Development of a Multichannel TCSPC System in a Spartan 6 FPGA LinoSPAD - Fluorescence Lifetime Imaging for Fluorescence Guided Surgery

Development of a Multichannel TCSPC System in a Spartan 6 FPGA LinoSPAD - Fluorescence Lifetime Imaging for Fluorescence Guided Surgery

Homulle, H.A.R.

Charbon, E. (mentor)
Powolny, F. (mentor)

Electrical Engineering, Mathematics and Computer Science

Microelectronics & Computer Engineering

Circuits & Systems

2014-07-04

For the master project work was carried out for the development of a fluorescence lifetime imaging probe for fluorescence guided surgery. For this project a prototype was designed. The work on the prototype was divided into three main parts, hardware, firmware / software, and system / optics. In this thesis the firmware / software of the system are described. An overview of the system is given and the performance is evaluated. The systems hardware consists of a mainboard with a Spartan 6 FPGA and a Cypress FX3 USB3 controller. Secondly a daughterboard was designed which houses the LinoSPAD chip, a line of 256 SPADs. The outputs of the SPADs are directly connected to the FPGA. The FPGAs firmware was developed. The system had to incorporate a Time Correlated Single Photon Counting structure for the 256 channels. TCSPC calculates the timing of individual photons and stores this information in a histogram. The Time to Digital Converters were implemented in the carrychain structure on the FPGA. A clock of 100 MHz is used, therefore 512 stages (128 Carry4 blocks) form the delayline of the TDCs. The TDC range is > 10 ns to fill the complete clock period. The histograms were stored in the FPGA RAM memory blocks. The final firmware has 8 TDCs, each having 32 channels. Those channels are connected to one TDC. Each channel has its own RAM memory in which the histogram is stored. The memories can be read-out in less than 2 ms over USB3. The performance of the TDCs was assessed. The resolution is 20-21 ps and the non linearities are DNL 4 LSB and INL 7.5 LSB. The results were confirmed both in simulation and in the physical design. Crosstalk from SPADs was removed by not storing timing information when multiple SPADs fire in one clock period. Secondly the systems software was developed. The software has a user-friendly GUI in which the main aspects of the fluorescence signals can be plotted. Both the raw TCSPC histograms and the extracted information as intensity and lifetime can be shown. In order to extract the fluorescence lifetime, two algorithms were studied and compared: Wiener filter and Centre of Mass method (CoM). The Wiener filter is based on deconvolution of the measured signal with the Instrument Response Function of the system (IRF). On the deconvoluted exponential, a fit is made to extract the lifetime. The CoM uses the centre of mass of the fluorescence histogram and the centre of mass of the IRF to make an estimation of the lifetime. Both algorithms show similar results in simulation with errors up to 30 ps over a lifetime range from 100 ps up to 1 ns. These errors are obtained for a histogram with 2500 photon counts, noise and systems non linearities being present. However the Centre of Mass method is up to 100x faster compared to the Wiener filter due to its lower complexity. The FluoCAM camera system, already developed in a previous project, was calibrated. It was also used to perform biological experiments in-vivo. Findings of fluorescence lifetimes are comparable to literature for well-known fluorescence compounds. Furthermore first tests in vivo (mice) reveal similar lifetimes over several days. The LinoSPAD system could not be tested for fluorescence as the LinoSPAD chip is not working. Instead a fluorescence emulator was designed in a second FPGA to test the LinoSPAD FPGA system. The emulator mimics fluorescence exponential signals, however due to FPGA limitations the derived exponentials are not analysable to extract lifetimes. Besides an emulator, some experiments were done with a single SPAD. The systems resolution and non linearities were confirmed using SPAD noise density tests. The noise rate of the SPAD was found in the same order as expected (100 Hz). Finally with the data collected from FluoCAM, the performance achievable with the LinoSPAD camera system was estimated. Main system requirements have been fulfilled during the project. However the system has to be tested with the working LinoSPAD chip to confirm the usability of the complete system. The first step towards a fluorescence probe has been set.

FLIM
surgery
lifetime
fluorescence
ICG
FPGA
TCSPC
TDC

http://resolver.tudelft.nl/uuid:86ecbaba-0711-40e8-8b10-1001b3772206

Student theses

master thesis

(c) 2014 Homulle, H.A.R.