The project consortium of CONcISE joins forces to develop biomedical imaging devices where the hardware and software are completely integrated and developed together, and that allow a more data-efficient and quality-oriented scanning of biological tissues.
DCs will work on individual doctoral projects in order to develop novel and unconventional techniques for multi-dimensional biomedical optical imaging for the detection and mapping of absorption, scattering and fluorescence, in biological tissues using visible and near-infrared light.
The objective of the network and the doctoral programmes is to overcome traditional limitations of the current biomedical instrumentation, which focus on the maximisation of the data acquired, regardless of its quality. This leads to a large amount of data to manage, transfer, and analyse and thus creates data bottlenecks and gaps in measurements.
Supported researchers must be doctoral candidates, i.e. not already in possession of a doctoral degree at the date of the recruitment.
Mobility Rule: researchers must not have resided or carried out their main activity (work, studies, etc.) in the country of the recruiting beneficiary for more than 12 months in the 36 months immediately before their recruitment date.
Additionally, DC applicants must fulfil the local requirements of the recruiting institutions listed in the project descriptions below.
Scan or click on the QR Code to read the MSCA Guide for applicants
CONcISE offers the DCs research and training excellence for the design of three novel systems to address relevant paradigmatic problems related to biomedical imaging. The project partners will lead three main research themes, in addition to an advanced training programme, organised in six ‘topical schools’, for the development and implementation of new cutting-edge systems for unconventional multi-dimensional biomedical optical imaging.
DCs will be offered the opportunity to work in a multidisciplinary way, moving from hardware to software development and vice versa, being at the best position to fill the gap between system development and data analysis, which is often the result of the traditional monothematic training typically adopted in biomedical imaging research.
To complement the academic and scientific goals of the DCs, the project will offer customised research projects, structured interdisciplinary, network-wide transferable skills training activities, and secondments at top-ranking European universities, research institutes, and industry partners.
Project Description: The activities for the candidate will be focused on the design and development of an optical fibre laser for microscopy applications. Most of the activities will be involved in the optical fibre manipulation, so ability and calm are important skills for the tasks. The candidate has to be a dynamic person with capacity to work in a team group and collaborate with the rest of the team. Electronic and programming experience will be also useful for the integration of the optical part inside with the electrical and software.
Expected Results: industrialised few-cycle femtosecond supercontinuum source and system for actual needs of the multiphoton microscopy industry. Trained DCs to bridging the gap between scientific achievements and their commercialisation in the bioimaging sector.
Project Description: Diffuse optical tomography (DOT) is an imaging modality where the optical properties of a biological sample are estimated from boundary measurements of visible or near-infrared light. The objective of the work is to develop a system for time-resolved multi-spectral diffuse optical tomography implementing structured light illumination and single-pixel camera detection working for biological tissues on the macro scale. This system will enable the implementation of adaptive, data-driven strategies to strongly reduce the number of acquisitions without information loss. The candidate will collaborate with DC2 for the modelling of light propagation in the tissue and image reconstruction, and with DC3 for the implementation of deep-learning strategies.
Expected Results: Construction of a system for multispectral diffuse optical tomography based on structured illumination and compress sensing. Implementation of an adaptive, data-driven acquisition strategy for maximising the information content. Performance assessment and validation of the system with dedicated phantoms mimicking the optical properties of biological tissues (work performed jointly with DC2).
Project Description: Diffuse optical tomography (DOT) is an imaging modality where the optical properties of the target are estimated from boundary measurements of visible or near-infrared light. The image reconstruction problem in DOT is an ill-posed problem that needs to be approached in the framework of inverse problems. Solving this problem requires modelling of light transport and applying methods of numerical optimisation for image reconstruction. The objectives and research activities of the DC include development of light transport modelling algorithms and image reconstruction methods for multispectral DOT, implementation of the algorithms for the multispectral diffuse DOT system based on structured illuminations, and evaluation of the developed methodologies. The DC will collaborate with DC1 and DC3 on the development of the smart DOT system.
Expected Results: implementation of software for simulating light transport in multispectral diffuse optical tomography system based on structured illumination and compress sensing. Implementation of image reconstruction algorithms for adaptive diffuse optical tomography system that maximise the information content. Performance assessment and validation of the system with dedicated phantoms mimicking the optical properties of biological tissues (work performed jointly with DC1 during secondment 3).
Project Description: in recent years AI proved to be extremely successful in providing decision support within both the healthcare and biology sectors. Specifically in the context of medical images, new developments in the field of deep and reinforcement learning could bring further enhancements when applied to visible or near-infrared light (diffuse optical tomography DOT). This multidisciplinary project expects to cover multiple areas including the development of a hardware and software framework for data acquisition; developing a software for online data source management; research methodologies for information quantification in biological tissues; generate simulated data; research and develop deep-learning based strategies for image processing; research and design a data processing pipeline for applied AI. The candidate will collaborate with an international team and be exposed to the latest software development best practices and technologies.
Expected Results: Implementation and testing of strategies for image reconstruction on experimental data acquired with a system based on structured illumination and compress sensing. Devise an adaptive strategy using CS combined with deep-learning (aCS+DL) strategy for diffuse optical tomography aiming at reducing the amount of acquired data without information loss. Performance assessment and validation of the system will be carried out using dedicated phantoms mimicking the optical properties of biological tissues.
Project Description: With a strong track record of clinical translation at IRCAD, the optimised endoscopic imaging prototypes will be tested during preclinical trials to validate their performance in real surgical situations. In particular, we will perform perfusion experiments, such as colon resection and anastomosis, as well as sentinel lymph node mapping in the lung. The results from these procedures will be compared to state-of-the-art clinical systems used today in surgery for fluorescence imaging.
Expected Results: Investigation of novel spatially-resolved multispectral imaging acquisition & processing methods for advanced (real-time & quantitative) fluorescence imaging. Design and fabrication of a clinical-compatible endoscopic platform integrating advanced fluorescence. Performance assessment and validation of the system with dedicated diffused and fluorescent phantoms. Work will be performed jointly with DC5 and DC11.
Project Description: Fluorescence imaging is a fundamental tool for clinical application. In particular, a multidimensional data set (spectrum, time, space) allows to improve the information capability and, hence, its diagnostic purpose. The objective of the work is to design, develop and validate a fast multispectral time-resolved imaging system based on a computational imaging approach. The system will exploit novel acquisition and processing methods and their integration within the system. The candidate will collaborate with DC6 for integrating and testing the developed computational architectures on experimental system, and with DC4 for the clinical strategies.
Expected Results: Investigation of novel spatially- and temporally- resolved multispectral imaging acquisition & processing methods for advanced (realtime & quantitative) fluorescence imaging. Design and fabrication of a multispectral time-resolved imaging system combined with spatial modulation for both illumination and detection. Performance testing of methods and systems for advanced fluorescence imaging with dedicated diffused and fluorescent phantoms. Work will be performed jointly with DC4 and DC11.
Project Description: the main objective of the DC will be to develop a modular optical system for multiphoton microscopy with light structured illumination and computational adaptive compressive strategies. In collaboration with other doctoral candidates of the network, the candidate will integrate adaptive optics techniques in the illumination stage of the optical system, will integrate machine learning approaches for data analysis and adaptive measurements and will apply the system for imaging of biological tissues by using phantoms.
Expected Results: Prototype of an optical system for wide-field multiphoton microscopy with optimised light structured illumination and computational adaptive compressive strategies. New algorithms for adaptive sampling with light structured illumination and image reconstruction providing high resolution in short measurement times. Performance assessment and validation of the system with phantoms adapted for multiphoton microscopy mimicking the optical properties of biological tissues.
Project Description: The main objective of the doctoral candidate will be the design of light detection modules for multidimensional nonlinear microscopy with structured light and compressive sensing. The candidate will develop new methods and algorithms to combine information from different sensors based on data fusion techniques. Also new algorithms for compressive sensing with spatial modulation and multiple detectors. In collaboration with other doctoral candidates of the network, the candidate will work on the 8integration of the detection modules in a nonlinear microscope and in the evaluation of the performance using phantoms.
Expected Results: Prototype of a detection module with multiple sensors for nonlinear microscopy with structured illumination and compressive sensing. New algorithms for image reconstruction with compressive sampling based on data fusion techniques providing high resolution in short measurement times. Performance assessment and validation of the system with phantoms.
Project Description: the aim of DC8 is to design an adaptive optics module for nonlinear microscopy based on the use of Multi Actuator adaptive Lenses and wavefront sensor optimisation algorithms. Comparison of results obtained with closed loop systems and sensor-less systems. Optimisation of the wavefront correction for volumetric image acquisition. Additionally, it aims to integrate adaptive optics module in the nonlinear microscope, evaluation of the performance and test with biological samples.
Expected Results: proof of concept design of the adaptive optics module with integrated multi actuators adaptive lens for nonlinear microscopy with wavefront sensor-less control algorithm. Demonstration of the increasement of the penetration depth in biological samples.
Project Description: A key challenge in many established imaging modalities is to reduce the number of measurements needed to sense all the relevant information contained in the image: For instance, high-resolution imaging of biological samples with wide-field 2-Photon Microscopy (also known as “non-linear microscopy”) relies on the spatial focusing of a femtosecond laser beam up to its diffraction limit for subsequent serial raster scanning, which usually requires the use of adaptive optics to correct for wavefront distortions and limits the temporal resolution obtainable. Traditionally, the information content of an image is associated with its Fourier representation, and spatial sensing patterns are designed to capture all frequencies up to a maximal one. The latter is often determined by the physical nature of the imaging process, e.g., images derived after propagation of light in biological tissues typically have a reduced Fourier bandwidth due to the high scattering. However, as the spatial complexity of many tissue structures is rather low, images recorded in a conventional fashion with N pixels can still often be compressed, i.e., stored compactly with M < N coefficients in a suitable representation system. This means that the original image contained redundant information. It may, therefore, be possible to speed up the image acquisition without a significant loss of image quality by exploiting this redundancy and directly measure a subset of the data chosen in such a way as to maximize its non-redundancy. To retrieve the image from this compressed data, non-linear image reconstruction techniques must be used. These concepts, established as the field of compressed sensing (CS), have been applied to many imaging modalities with success. In conventional CS, a fixed sensing pattern is designed for a given class of images a-prior, i.e., before any measurements of a particular image are taken. The idea of adaptive compressed sensing (aCS) is to further adapt the sensing pattern to an image during the measurement, i.e., a-posteriori. Conventional approaches to perform aCS such as optimal experimental design lead to complex numerical optimization problems that can rarely be solved fast enough to realize aCS in real-world applications.
The DC will examine the use of adaptation approaches based on deep learning (DL), in particular deep neural networks with convolutional layers (CNNs) for the image reconstruction from compressed data and deep reinforcement learning (RL) for the adaptation. We will develop mathematical formulation, numerical algorithms and implementation of adaptive compressed sensing using deep learning techniques (aCS-DL) and use the new methods to improve wide-field 2-Photon Microscopy: Within the CONcISE project, the DC is part of the “SMART-2PM” team which will develop a system that reaches higher penetration depth, sensitivity, and imaging speed.
Expected Results: Proof-of-concept of aCS-DL with a simulation study. Design and integration of dedicated aCS-DL software for SMART-2PM system. Demonstration of aCS-DL for non-linear microscopy for biological tissues.
Project Description: The project will be within the SMART-FLUO theme of the doctoral training network. The overall aim will be to develop a system to perform multispectral fluorescence imaging, absorption and scattering using endoscopes and structured light illumination. The system will enable quantitative fluorescence endoscopy imaging, making multidimensional fluorescence measurements repeatable and interpretable in real-time through an endoscope. The system aims to provide the physician with a fast-imaging tool to be used during guided-surgery with exogenous contrast agents. The DC will (1) investigate whether learned CNN based decoders can be combined with an experimental encoder system based on spatial light patterns; (2) develop an architecture for generating an optimal encoding adapted to experimental data; (3) test the developed architectures on experimental systems developed at POLIMI and IRCAD.
Expected Results: a novel network architecture that is capable of generating an optimal encoder for any given experimental conditions together with a CNN based decoder. Demonstration of fast real-time adaptive image acquisition using the developed architecture. Simultaneous development of fast reconstruction using this network architecture. Work will be performed jointly with DC4 and DC5.