Penso que o LED é mais adequado de vido ao seu custo, largura de espectro, vida útil e custo da fonte de alimentação, além de aceitar um circuito de driver mais flexivel (menos crítico).

I think you should to make sure the excitation and emission spectra  of your fluorescence. According to the spectra, we can choose the  LED with the wavelength to maximum the signal-noise ratio. In fact, it is not difficult to purchase the LEDs with output power higher than 3-5 W with the wavelength ranges of 450-470nm, 525-535nm and 610-620 nm.

Dear Casper,

I agree with Shang, i think one has to specify the excitation wavelengths of the molecules labeled with fluorescence and the emitted radiation wavelengths. You have also the maximum allowed intensity allowed to strike such molecules. Laser may adversely affect the samples as it is strong radiation.

I think laser may less suitable for such imaging applications. One may use power led lamps. If you can describe the conventional imaging system, one can find the suitable substitution. 

From the practical point of view, one can try different sources and use the most suitable one since these sources are not expensive.

Best wishes