Many thanks to everyone who has taken part so far with some excellent contributions. A few more participants are still needed so if you have a few minutes to spare I would really appreciate your input!
Image analysis, used to describe particular features of an image, is important in science. The need to characterise brain pathologies in as standard a way as possible underpins much of the work I do in the area of brain ageing. Most of the work describing what happens in the brain during ageing relies on neuropathologists examining very thin sections from the brain for features associated with disease. Often these features are marked by special stains that pick out specific forms that represent abnormalities inside cells and these can be counted to estimate the severity of disease processes. It is painstaking work involving many hundreds of sections and can take months to do. It is the kind of job that could benefit from a computerised automatic system and integrated microscope and image analysis equipment is available, though very expensive, at around $300,000 to buy and around $35,000 per year to maintain.
As part of the Biofilter Project (salhunter.deviantart.com/galle…), I want to see if these automated image analysis systems really can compare to human interpretations of complex images. Therefore, I am asking the good citizens of the interweb for participation.
One feature that often turns up my drawings is “Holes”. These are mostly quite small dot like structures which are a fairly consistent feature across the set of drawings that would be ideal for image analysis. In order to get an idea of how well humans do, I need a baseline set of counts to compare with analysis by computer. Anyone with a graphics program that allows the addition of a new layer can participate.
The idea is to mark each hole with a red dot and send the resulting red dot layer to me for counting - the following detailed instructions describe how to do this. It should take around an hour. In return, I give everyone who completes the task the permission to print out the high resolution image for their own non-commercial use as a gift to say thank you!
1. Download the 3000x3000 pixel experimental image file "Specimen - 34" via the download button here - salhunter.deviantart.com/art/S…
2. Open the file in your graphics program
3. Check the file is 3000x3000 pixels in dimension.
4. Add a new layer
5. Select a paint brush or drawing tool at around 10 - 20 pixels with a hard edge
6. Select a strong red colour
7. While on the new layer, mark each hole that you find with a single red dot using your selected drawing tool
8. It is advisable to “save as” a new file and keep saving every so often!
9. Once you are happy that you have marked all the holes you can find, save the new layer as a separate file in .jpg format at high quality (image quality of around 5 should be OK). The image should still have dimensions of 3000x3000 pixels10. Send me the resultant .jpg file of red dots - email to sal DOT hunter AT ntlworld DOT com (I'm sure you all know how to edit this no spaces)
11. In your email I would also like to know if you are mainly an artist, a scientist, neither or both
Many thanks to anyone who participates!