Êtes-vous un étudiant de l'EPFL à la recherche d'un projet de semestre?
Travaillez avec nous sur des projets en science des données et en visualisation, et déployez votre projet sous forme d'application sur Graph Search.
Fluorescence lifetime imaging microscopy (FLIM) is an imaging modality often used to monitor biochemical properties of a cell or a tissue. In addition to conventional fluorescence microscopy features, such as selective labeling and non-invasiveness, FLIM enables high background rejection and insensitivity to fluorophore concentration, tissue thickness and photobleaching. In this context, the phasor method has recently gained popularity in FLIM owing to its simple arithmetic operations and graphical representation. However, its widefield implementation has been mostly limited to systems that are expensive, fragile, or of limited flexibility. This thesis aims to overcome these challenges using SPAD technology by exploiting its single-photon sensitivity and CMOS integration capability. To achieve this goal, two large-format time-resolved SPAD imagers were designed. Time gating was selected over time-stamping due to simpler pixel implementation and higher photon-count rates, along with high spatial resolution. Of the two imagers, the first, SwissSPAD2, employs a 512×512 SPAD array, which generates photon-counting binary frames at up to 97,700 frames-per-second. SwissSPAD2 features a single time gate, which can reach a minimum width of 10.8 ns separated by steps of 17.9 ps. The SPAD in each pixel achieves one of the best photon detection probabilities (up to ~50% at 520 nm) and dark count rates (median of 7.5 cps) among current standard CMOS SPADs. The pixel native fill factor is 10.5%; microlenses were deposited on the pixel array achieving a concentration factor of up to 4.2 at the optimal photon angle of incidence. This resulted in an effective fill factor of 44.1%. The second imager, SwissSPAD3, comprises 500×500 SPAD pixels. Each pixel exploits two contiguous parallel time gates, allowing 100% duty cycle. It also has a minimum gate width of 1 ns, which, to the best of our knowledge, is the lowest gate width achieved by a large-format SPAD imager to date. The microlensed version of SwissSPAD3 is also available; its characterization is expected to yield similar concentration factors as for SwissSPAD2. Using SwissSPAD2, widefield phasor-FLIM was demonstrated with a time-gated large-format SPAD imager for the first time. It was shown that the camera is able to distinguish two single-exponential lifetimes with 1.4 ns difference using as low as 16 gate positions, corresponding to an equivalent acquisition frame rate of 12.1 fps. Moreover, FLIM-FRET analysis capabilities were validated by a double-exponential mixture analysis. In this experiment, the relation between the phasor ratios of five mixtures of two single-exponential fluorescence dyes and their respective volume fractions were consistent with expectations. The photon economy was found to be mainly determined by the overall temporal resolution of the photon time of arrival, and is largely independent of the detected photon count. This finding indicates that the gate width does not pose a fundamental limitation to the minimum detectable lifetime, allowing the estimation and separation of lifetimes that are much shorter than the gate width. This thesis demonstrates that time-gated SPAD imagers are competitive detectors for widefield FLIM. Future steps, such as integrated data processing, pixel pitch miniaturization, increased number of shorter gate channels, and microlens optimization are expected to further exploit the potential of SPAD technology.
Edoardo Charbon, Francesco Piro, Ashish Sharma
Edoardo Charbon, Pouyan Keshavarzian, Jiuxuan Zhao, Francesco Gramuglia, Myung Jae Lee
Edoardo Charbon, Claudio Bruschini, Arin Can Ülkü, Yichen Feng