🚀 go-pugleaf

Well you could help me by finding a method to automatically edge detect
the cardiac shadow on cardio angiograms

chris@uxb.liv.ac.uk

In article <C0ozIH.AMJ@compsci.liverpool.ac.uk> chris47@uxb.liv.ac.uk (Chris Brown ) writes:
>Well you could help me by finding a method to automatically edge detect
>the cardiac shadow on cardio angiograms

Well, I suppose I could. :-)

There are a number of edge detection algorithms.  The usually involve
a convolution of some sort.  One of the more popular is the Sobel
filter, the kernel of which looks like

	| -1  0  1 |
	| -2  0  2 |
	| -1  0  1 |

or

	| -1 -2 -1 |
	|  0  0  0 |
	|  1  2  1 |

Check your favorite image processing book for more details.  Here's a
list of mine:

Russ, John C., _The Image Processing Handbook_, ISBN 0-8493-4233-3

Pratt, William K., _Digital Image Processing, Second Edition_, 
ISBN 0-471-85766-1

Lim, Jae S., _Two-Dimensional Signal and Image Processing_, ISBN
0-13-935322-4.

Gonzalez, Rafael C. and Woods, Richard E., _Digital Image Processing_, 
ISBN 0-201-50803-6  (There's also an earlier edition by Gonzalez and
Wintz that some have criticised for minor errors in math, etc.)

Jain, Anil K., _Fundamentals of Image Processing_, ISBN 0-13-336165-9

Castleman, Kenneth R., _Digital Image Processing_, ISBN 0-13-212365-7

although it's a little old, these days, I also like

Pavlidis, Theo, _Algorithms for Graphics and Image Processing_,
ISBN 0-914894-65-X

							spl
-- 
Steve Lamont, SciViGuy -- (619) 534-7968 -- spl@szechuan.ucsd.edu
UCSD Microscopy and Imaging Resource/UCSD Med School/La Jolla, CA 92093-0608
"Personally, I don't care whether someone is cool enough to quote Doug
Gwyn--I only care whether Doug Gwyn is cool enough to quote." - Larry Wall

 spl@ivem.ucsd.edu (Steve Lamont) writes:
>In article <C0ozIH.AMJ@compsci.liverpool.ac.uk> chris47@uxb.liv.ac.uk (Chris Brown ) writes:
>>Well you could help me by finding a method to automatically edge detect
>>the cardiac shadow on cardio angiograms
>
>Well, I suppose I could. :-)
>
>There are a number of edge detection algorithms.  The usually involve
...description of Sobel filter omitted...

In the cardiac shadow domain, you could exploit:
* the fact that the edge is moving, so you can use motion-enhanching
techniques like image differencing
* the fact that the edge has a roughly known shape, ie, a long slowly
curving arc, which can be exploited to generate better matched filters

The more you know about what you're looking at, the better you can be
at detecting it.  If you know that its motion is smooth, you can exploit
this better.

Check out the books recommended, and if your budget allows hire an EE type who knows how to code, or a CS type who knows somthing about image processing.

--
David Honig

"Q: What does a C programmer say to an Ada(tm) programmer?
 A: I'll have a burger and fries, no onions."