Periodic Reporting for period 1 - NEXTSCREEN (Training network for Next generation cellular screening)
Periodo di rendicontazione: 2023-12-01 al 2025-11-30
Biomedical screening at single-cell and bioparticles level has the potential to transform clinical diagnostics, but the research and development in this field are scattered in different disciplines: biophotonics, micromanipulation, machine learning, in vitro diagnostics, and clinical regulations are traditionally imparted in separate training programs. NEXTSCREEN aims to train and establish a network of researchers (Doctoral Candidates - DCs) with the expertise required for the development of next-generation screening methods, based on automatic imaging and classification of samples moving along a liquid stream. The researchers have the objectives to reduce the cost and complexity of imaging flow cytometry; empower it with novel contrast mechanisms; build high-resolution automatic microscopes at the diffraction limit and beyond; develop real-time data processing tools able to detect and recognize the samples, circumventing the need for manual annotation. Using these technologies they will characterize blood cells and bioparticles, screening large cellular populations, with the goal to identify and characterize cancer biomarkers, in samples derived from liquid biopsies. The ultimate goal is to initiate the development of diagnostics tools, that could be adopted in clinical settings on a large scale, democratizing the use of automatic screening. The project brings together research groups, small and large companies that are leading the field of imaging flow cytometry, with complementary know-how in high-resolution microscopy, high-precision microfluidics, biotechnologies, and weakly-supervised deep learning, that will facilitate the development of data-driven cells and bioparticle screening.
The Scientific Objectives are:
SO1: to lower the cost and complexity of automatic cellular and particle screening while maintaining high resolution and throughput;
SO2: to extend the capabilities of imaging flow cytometry with novel contrast mechanisms and higher resolution;
SO3: to demonstrate blood-cell classification by deep learning while avoiding manual annotation in imaging flow cytometry;
SO4: to introduce protocols for quality control, reporting, and reproducibility in imaging flow cytometry;
SO5: to adopt imaging flow cytometry for the identification and characterization of tumour biomarkers in blood-derived samples.
Progress toward SO1 includes the development of components and subsystems. DC1 demonstrated a modular optofluidic architecture combining disposable PDMS microfluidic cartridges with a reusable fused-silica light-sheet optical engine, explicitly addressing reduction of alignment complexity; inertial focusing was experimentally validated at flow velocities approaching ~1 m/s, and a glass chip fabricated with femtosecond laser micromachining enables light-sheet imaging with diffraction-limited performances (≈1 µm beam waist). In parallel, DC2 established a reproducible workflow for low-refractive-index, biocompatible polymer materials compatible with two-photon polymerization, potentially enabling compact and scalable optofluidic probes. Integration toward a fully monolithic lab-on-chip device is ongoing.
Progress toward SO2 includes the implementation of new imaging and phenotyping modalities. Light-sheet IFC concepts were implemented and validated on standardized and biological samples, with demonstrated optical sectioning and sub-micron lateral resolution (DC3, DC9, DC10). High-throughput deformability cytometry was validated on clinical samples, achieving discrimination between healthy donors and leukemia patients (DC4). Super-resolution imaging under flow-compatible constraints was advanced by the undergoing development of millisecondscale single molecule localization acquisitions for extracellular vesicles (DC5).
Progress toward SO3 is well advanced. DC8 developed a weakly supervised segmentation framework enabling real-time, label-free analysis compatible with acquisition-scale data rates and achieving ~95% classification accuracy. DC7 benchmarked state-of-the-art and foundationmodel-based segmentation approaches across heterogeneous datasets, demonstrating improved generalization. Consolidation of these results and publication preparation are in progress.
Progress toward SO4 involves the characterization of samples and the development of reproducible measurement strategies. DC6 established quantitatively validated, feedbackcontrolled flow-focusing platforms with reproducible performance across wide flow-rate and viscosity ranges, providing design rules relevant for industry standardization. Across WP1–WP3 activities, standardized beads and biological references were consistently used to assess resolution, flow stability, and imaging performance. The study of cell and extracellular vesicles mimicking samples has been initiated.
Progress toward SO5 is, as expected at early stage. Yet, deformability and morphology based biomarkers for hematological disease were validated (DC4), supported by real-time computational pipelines (DC8). On the extracellular vesicle axis, DC11 optimized the production of fluorescence-tagged EV reference materials, while DC9 and DC10 developed high thougput strategies that are ideally suited for EV characterization. The study of specific biomarkers and fully integrated diagnostic workflows is ongoing.
SO1: to lower the cost and complexity of automatic cellular and particle screening while maintaining high resolution and throughput;
SO2: to extend the capabilities of imaging flow cytometry with novel contrast mechanisms and higher resolution;
SO3: to demonstrate blood-cell classification by deep learning while avoiding manual annotation in imaging flow cytometry;
SO4: to introduce protocols for quality control, reporting, and reproducibility in imaging flow cytometry;
SO5: to adopt imaging flow cytometry for the identification and characterization of tumour biomarkers in blood-derived samples.
Progress toward SO1 includes the development of components and subsystems. DC1 demonstrated a modular optofluidic architecture combining disposable PDMS microfluidic cartridges with a reusable fused-silica light-sheet optical engine, explicitly addressing reduction of alignment complexity; inertial focusing was experimentally validated at flow velocities approaching ~1 m/s, and a glass chip fabricated with femtosecond laser micromachining enables light-sheet imaging with diffraction-limited performances (≈1 µm beam waist). In parallel, DC2 established a reproducible workflow for low-refractive-index, biocompatible polymer materials compatible with two-photon polymerization, potentially enabling compact and scalable optofluidic probes. Integration toward a fully monolithic lab-on-chip device is ongoing.
Progress toward SO2 includes the implementation of new imaging and phenotyping modalities. Light-sheet IFC concepts were implemented and validated on standardized and biological samples, with demonstrated optical sectioning and sub-micron lateral resolution (DC3, DC9, DC10). High-throughput deformability cytometry was validated on clinical samples, achieving discrimination between healthy donors and leukemia patients (DC4). Super-resolution imaging under flow-compatible constraints was advanced by the undergoing development of millisecondscale single molecule localization acquisitions for extracellular vesicles (DC5).
Progress toward SO3 is well advanced. DC8 developed a weakly supervised segmentation framework enabling real-time, label-free analysis compatible with acquisition-scale data rates and achieving ~95% classification accuracy. DC7 benchmarked state-of-the-art and foundationmodel-based segmentation approaches across heterogeneous datasets, demonstrating improved generalization. Consolidation of these results and publication preparation are in progress.
Progress toward SO4 involves the characterization of samples and the development of reproducible measurement strategies. DC6 established quantitatively validated, feedbackcontrolled flow-focusing platforms with reproducible performance across wide flow-rate and viscosity ranges, providing design rules relevant for industry standardization. Across WP1–WP3 activities, standardized beads and biological references were consistently used to assess resolution, flow stability, and imaging performance. The study of cell and extracellular vesicles mimicking samples has been initiated.
Progress toward SO5 is, as expected at early stage. Yet, deformability and morphology based biomarkers for hematological disease were validated (DC4), supported by real-time computational pipelines (DC8). On the extracellular vesicle axis, DC11 optimized the production of fluorescence-tagged EV reference materials, while DC9 and DC10 developed high thougput strategies that are ideally suited for EV characterization. The study of specific biomarkers and fully integrated diagnostic workflows is ongoing.
From the scientific perspective, the project has already generated results in the field of imaging flow cytometry, including microfluidic handling, advanced optical imaging, and real-time data analysis. Quantitative outcomes include the demonstration of stable inertial focusing at flow velocities approaching 1 m/s, diffraction-limited light-sheet generation (~1 µm beam waist), sub-micron volumetric imaging under flow conditions, millisecond-scale super-resolution imaging of extracellular vesicles, and deformability-based blood cell classification. These results address the project’s baseline assumptions, namely that IFC performance can be extended beyond current commercial systems by combining modular optofluidics with data-driven analysis.
From an industrial and economic perspective, the project has started the path toward developing modular, scalable, and reusable components that reduce system complexity and facilitate future manufacturability. The separation between disposable microfluidic devices and reusable optical modules, the use of polymer materials compatible with scalable fabrication techniques, and the implementation of feedback-controlled flow systems provide could be the basis for industrial translation. While, as expected this development are at a preliminary stage, the interaction with industrial partners and the identification of exploitable foreground in microfluidic flow control, optofluidic probes, and imaging subsystems indicate that the expected economic and technological impacts remain valid and achievable within the project timeframe.
Societal impact is being built through the project’s focus on minimally invasive diagnostics and liquid biopsy applications, particularly for hematological diseases. The validation of deformability- and morphology-based biomarkers on patient samples, combined with real-time, label-free analysis pipelines, demonstrates the potential of NEXTSCREEN technologies to support earlier, faster, and more accessible diagnostics, aligning with the long-term goal of deployment in clinical settings by non-technical users.
From an industrial and economic perspective, the project has started the path toward developing modular, scalable, and reusable components that reduce system complexity and facilitate future manufacturability. The separation between disposable microfluidic devices and reusable optical modules, the use of polymer materials compatible with scalable fabrication techniques, and the implementation of feedback-controlled flow systems provide could be the basis for industrial translation. While, as expected this development are at a preliminary stage, the interaction with industrial partners and the identification of exploitable foreground in microfluidic flow control, optofluidic probes, and imaging subsystems indicate that the expected economic and technological impacts remain valid and achievable within the project timeframe.
Societal impact is being built through the project’s focus on minimally invasive diagnostics and liquid biopsy applications, particularly for hematological diseases. The validation of deformability- and morphology-based biomarkers on patient samples, combined with real-time, label-free analysis pipelines, demonstrates the potential of NEXTSCREEN technologies to support earlier, faster, and more accessible diagnostics, aligning with the long-term goal of deployment in clinical settings by non-technical users.