By Cnr Ifn
This research activity aims at developing novel, unconventional techniques for multi-dimensional biomedical optical imaging to detect and map absorption, scattering, and fluorescence in biological tissues using visible and near-infrared light. The possibility of detecting other dimensions like time and wavelength, together with space, dramatically improves the imaging and diagnostic power of biomedical devices. Under a classical, conventional approach to imaging, more dimensions mean more and more data to acquire but…more and more data does not necessarily mean more information! This is why we aim to optimize the sensing strategy to acquire less data as possible without losing information. Our research is based on a Compuational imaging approach: the final image is not directly measured but obtained by hard computational processing where different techniques, such as single-pixel imaging, compressed sensing, data fusion, and others, can be combined together. Hereafter is a short description. Single-pixel imaging Developed at the RICE University, single-pixel imaging allows imaging through a single-element detector (also called "bucket detector") taking advantage of spatial light modulators, such as a Digital Micromirror Device (DMD). A series of spatial patterns are overlapped to the target image and the resulting light is focused on a single-element detector. After a reconstruction phase that can range from a simple matrix inversion to a more complex constrained minimization, the target image is eventually reconstructed. The fact that we can compress the spatial dimension into a single-element detector opens the way to exploring other dimensions (such as lifetime, spectrum, polarization, and depth) or other spectral ranges where cameras are not available. Compressed sensing This computational framework allows the recovery of an image which has a sparse representation by a number of measurements that is much less than the number of pixels. This can be done by a constrained minimization enforcing sparsity in different domains (e.g. pixel, gradient, Wavelet, etc.). The single-pixel camera is the enabling technology of compressed sensing: in fact, we can directly compress during the acquisition phase by acquiring a small set of measurements. Data fusion Developed in collaboration with the Photonics Research Group (GROC) @ UJI (University Jaume I) this technique allows us to fuse together datasets obtained with different detectors. In particular, we have recently applied and published this technique to recover a high-resolution 4D hypercube (space, time on the ps, spectrum) by fusing a low-resolution multi-dimensional microscopy dataset derived bt a low-resolution single-pixel camera and a high-resolution image obtained by a CMOS camera. Experimental Setups We mainly work on two setups: one dedicated to imaging and tomography in diffusive media on the macro scale, and another working on the microscopy scale. Time-resolved Diffuse Optical Tomography (TR-DOT) The system allows the 3D mapping of absorption and scattering in diffusive media. Making use of the time-resolved technique it is possible to discriminate absorption and scattering. The system embeds two DMDs for both structured illumination and single-pixel detection. Multi-spectral Fluorescence Lifetime Microscopy (λ-FLIM) The system allows for the multidimensional mapping of the fluorescence emission from biological samples on the microscopy scale. The system embeds a DMD for structured illumination and different output paths for exploring both multi-spectral and high-throughput time-resolved single-pixel imaging in collaboration with the Department of Electronics @ Politecnico di Milano.
