OrganVision are:
• Create a ground-breaking comprehensive science & technology platform that alters state-of-the-art across multiple disciplines by leaps, and thereby infusing new and promising R&D streams in each of them
• Redefine the possibilities for organoid research and its impact on finding new age solutions for diseases and supporting new knowledge discovery
• Create a lasting impact on global cardiovascular health
• Build a strong consortium of young research leaders
The scientific objectives of OrganVision are:
1. To develop a novel multi-scale imaging technology for real-time living tissue imaging
2. To develop a novel artificial intelligence engine that is able to model collective dynamics of life processes in organoid
3. To undertake biological studies for understanding cardiac toxicity, cell-cell interaction mechanisms and their impact on post-injury cardiac health, and changes in mitochondrial transfer and turnover in a cardiac injury setting
Importance to society: Cardiovascular disease (CVD) causes over 1.8 million deaths in EU each year (~37% of all deaths in the EU) and is estimated to cost the EU economy €210 billion a year . It is the leading cause of human deaths globally, and a high priority in the UN’s sustainable development goal no. 3 (good health and well-being). Understanding the mechanisms of CVDs and regenerative and repair potential definitely has a direct impact on the global cardiovascular health, potentially leading to better prevention, treatment, and cure of CVD. The technology platform will also be tested on cancer spheroids, therefore readying it for a big impact in the future. The technologies developed will serve the society and the globe by advancing understanding and resolving other diseases as well. Lastly, with better exploitation of organoids, the number of primary cells and tissues harvested from animals for biological studies will reduce. This will contribute to animal welfare.
The broad objectives of OrganVision are:
• Create a ground-breaking comprehensive science & technology platform that alters state-of-the-art across multiple disciplines by leaps, and thereby infusing new and promising R&D streams in each of them
• Redefine the possibilities for organoid research and its impact on finding new age solutions for diseases and supporting new knowledge discovery
• Create a lasting impact on global cardiovascular health
• Build a strong consortium of young research leaders
The scientific objectives of OrganVision are:
1. To develop a novel multi-scale imaging technology for real-time living tissue imaging
2. To develop a novel artificial intelligence engine that is able to model collective dynamics of life processes in organoid
3. To undertake biological studies for understanding cardiac toxicity, cell-cell interaction mechanisms and their impact on post-injury cardiac health, and changes in mitochondrial transfer and turnover in a cardiac injury setting
Progress in WP1: We have made progress on two different fronts. First, we assembled the components required for rapidly creating tunable two-dimensional (2D) light patterns in a commercial optical microscope. After careful alignment, we characterized the optical performance of such a microscope, including signal-to-noise ratio, spatial resolution, and imaging speed. We also performed some proof-of-concept experiments consisting of imaging static and dynamic fluorescent and non-fluorescent samples. Secondly, we worked on the high-speed control of light along the optical axis (z-axis). The obtained speed, in the order of kilohertz, enabled us to extract quantitative phase information (QPI) from non-fluorescent specimens in real-time. We applied such a system to perform QPI of unlabeled histopathological samples with sub-cellular resolution over centimeter-squared areas in seconds.
Progress on WP2: We have developed a blind 3D deconvolution and point spread function (PSF) estimation algorithm. We have developed a strategy for illumination and sample-agnostic sample reconstruction. Our approach allows reconstruction of the sample despite the obfuscation of the illumination. We have done initial work in adapting multiple-scattering based solvers for phase-less polarization-insensitive measurements in darkfield label-free microscopy.
Progress on WP3: A data pipeline has been developed to acquire, pre-process, and store correlated fluorescence and label-free (Co-Fl-LF) images. A large dataset from the Allen Institute for Cell Science provides the basis for the large-scale training of the digital-staining model for transmitted light (bright-field) label-free inputs. Various architectures and combinations of architectures were evaluated for the final model. A single, combined model, predicting all the structures of interest in a single shot from bright-field input images, has been trained and optimized.
Progress on WP4: We have developed an easy to assemble EHT holder that is microscopy compatible and can take chemical stimuli. We have also shown its compatibility for non-linear label-free imaging, for example using CARS.
Progress on WP5: We have been able to develop several key 3D in vitro systems that act as perfect validation models for MUSIT and provide platforms for achieving intermediate milestones to test the full depth of capabilities and limitations of MUSIT. In addition, we have been able to show the results of integration of the MUSIT instrument and MUSIT algorithm for a fluorescently labeled thin sample. We have developed sophisticated 3D cell culture models for benchmarking the performance of microscopes on thick samples.
Progress on WP6: We have generated multicellular engineered heart tissues (EHTs), containing hiPSC derived cardiomyocytes and hiPSC-derived fibroblasts. We have developed an engineered heart tissue-based injury model to undertake injury response studies. We have also established protocols for studying mitochondria degradation and transfer in EHTs. We have shown new biological insights related to EHT maturation and have shown that cardiac recovery can be supported if cardiac work load on cardiomyocytes can be temporarily switched off.
Progress on WP7: We have opened a new field of scalable AI and developed variety of AI architectures and learning mechanisms for them. We have developed generative AI based sophisticated shape and behavior simulators for mitochondria. We have demonstrated simulation supervised learning using these simulators.
Progress on WP2: We have developed a blind 3D deconvolution and point spread function (PSF) estimation algorithm. We have developed a strategy for illumination and sample-agnostic sample reconstruction. Our approach allows reconstruction of the sample despite the obfuscation of the illumination. We have done initial work in adapting multiple-scattering based solvers for phase-less polarization-insensitive measurements in darkfield label-free microscopy.
Progress on WP3: A data pipeline has been developed to acquire, pre-process, and store correlated fluorescence and label-free (Co-Fl-LF) images. A large dataset from the Allen Institute for Cell Science provides the basis for the large-scale training of the digital-staining model for transmitted light (bright-field) label-free inputs. Various architectures and combinations of architectures were evaluated for the final model. A single, combined model, predicting all the structures of interest in a single shot from bright-field input images, has been trained and optimized.
Progress on WP4: We have developed an easy to assemble EHT holder that is microscopy compatible and can take chemical stimuli. We have also shown its compatibility for non-linear label-free imaging, for example using CARS.
Progress on WP5: We have been able to develop several key 3D in vitro systems that act as perfect validation models for MUSIT and provide platforms for achieving intermediate milestones to test the full depth of capabilities and limitations of MUSIT. In addition, we have been able to show the results of integration of the MUSIT instrument and MUSIT algorithm for a fluorescently labeled thin sample. We have developed sophisticated 3D cell culture models for benchmarking the performance of microscopes on thick samples.
Progress on WP6: We have generated multicellular engineered heart tissues (EHTs), containing hiPSC derived cardiomyocytes and hiPSC-derived fibroblasts. We have developed an engineered heart tissue-based injury model to undertake injury response studies. We have also established protocols for studying mitochondria degradation and transfer in EHTs. We have shown new biological insights related to EHT maturation and have shown that cardiac recovery can be supported if cardiac work load on cardiomyocytes can be temporarily switched off.
Progress on WP7: We have opened a new field of scalable AI and developed variety of AI architectures and learning mechanisms for them. We have developed generative AI based sophisticated shape and behavior simulators for mitochondria. We have demonstrated simulation supervised learning using these simulators.
A high throughput single pixel imaging system with automatic identification of position of fluorescent samples has been developed. A new approach to estimate the PSF of microscopes, and illumination agnostic reconstruction of fluorescence samples has been developed. New multicolor 3D virtual staining engine has been developed. New EHT holding platforms with control of stimuli have been developed. New 3D cell cultures for performing a variety of biological assays have been developed. New protocols for developing mitophagy reporters in iHPC derived cardiac cells have been developed. Also, new EHT protocol has been developed such that optical signal can be used to switch off the functional load of cardiomyocytes. Several new AI approaches have been developed to deal with uncertain number and nature of features, data deficient learning models, and simulation engines for subcellular structures.
3 intellectual properties have emerged. More are underway.
3 intellectual properties have emerged. More are underway.