Our group does a lot of work with probe and wide-field quantification algorithms for neurosurgical guidance. In brief, we have found that our reflectance analysis models are failing to decouple scatter from absorption for in vivo measurements of brain. What we tend to see is the μs′ , modeled as μs′ = a(λ/λ0)-b , scatter slope (b) estimated as a near zero or negative value, resulting in either a flat or increasing with increasing wavelength scatter profile. I have come to suspect that we may be missing an absorber (i.e. oxy- and deoxyhemoglobin are not the only significant chromophores). My main suspects are melanin or endogenous fluorophores like NADH, FAD, and collagen.
Now here is where my problem lies. I have made simulated spectra to explore the effects of having melanin or other fluorophores present but not fitting for them and found that melanin produces a lot of similarities to measured in vivo data, including negative estimates of b when not fit. Unfortunately, the spectrum of melanin is featureless over the entire visible region and produces cross-talk with the equally featureless scatter profile. Similarly with the endogenous markers, I found some papers that have used a generic exponential of the form Ae(-Bλ/λ0) to account for their combined aborption but it is also featureless. Is there a good way to prevent cross-talk between a featureless aborber and scatter in the visible region? We have data between 450 nm - 700 nm.
Thanks!
Fit your curve to an equation consisting of log-normal function and a scattering function should help you to separate out the contributions. Check if this helps.
V. N. Ravi Kishore V., Thank you for your response but could you clarify a bit more what you mean? Is the attached link what you are referring to with a log-normal function? If so, it is not much different than the exponential decay component I have already tried.
https://en.wikipedia.org/wiki/Log-normal_distribution
I am asking it to fit your data to a function which is an addition of two different functions. One function accounts for the absorption and the other accounts for the scattering. Generally absorption spectra of organic molecules are asymmetrical and hence are fit to a log-normal function. I hope this helps.
Sorry for the delayed response! OK, I see what you are saying. I beloeve I am already doing that, just in a different manner. What I don't have in the description is the absorption coefficient (μa), defined as
μa = BVF*(StO2*εaoxyHb(λ) + (1 − StO2)*εadeoxyHb(λ)) + MVF*εamelanin(λ),
where BVF is blood volume fraction, MVF is melanin volume fraction, StO2 is oxygen saturation, and εaoxyHb, εadeoxyHb, and εamelanin are the wavelength-specific absorption coefficients [mm-1] for oxyhemoglobin, deoxyhemoglobin, and melanin (if used), respectively. While my μa and μs' are different functions they are fit as inputs into either a diffusion theory model or Monte Carlo look-up-table and are therefore fit simultaneously. If I swap the melanin component for a generic exponential Ae-Bλ/λ0, where A and B are fit parameters, it closely resembles a lognormal function (eμeσZ). Is this different from what you are recommending?
The expression you are using takes care of the absorption contribution due to different components present in the sample. It does not consider the losses due to scattering. So you need to add an additional term which takes care of scattering.
The μs′ function accounts for that (i.e. μs′ = a(λ/λ0)-b). We then consider reflectance to be a function of μa, μs′, and rho, the source-detector separation (e.g. R = f(μa,μs′,rho))
What type of functional dependence are you considering? Coz even that is very important.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Physical Phenomena
- Luminescence
- Fluorescence