Help Needed -- Near Infrared Intensity Sensor

[leej12]

Hello,

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?

Display: I want to display the results on an LCD and possibly want to show a intensity graphic similar to a heat map that updates in real time. What LCD should I choose and how would I go about creating this? Would it be easier to use Bluetooth and communicate the data to an Android device or website rather than use an LCD?

Other Questions
1) Will using a lux converter create noise that can not be accounted for? IE Is it better to take the analog signal directly from the sensor and then convert it to digital thorugh code or does it not make a difference using hardware components to convert analog to digital?Please give as much detail as possible as I am quite new to everything! Specific procedure and components would be amazing!

 

[tech99]

Hello,

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?

Display: I want to display the results on an LCD and possibly want to show a intensity graphic similar to a heat map that updates in real time. What LCD should I choose and how would I go about creating this? Would it be easier to use Bluetooth and communicate the data to an Android device or website rather than use an LCD?

Other Questions
1) Will using a lux converter create noise that can not be accounted for? IE Is it better to take the analog signal directly from the sensor and then convert it to digital thorugh code or does it not make a difference using hardware components to convert analog to digital?Please give as much detail as possible as I am quite new to everything! Specific procedure and components would be amazing!

There are some standard circuits for using a photo diode with an op amp, and special low noise op amps for the task I believe. The usual circuit is to place the diode in the feedback path.
How will you prevent your diode responding to the incident 670-702nm radiation?

 

[leej12]

There are some standard circuits for using a photo diode with an op amp, and special low noise op amps for the task I believe. The usual circuit is to place the diode in the feedback path.
How will you prevent your diode responding to the incident 670-702nm radiation?

Will a band pass filter work? If so, how much will the emitted wavelength have to be? In addition, can you "remove" noise by accounting for it through modelling? Or will it still affect measurement regardless?

 

[Nidum]

Google ' Thermal imaging near infrared '

 

[berkeman]

My goal is to create a sensor that will be able to detect this emitted wavelength through human tissue

Human tissue is transparent to this wavelength? What is the attenuation coefficient for this wavelength in human tissue. What depth of tissue are you trying to detect this through?

 

[leej12]

Human tissue is transparent to this wavelength? What is the attenuation coefficient for this wavelength in human tissue. What depth of tissue are you trying to detect this through?

The extinction coefficient is 96000. The depth is 2.5cm

 

[berkeman]

The extinction coefficient is 96000. The depth is 2.5cm

What are the units for that extinction coefficient? You are trying to detect IR through an inch of tissue?

 

[leej12]

What are the units for that extinction coefficient? You are trying to detect IR through an inch of tissue?

I should have clarified, the molar absorption coefficient is 96000 M−1⋅cm−1.
Yes, will it be too difficult? In-vivo near infrared imaging has been done and all we need is a relative measurement of intensity.

 

[Jeff Rosenbury]

This is a complex technical problem. There are several areas that need to be addressed. Human tissue is one, infrared optics another, and the hardware a third. There's enough here for several credit hour classes. Asking for everything is a bit to much for this forum. (That doesn't even address the ethics, medical, or biological issues.)

Perhaps you have some more specific questions?

Good luck.

 

[leej12]

Yes, what I would like to know the most is the general hardware that I will need for this IR sensor. Such as what type of photodiode would be most suited, op amps, digital to analog converters, etc.

 

[Jeff Rosenbury]

You will apply the sensor for how much time? Do you need a reading every second? minute? microsecond?

How much light are we talking about. This will depend on both the amount generated and how much is transmitted through the flesh.

Both the measurement time (bandwidth) and amount of light are key factors in part selection. Frequency of the light is also a consideration. How the part will be used also plays a minor role. (Will the detector and first amplifier stage be co-located? Ideally yes.)

So the problem isn't a simple, "This part is what you want." question. What you need depends on what you're measuring.

But basically you will need a photodetector (possibly a diode?), an amplifier (an op-amp/OTA for reasonable measurement bandwidths), an ADC (analog to digital converter), a medical grade processor (i.e. software on it can be verified for human use.) and a Bluetooth transceiver. Unless you need a photomultiplier (which I know little about) it should all fit in a handheld case with battery power.

 

[leej12]

You will apply the sensor for how much time? Do you need a reading every second? minute? microsecond?

Ideally I would like as many readings within the time period that the device is on. I think that this would allow for an average of the readings and therefore higher accuracy, correct?

How much light are we talking about. This will depend on both the amount generated and how much is transmitted through the flesh.

At this stage, I do not know how much light will be emitted and attenuated through tissue. We will be doing testing to determine this. But I'm hopeful that a photomultiplier will not have to be used and enough light will be able to reach a economical photodiode.

Both the measurement time (bandwidth) and amount of light are key factors in part selection. Frequency of the light is also a consideration. How the part will be used also plays a minor role. (Will the detector and first amplifier stage be co-located? Ideally yes.)

We will be doing testing to determine the frequency, but do not know at this moment.

Will an Arduino be able to handle the processing? What resources are there to learn about all the things I will need to know?

 

[berkeman]

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).

BTW, if you are illuminating at the same time you are trying to detect, you will probably need to use some pretty sharp optical filters to reject the illumination light at the sensor. Alternately, can you pulse the illumination and detect the residual emitted light? What is the expected time constant of the light emitted from the cells after the illumination is turned off?

 

[leej12]

BTW, if you are illuminating at the same time you are trying to detect, you will probably need to use some pretty sharp optical filters to reject the illumination light at the sensor. Alternately, can you pulse the illumination and detect the residual emitted light? What is the expected time constant of the light emitted from the cells after the illumination is turned off?

We have yet to determine the time constant, but the emission has a half life of 4-5 hours. Preferably it would be easier to use some sort of band pass filter, but if the difference between the two wavelengths is too small this would be difficult.

 

[berkeman]

emission has a half life of 4-5 hours.

Pfftt. With that long time constant, illuminate for however long, then turn off the illumination and gather the emission data. Do it in a dark room...

 

[Jeff Rosenbury]

BTW, if you are illuminating at the same time you are trying to detect, you will probably need to use some pretty sharp optical filters to reject the illumination light at the sensor. Alternately, can you pulse the illumination and detect the residual emitted light? What is the expected time constant of the light emitted from the cells after the illumination is turned off?

I guess I thought they were modifying the cells to emit IR. I know this is possible, though other methods might be used.

If this isn't true, let me know Lee.

Ideally I would like as many readings within the time period that the device is on. I think that this would allow for an average of the readings and therefore higher accuracy, correct?

At this stage, I do not know how much light will be emitted and attenuated through tissue. We will be doing testing to determine this. But I'm hopeful that a photomultiplier will not have to be used and enough light will be able to reach a economical photodiode.

We will be doing testing to determine the frequency, but do not know at this moment.

Will an Arduino be able to handle the processing? What resources are there to learn about all the things I will need to know?

No, more detections will not increase accuracy. This is because the noise also scales with the time window. Ideally you filter the noise band to match the bandwidth of your signal. Longer is better (up to a point of course). If you don't need sub-second timing, you might take about 10 samples a second which is easily filterable (for the noise).

Berkman's idea of pulsing light at the cells might have merit even if you intended them to self-glow. The same structures that cause the glow might respond if pulsed. This in turn might increase the signal to noise ratio.

He is also right about the optical filtering, but that's outside the strict scope of the electronics. It will affect design choices such as light frequency and detector selection though.

An Arduino likely has the raw power for the job, but I question the ethics of using a hobby kit when someone's life might be at stake. (Even a misdiagnosis due to a computing error might be a problem.) Medical electronics design is a specialized field and they don't like shortcuts. Still, it might work for a preprototype to prove funding. There are likely ADCs for the Arduino. Hopefully the conversion will be straightforward and you can COTs it (Commercial Off the Shelf).

You might read some communications theory. You need to know what signal to noise ratio means on an intimate level. All the designs you might want should be online, but understanding them will require some background in analog design. The Art of Electronics is a good guide.

I know you want to do this with as little cost as possible, but Electric Engineering is a complex endeavor and Electrical Engineers deserve to be paid like everyone else. You might consider hiring one.

 

[leej12]

I guess I thought they were modifying the cells to emit IR. I know this is possible, though other methods might be used.

Yes, this is what we are doing.

Thank you for your advice, Jeff. I will look into everyone's suggested topics and texts and continue diving into this project.

 

[mfb]

4-5 hours of half-life are problematic for other reasons: the light intensity will be low, because you spread your emission over such a long timespan. I would expect that you need some control reading to estimate the background, then an excitation, and then another reading to see the signal. Ideally the measurement device is not moved at all during that cycle, so it cannot be too long.

A specific design will need at least some rough estimate of the signal and the background you'll get.

 

[Jeff Rosenbury]

4-5 hours of half-life are problematic for other reasons: the light intensity will be low, because you spread your emission over such a long timespan. I would expect that you need some control reading to estimate the background, then an excitation, and then another reading to see the signal. Ideally the measurement device is not moved at all during that cycle, so it cannot be too long.

A specific design will need at least some rough estimate of the signal and the background you'll get.

That depends on the light level. If it is faint, you are correct. If it glows like a light bulb, not so much. I've seen glowing fish which would seem to be detectable in a dark room. I assumed it was something like that.

 

[mfb]

4-5 hours are about 15000 seconds. With 100% conversion efficiency and no losses, which is horribly unrealistic, 10 seconds of illumination would lead to an emission of 1/1500 of the illumination. Add an order of magnitude for the detector area, an order of magnitude for absorption on the way, probably much more for conversion efficiency, ...