Embedded Files
We set out to prove three things with our proof-of-concept prototype:
1) Our device can detect blood.
2) The sensor and alarm units of our device can communicate wirelessly via Bluetooth.
3) Our device can sound an alarm when blood is detected.
We proved all three of these things through our testing!
Preliminary Color Testing
Preliminary Color Testing
During our first stage of testing, we investigated the color accuracy of our RGB sensor unit . To do this, we scanned 50+ different paint samples with our RGB sensor and recorded the minimum and maximum RGB values of the color that our device detected. Then, we compared our sensor RGB values to the actual RGB values of the paint sample. The right two columns of the table below compare the ideal paint color and the color detected by our sensor. This comparison shows that our RGB sensor produced RGB values that were similar to actual RGB values of the paint samples.
Determining the Color Red
Determining the Color Red
In order to prevent false alarms, our device needs to be able to accurately detect the color red. We found that the key indicator that a color is red is the ratio of the red and green RGB values, so we based our "red" color categorization off of this ratio. The color dots in the graphs below are plotted using their green value as the y-value and the red value as the x-value. Colors detected with our sensor are shown on the left, and the ideal colors are plotted on the right. Despite the fact that our blood sensor observed lower-than-ideal green values, similar red vs green trends are observed across both graphs, indicating that our sensor accurately identifies colors.
The possibility of colors other than red fitting within stand alone minimum and maximum red and green value ranges is what prompted the use of a ratio comparing the red and green input values. To determine the proper ratio value, all of the red paint sample results were compiled and the red to green values were compared. The average ratio of that sample set was determined and through testing on samples not included in the sample population, a range of +/- 1.5 standard deviations from the average ratio value was determined to only set off the alarm for the color red.
Animal Blood Testing
Animal Blood Testing
After determining our criteria for red, we tested our device with bovine and porcine blood. We found that our device successfully set off the alarm code in the presence of both types of animal blood. The image to the right shows two of the blood-soaked gauze pads that we used for testing. The pad on the left had one drop of blood and the pad on the right had 5 drops of blood on it. The alarm only consistently went off for the pad with 5 drops of blood, which is an ideal result since it will alert the patient for active bleeding events, but not for trace amounts of blood that may be present from the insertion of the needle.
Page updated
Report abuse