almost done...

mDescape 2 years ago
parent 46a71b8930
commit 875682ed06
  1. BIN
  2. 63
  3. BIN

@ -60,3 +60,66 @@ the system parts. The spectrometer, the light sensor and the projector.
In this slide I'll give a brief explanation of the characteristics of each system's component. This is the
implementation made for the solution presented on the last slides.
About the stimuli generated by the system, they are a plain white image, of diferent intensities projected for less
than 0.25s.
The spectrometer takes measures in an automatized maner. It is controlled by a Visual Basic program which also
comunicates to the main script in order to take orders from it and send back the measures made.
Since the comunication between the programs is build under the TCP/IP protocol, the spectrometer and the computer
runing the main program could easily be separated by some distance, giving the system a bit more of flexibility.
The light sensor has actally two parts, the light sensor itself and a microcontroller, which digitalizes the
sensor's output, measures the time past between the start of each stimulus and the end of it, and send an order
to finish the experiment in case a stimulus is detected to be on screen more than 220ms, in order to prevent retinal
The main experiment program accomplish diferent tasks in order to be able to run the experiment.
Firstly it creates a Grafical User Interface in order to allow the researcher to enter the participants
information. Such information is necesary to keep track on the participants ID, the test number and if it is a
migraine patient or a control; in order to extract the results once the whole experiment is over.
It also is in charge of generating the diferent stimuli, presenting them according to the SIAM method and storing
the test results, to allow a later study of them.
Finally, it is also is aware of the information comming from the light sensor, in order to finish the test when
a hazardous stimulus is detected, in order to prevent retinal damage.
The stimuli generated by the projector are not static, they are associated whith an aleatory error which make its
illuminance change over time. In the worst case scenario caputred, the standard derivation of a stimulus is of only
6.3%, which is provably small enough to not be detected by the human eye.
When the system is calibrated, after 12h it keeps producing the desired stimuli precisely, within the same error
margin it produces them the first minutes after the calibration.
Because the projector the system uses is a LED based one, the light beam is Pulse With Modulated. Because of this,
to detectd when a stimulus is on screen or not, the light sensor needs to average the readings taken. When such
average is above 0.25V it interprets the stimulus is on screen and when it falls under 0.25V again, it interprets
it has finished.
In order to give the participants the instructions on how to perform the test, a screen is shown before the test
starts where the instructions can be read.
On this image we see the diferent readings made over 70 seconds (the blue circles) of the same stimulus.
The red line is the average of all the measures.
We can apreciate the furthes point is only 33 lux away from an average of 2513 lux.
We can also apreciate that the quantization error of the spectrometer in this range is of 10 lux
In this next image we see the transfer function from the image generator to the stimuli illuminance.
The method of creating this image is the following: For each posible stimulus, present it on screen and perform
10 measures. Averge all the results. Save such average and also the maximum and minimum illuminance values measured.
In this image, each circle is the average of one stimulus, and its error bars represents the maximum and minimum
illuminances found for it.
As can be seen, after the first 75 images the system has a liniar behaviour. It is also worth mentioning that the
first section of the calibration curve has a negligible aleatory error.