Date: Mon, 29 Sep 1997 14:58:57 +0000

To: lodder@pop.uky.edu

From: Neil Lewis <neil@spy.niddk.nih.gov>

Subject: Manuscript

Bob...

I've finally had an opportunity to look over the manuscript!! It looks

good.. I particularly like the application of the TED and FRED algorithms

to reduce spectral artifacts from subject motion...

Look over the following comments/questions though and let me know if they

make any sense....

1) What is known about the correlation between surface temperature and

lipid metabolism? Obviously there is some correlation, but what about

situations where there is a reduction in peripheral blood flow but a

commensurate increase in metabolism. Adrenalin, for example, reduces

peripheral circulation. Could this lead to the recording of surface

temperature measurements that were not as well correlated to metabolism?

Is there any way to record something like rate of weight loss over the two

week period to compare to the increased temperature measurements?

*** Surface temperature and metabolic rate are correlated. This is a published observation in rats and humans (I will cite the references in the revised paper since you mentioned it), but is also easy to understand when you consider that the heat generated by increased metabolism must flow somewhere, even in the presence of peripheral vasoconstriction. In rats, for example, an increase in metabolic rate raises core temperature. Thermoregulatory mechanisms will prevent this core temperature from rising to dangerous levels. These thermoregulatory mechanisms must remove the heat by conduction, convection (including losses through the latent heat of vaporization of water), and radiation even in the presence of vasoconstriction. Accurate metabolic measurements in the rat must be made using cameras attached to a metabolic calorimetry cage, which permits heat losses through conduction and convection to be measured along with radiation. In this way, interferences resulting from vasoconstriction can be removed. If this information is not in the paper, it should be (probably in the introduction, don't you think?).

It should be noted that the largest thermoregulatory organ in the rat is the tail. Acute administration of agents that raise core temperature a few degrees result in tail temperature increases of 20 degrees within a few minutes. This has been documented by IR imaging as well as in the literature by more conventional temperature measurements. Perhaps this should be noted in the Introduction to the paper.

We have recorded weight losses in rats correlated to changes in temperature measurements. These more extensive experiments are described in a paper submitted to the Journal of Clinical Investigation. Do you think that we should mention that in the paper?

We have also done experiments using hydralazine administration to counteract the effect of vasoconstriction in angiotensin-II administration, and assessed the difference in surface temperatures in the presence and absence of vasoconstriction. These experiments, too, are in the paper submitted to the Journal of Clinical Investigation.

2) Not that it's particularly important for this manuscript, but how many

animals did you look at? Statistics for physical types means recording lots

of data on the same subject, whereas for the bio types it's recording data

on lots of subjects?

*** In these experiments we scanned n=10 animals (5 control and 5 test), with at least 256 video frames recorded at each wavelength. Blackbody images were also obtained from at least 256 collected frames. I will make sure that this gets into the revised manuscript as well.

3) In your conclusions you indicate that you are able to monitor lipid

metabolism in the ISF with the IR camera. Is this true? I wasn't sure from

reading the paper how the NIR spectra of the rats were recorded. Whether it

was the Bran+Luebbe system or the camera with filters. If it was the

camera then that's great, because these measurements be simultaneously

correlated with the increase in surface temperature.

*** We have measured lipids in the ISF with the camera and with a conventional spectrometer and fiber-optic probe (the ISF is right under the skin so it is pretty accessible in reflectance mode). Either way, when we attempt to measure spectra of the ISF, we shave a tiny spot to eliminate interference from hair. The aim, naturally, is to get temperature measurements nearly simultanously with spectral images. We can't usually collect both NIR and IR images simultaneously because the broadband near-IR light source skews the observed blackbody light, unless we employ a second camera with a 4.7 micrometer narrow-band cold filter, and narrow-band near-IR illumination (e.g., laser). When you do this it is best to inject the light into the ISF by a touching it with a fiber-optic probe, to keep the light intensities recorded by a camera from the skin inside a reasonable dynamic range. This would probably be good to mention in the Conclusion.

Talk to you later... Neil

Neil Lewis: National Institutes of Health

neil@spy.niddk.nih.gov

I enjoyed your paper regarding "Near-IR and Imaging in Lipid Metabolism

and Obesity". My comments are relatively minor.

1. In the Materials and Methods section:

a) The first two sentences appear to be somewhat redundant.

*** OK. One sentence will do.

b) What is the significance of the sound annotation capability of

the camera? I see no mention of your using this feature.

*** As data collection from the cameras becomes faster and faster (and in fact, it is now free-running, real-time video capture), it becomes increasingly important to be able to index the frames for analysis later (valuable indices include facts like subject ID number, right ear or left ear, changes in camera settings made to keep signal intensities near the middle of the A/D range, etc.). Voice annotation is fast enough to annotate spectral frames from freely moving subjects, while collecting frames / typing comments / collecting frames is not. This can be mentioned in the experimental section.

c) Where did the full spectra from a B+L instrument get used in this

experiment?

*** Full spectra were used only in reference scans of materials like the black and white optical reference standard spheres (whose scans are not shown in the manuscript).

d) Have you used the FRED algorithm in this study or another. Unless

you can compare actual performance with TED algorithm I think you

would be better off to leave out any discussion of FRED.

*** FRED is an obvious subset of TED that showed similar performance characteristics on our thermal image data (i.e., TED thermal images were qualitatively indistinguishable from FRED thermal images). For this reason, the data were not shown. However, I agree that a sentence stating this fact should be added.

e) What is the "standard software based on Stefan's Law"?

*** You can buy radiometric software from the camera manufacturers to measure surface temperatures.

2. In the last paragraph of the Results and Discussion you discuss

the changes in major lipid peaks after administration of two

drugs. How were these NIR spectra collected?

*** Variable-wavelength near-IR light was "injected" into the fat pads using a fiber-optic probe.

Do the spectral changes correlate with expected physiological changes?

*** Yes, for the prototype compound (norepinephrine)(see below). The lipid mobilization observed when angiotensin-II is administered is a unique observation.

What biochemical/physiological changes may describe the particular

spectral phenomena that are witnessed?

*** Norepinephrine causes lipid mobilization from fat pads (saturated and unsaturated lipids equally) to act as fuel for the fight-or-flight response. Angiotensin appears to mobilize first lipids, then lipids. The reason(s) for the differential mobilization are not yet clear.

Generally, I find the paper to be novel and of sound method. The paper should be accepted with minor revision regarding the clarification and interpretation of several issues that I have listed above.

Best Regards,

Jim Drennen

drennen@nir.pharm.duq.edu