Recently I've been spending most of my time working on my master research project: 'Fluorescence Lifetime Imaging Microscopy with Compressive Sensing'. I thought I'd write about it while all the knowledge is still fresh.
What are we trying to do?
Ultimately, the goal is to develop an imaging system that's fast and clear and non-invasive that's useful in cancer diagnosis and even during operation.
Traditional Fluorescence imaging is very fast, you shine a light onto the sample and pretty much immediately you can see the fluorescence. So why do we bother to go through the trouble?
There's something that's quite useful in fluorescence sensing, it's a parameter called 'quenching rate'. It can measure the 'micro-environment' around the cell e.g. temperature or pH. Traditional fluorescence is not very good at it because it measures intensity, which can also be effected by local concentration.
This is why we use lifetime, because lifetime of the fluorophore is independent of the concentration.
There are many ways to measure lifetime, the most prominent technique is called TCSPC (Time Correlated Single Photon Counting). TCSPC works by exciting the fluorophore for a second, then turn the illumination off, count which photon came at what time, and eventually the distribution of photons emitted correlating to time will show a decay curve, then the lifetime can be worked out.
But this process is very slow. TCSPC can only count the photons pixel by pixel. So if you want a 64*64 image, you'll have to do the same thing 4096 times. This normally comes out at timesacle of minutes for an image to be acquired, which can be problematic for living cells, as they move all the time.
This is the problem we are trying to overcome. And our solution is to achieve a faster rate through compressive sensing.
So what is compressive sensing? In a few words, it takes less pixels than needed and fill the gap by using some algorithms to reconstruct the image. By taking less pixels we are effectively saving a lot of time.