Post-doctorant.e en Imagerie hyperspectrale computationnelle rapide

Type de structure : The optical imaging research group of CREATIS is dedicated to advancing computational imaging systems that harness the synergy of hardware and software innovations [1, 2], with a specific focus on acquiring hyperspectral data cubes characterized by two spatial dimensions and one spectral dimension. To this end, we’ve developed an in-house hyperspectral imaging setup (as depicted in Figure 1a) comprised of a digital micromirror device (DMD), a spectrometer, and a camera. In this setup, the spectrometer captures a series of spectra associated with predefined light patterns loaded onto the DMD, allowing us to reconstruct the hyperspectral data cube. This innovative system design stems from our main objective: optimizing glioma resection to enhance the life expectancy of patients.Currently, glioma removal involves a surgical procedure that entails opening the cranial cavity and extracting the tumor. During this operation, neurosurgeons employ a fluorescence microscope to observe the emission of Protoporphyrin IX (PpIX), a fluorophore selectively present in tumor cells. However, the current fluorescence microscope can only provide information about the total fluorescence intensity. Our research team has demonstrated that by leveraging the complete PpIX spectrum present in gliomas, it is possible to distinguish the boundaries between tumor tissue and healthy tissue [3].Consequently, our initial setup is transported to the neurosurgical operating room, allowing us to acquire hyperspectral data cubes from biopsies taken during the procedure (as illustrated in Figure 1b). These biopsies exhibit distinct spectral characteristics (as depicted in Figure 1b) at various locations, enabling us to pinpoint the boundary between healthy and tumoral cells. This innovative approach holds great promise for improving the precision and effectiveness of glioma resection procedures.Contexte et mission : OBJECTIVES:We are actively seeking a postdoctoral fellow or an engineer to develop a high-speed computational imaging system tailored for hyperspectral imaging. This camera should be capable of detecting Protoporphyrin fluorescence signals, specifically in the context of glioma resection. The role, funded by ANR, will encompass responsibilities from conceptualization and simulation to system design, implementation, and characterization. The project includes acquiring hypercubes from ex-vivo and in-vivo glioma samples directly within the hospital’s operating room.CHALLENGECurrently, the acquisition time is suitable for biopsy samples but impractical for in-vivo imaging, especially considering the limited spatial resolution. Our research group has developed compressive imaging strategies to reduce the number of patterns needed for acquisition, thereby decreasing total acquisition time. This involves deep neural networks to reconstruct hyperspectral cubes from observed data, with various architectures proposed [4], which have been effective in denoising experimental data. However, hardware development is necessary to further increase speed for clinical in-vivo applications.MAINS TASKS :Enhance the existing optical setup (software and hardware improvements)Develop a second optical setup for faster acquisitionDevelop instrumentation control software based on the SPAS package [5]Implement advanced reconstruction algorithms using deep learning (based on the spyrit package [6])Characterize the imaging device (sensitivity, spatial and spectral resolution)More specifically, the project involves two main tasks:Optical Setup Enhancement and Biopsy Imaging: Improve the current setup for hyperspectral data acquisition from biopsies, compare results with pathological analysis to distinguish tumor from healthy tissue.Development of a 2D Sensor Spectrometer: Create a new optical setup with a 2D sensor to accelerate data acquisition. This involves constructing a spectrometer with a cylindrical lens, prism, or grating, and updating software to handle 2D patterns, enabling faster imaging, e.g., 64x faster for a 64x64 image.Rémunération : Salary: €2000-€2400 net/month depending on experience. Duration: 18 months.Diplômes requis : Strong background in optical setup design and proficiency in Python programming are required.Contact : Send CV and motivation letter to mahieu@creatis.insa-lyon.fr and Bruno.montcel@creatis.insa-lyon.fr. More information at this link. #J-18808-Ljbffr