Question
Question Posted 11/21/16:
What is the best colormap for PET imaging in image processing?
What is the best colormap for PET imaging in image processing?
Response posted 11/21/16 by Bob Koeppe:
There really is no correct answer to this question. It is definitely a matter of preference and what one is used to looking at. To a trained eye, a black and white scale is as good as any, however some differences are more easily seen in color (particularly when comparing regions that are remote from one another). There are many instances where a particular color map may be mis-leading. For example, when one section of the color table covers many quantitative values (for example, all the greenish colors), but then when the quantitative values changes just a little more, then the color changes dramatically (or example into a bright yellow or red).
Response posted 11/21/16 by Emimal jabason:
Thank you so much for your prompt response. I was trying to use molecular imaging (PIB), and in the gray scale, I could not identify anything. Do you have any suggestions? Thanks
Response posted 11/21/16 by Bob Koeppe:
There is no correct answer to this question. It is definitely an individual preference and is influenced by what one is used to looking at. Black and white is actually preferred by many (particularly radiologists trained on film images) and I agree that subtle differences are seen well in black and white. B/W is harder to compare regions that are remote from one another, however. Color maps, on the other hand, can be misleading as a large section of the color scale may look similar (all in a greenish hue for example), yet a small change could move scale into a totally different color. I have seen very specific color scales used (in talks and posters) that were set to highlight small difference when these differences weren't really very significant at all. A good color map will be one that changes in appearance across out the entire range of values, yet will not be confusing as to what color represents a higher number than another.