发信人: leohawk (leohawk), 信区: Biology
标 题: Re: Image Analysis (2)
发信站: The unknown SPACE (Tue Jun 19 16:19:12 2001), 转信

should you pay more attention to the intensity of the signal instead of
the size of the signal area?

you can write a program that reads a fixed area in the matrix, each read out
can be an instance of an object, then you can manipulate the object. (in C++
or in Java). I think this can even be done in Matlab.

As to the site of the spot, in the object design for spot, can you let the
boder pixels remember whether their outside pixels also have signal or not?
This information might be useful for some analysis.

or, as an alternative, let the program pick up an area where all pixels in it
share similar signal and treat it as a signal spot, then the area size
varies, you will have to design your object differently.

In each spot, you can do all kind of manipulation of the pixel values... And I
think these is the fun part as there seem to be many choices.

can you keep me updated with your proj, it's quite interesting.

【 在 springtime (Spring) 的大作中提到: 】
: Thanks! my problem is: Affymetrix chooses 8x8 pixels for
: each spot, but
: it is not very accurate for every spot cause you could see
: there are some
: spots for which 7x10, 6x8, ....is the best. Now I don't know
: how to transform
: this case into mathematical representation.
: Spring


--
※ 来源:．The unknown SPACE bbs.mit.edu．[FROM: 165.230.116.134]