Best Sensor for Near Infrared Intensity Detection Through Human Tissue

I am part of a synthetic biology team whose goal is to modify immune cells to respond to an incident wavelength of 670-702nm and emit a near-infrared wavelength of 650-900nm(to be determined). 

My goal is to create a sensor that will be able to detect this emitted wavelength through human tissue and convert the data from the sensor to the si unit lux of intensity. The sensor should be as accurate as possible.

Everything I know is from simple googling so any help would be appreciated!!!


Sensor Selection: I am thinking of using a photodiode, assuming that the emitted wavelength is strong enough to detect.I assume I have to use an op amp. If the emitted wavelength is too weak from attenuation through the tissue, what options do I have? An avalanche photodiode? 

Look at the sensitivity curves for solid state diode detectors, they peak in the red.
When you say the sensor should be as accurate as possible, what do you mean, in absolute terms, or relative. How do you propose to calibrate the whole system, light measurement is at best complicated, some quite specialized equipment is needed if you want absolute values. Lux meters use integrating spheres, and different coatings are used depending on the wavelengths of interest. I am not sure how in your circumstance a calibrated lux meter would be of much value.
Is the device in contact in such a way that ambient light will not be a problem.
If not, the use of optical filters can help. Modulated light and synchronous detection is often useful, since you can have high gain with AC coupling in the preamplifier stages. It may even be possible to put the whole system in a feedback loop, controlling the source intensity to get constant illumination at the detector, which should give very good linearity. Look up the articles by Bob Pease on trans-impedance amplifiers. Avalanche diodes are a possibility, but you need to be aware that their noise performance is atrocious at high gains. Best way to get a good S/N is to use the brightest source that is possible.
Just a few thoughts, I could be barking up the wrong tree,
cheers,
Richard

Thank you for your response Richard,
It should be relative. I simply want to determine the intensity of the wavelength emitted by a certain amount of immune cells and see the increase in intensity when there is more immune cells. We will be using a band pass filter to block out unwanted wavelengths.