This site is devoted to mathematics and its applications. Created and run by Peter Saveliev.

Image analysis for a hand-held diagnostic device

From Mathematics Is A Science

Jump to: navigation, search
The image produced by the DNA detector. Not aligned with the rows, no matter.

The image came from a hand-held DNA detector, which has potential markets in security, vet, medical, agricultural, etc.

The job was to customize Pixcavator for these purposes. The main difference is in the output table. The table should be a table of 0s and 1s, with 1s indicating the found dark spots located in the wells.

As always, I received a few analyzed images. The user applied Pixcavator (version 2.4) and once he thought that he found everything necessary, the image and the table were saved (Pixcavator saves them with the settings of the first two sliders automatically appended to the image name). I used these images to find the locations of the wells, sizes of the wells and spots, intensities of spots of different kinds and the background etc, and then use them to search for and identify dark spots.

Unfortunately, no a priori information about the content of the images (locations, sizes, intensities etc) was available.

The image file is a microarray plate of samples with a fluorescent signal from the wells. The table can be used to find the values where our thresholds can convert the value to a 1 or a 0. Basically the light gray spots are recorded as zeroes and the darker spots are ones.

In the final image, I catch all wells and then classify them based on their intensity (instead of contrast and size). This certainly can be done in a number of ways.

Analysis results: the table of responses is inside the blue rectangle.

The data analysis part is fairly east...

Other image analysis examples.