Hey everyone,

I'm new here and to the electronic side as well. So I'm pretty clueless.

I need to design a circuit using thermistor to measure the temperature distribution on skin surface from the throat region.
I need accuracy greater than even 0.1 degrees.

I have no idea as to how to proceed or where to make a start.

Any help would be appreciated.
thank you

 

I need accuracy greater than even 0.1 degrees.

That's really tough. Do you possibly mean precision (ability to reliably detect a small difference) and not accuracy (readings against a calibrated standard)? Precision is still tough but doable. That level of accuracy will add a whole new layer of difficulty.

I think one of the biggest challenges at that level of precision will be eliminating ambient effects.

Many thermometer projects start with a LM35 IC. It's worth looking at, but may not be enough for this project. IMHO, I'd forget about using a thermistor for a project this challenging. Calibrated platinum RTD probes might work.

Have you considered IR thermal imaging?

 

I like thermistors because they are small and respond quickly but, in this case, they also respond quickly to minor drafts of air. Not good.

An infrared camera can do 10,000 measurements in a few milliseconds. Much better than moving a thermistor around on a person for several minutes.

 

Thermistors are used on most of the typical medical grade skin temperature probes.. The best have .1deg C accuracy (never seen one with better than that).
Others feature thermocouples.

Any reason to not use the ones already out there with NIST/FDA certification already?
All I've seen do stop at .1 deg C.. You say greater than..But haven't defined how much greater than and you really need to.
Are you aware of the costs for obtaining medical certifications?

What response time do you require?

Have you talked to Omega? (you should)

 

It is a class of PT100 elements named "tenth-DIN sensors" they have accuracy down to (±0.03 °C). A long time ago I used such sensors for measurements in fluid. Using a so called immersed temperature sensor and a lock-in amplifier. However with such temperature accuracy it is a lot of inherent problems. Kind of glad I do not have to deal with it

 

The trouble with taking temperature readings with physical things is that the physical thing effects the temperature of the other thing.

The smallest temperature-information-bearing device would be best; a remote device would be better still.

I vote for infrared imaging.

Also, as an aside, an optical mouse contains a small CCD camera (usually around 300 pixels total). It is sensitive to both visible and infrared, to an extent. You could theoretically focus the light onto the sensor, dump the internal data pixel map, and use this scheme to calibrate-then-measure sections of skin. It's like an infrared camera, except it's a helluva lot cheaper than an infrared camera camera (2 x camera on purpose).

Just my 2¢.

 

Thank you so much for your replies, I'll definitely research along these lines.

Thermal imaging is definitely the best option but it's also very expensive. My project does not allow for such expensive technology to be used.

What I basically need is to measure the temperature and set a standard for normal people. And then take reading on diseased people and note the temperature variations.

So I need accuracy in terms of the standard reading and precision to note even minute changes.

I was told that thermistor would be one of my better options to achieve my target.

Also, I made a mistake. I don't need greater than 0.1. 0.1 degree should be good

 

What I basically need is to measure the temperature and set a standard for normal people.

Good luck with that. Have you noticed that there is no such thing as a "normal" person? There is going to be so much natural variation that you're going to have to measure a LOT of people, and you'll never know for one individual if they are hot because they're sick or hot because they're an outlier in the population.

Imagine you're measuring height, something easy to measure accurately with great precision. You could compare two populations, say men versus women, and you'll get two overlapping but slightly offset bell curves, each with a mean and a standard deviation of maybe 6 inches or more. With enough measurements of both groups, you could establish with statistical significance a 6" difference in mean height between men and women.

But you'll NEVER get to the point you could judge someone's sex by measuring just their height. You could improve the odds of your estimate to be better than 50:50, but that's it.

With facial temperature, add in the measurement difficulty and the variation that you'll see in even a single person (are they flushed because they're hot for the nurse?), and you've got a real challenge to extract meaningful signal out of the noise.

I like the idea of an imaging camera spotting hot people in a crowd though. I'm a big fan of quarantine! Infectious disease kills a lot more people than war.

 

Good luck with that. Have you noticed that there is no such thing as a "normal" person? There is going to be so much natural variation that you're going to have to measure a LOT of people, and you'll never know for one individual if they are hot because they're sick or hot because they're an outlier in the population.

Imagine you're measuring height, something easy to measure accurately with great precision. You could compare two populations, say men versus women, and you'll get two overlapping but slightly offset bell curves, each with a mean and a standard deviation of maybe 6 inches or more. With enough measurements of both groups, you could establish with statistical significance a 6" difference in mean height between men and women.

But you'll NEVER get to the point you could judge someone's sex by measuring just their height. You could improve the odds of your estimate to be better than 50:50, but that's it.

With facial temperature, add in the measurement difficulty and the variation that you'll see in even a single person (are they flushed because they're hot for the nurse?), and you've got a real challenge to extract meaningful signal out of the noise.

I like the idea of an imaging camera spotting hot people in a crowd though. I'm a big fan of quarantine! Infectious disease kills a lot more people than war.

You do make a fair point but it's a non optional project. I have to give it a go.
I'll let you know how well it works out (or doesn't)
Sigh.

 

Do you have a hypothesis you're testing? I mean, something like "hot nose with cool cheeks implies imminent death"? I really have no knowledge of this but I imagine a temperature DIFFERENCE between facial areas might be more meaningful than absolute temperature. That'd be lucky because I suspect it would be far easier to detect.

That earlier suggestion to learn from the folks at Omega is good advice. Industry leaders.

 

If you decide to go with contact sensors...thermistor...thermocouple. How are you going to attach them? Attaching "things" to the skin can cause localized changes in skin temp at that point. If you are concerned about 0.1°C variations, you may be actually causing them.

From you description, thermal imaging seems the only logical way. Have you thought of renting the equipment? There are a lot of companies out there that do that.

Ken

 

...thermal imaging seems the only logical way.

+1
A clever solution to get the cost of that down would be a better investment than trying to avoid it. Camcorders can "see" IR. Maybe all you need is a filter and a software solution to process the image. An old camcorder and a computer would be nearly free.