IMPROVEMENT OF CROSS-CORRELATION PHASE COUNTING METHOD FOR DIGITAL FREQUENCY DOMAIN FLUORESCENCE LIFETIME IMAGING MICROSCOPY by Zhang Baoning Thesis Supervisor: Dr. Sio Hang PUN Thesis Co-supervisor: Dr. Peng Un MAK Department of Electrical and Computer Engineering Commonly, the differences in the exponential decay rate of the fluorescence from a fluorescent specimen are very important for constructing an image by Fluorescence lifetime imaging microscopy (FLIM). FLIM can be used in Laser Scanning Confocal Microscopy (LSCM) or Camera-based Microscopy. By contrast with conventional confocal intensity imaging technique, fluorescence lifetime image is more colorful and effective especially in minimizing photon scattering for thick-layer samples. According to Jablonski diagram, fluorescence lifetime is the average time when electron energy decreases from the fluorophore’s excitation state to return to the ground state, when emission of a photon happens simultaneously. Fluorescence lifetime imaging acquisition can be attained by time-domain or by frequency domain techniques. A classical time-domain fluorescence lifetime acquisition method is Time-correlated single photon counting (TCSPC), which single photons of a fluorescence signal are precisely transmitting in repetition. On the other hand, for frequency domain techniques, the phase shift between the emission of the sample and the modulated or pulsed excitation is used at the fundamental or harmonic modulation frequency. For analogue frequency domain (AFD) technique, the excitation light source is modulated to a sinusoidal waveform, thus, the emission fluorescence is also sinusoidal with the same frequency but the different phase and relative modulation which is decided by fluorescence response. For digital frequency domain (DFD) technique, A novel method based on Field-Programmable Gate Array (FPGA) implementation was firstly proposed in 2008. The DFD-FLIM can provide high-quality lifetime resolution comparable to TCSPC methods, the cost of its electronics is relative much cheaper, its acquired data which are in the form of phasors make data calculation and analysis much simplified. Moreover, as the development of FPGA efficiency, its performance will make huge advances. Overall, this approach may make FLIM technology more widely available. However, the DFD-FLIM technique is just proposed, thus, the efficiency of system and the performance of implementation would be much improved. My topic is focused on FPGA implementation of DFD FLIM data acquisition by cross-correlation phase counting method. In this paper, I will introduce DFD FLIM theory and revised some mathematical modeling based on it. To verify the result of revised model, we firstly used Matlab to simulate this process and compared the results between four and eight sampling windows. And then I give the DFD-FLIM MATLAB simulation and DFD-FLIM FPGA implementation of data acquisition. And then we implemented both four and eight sample window system on FPGA, which exhibits that eight sample window performs better than four sample window.