Last class, Serge suggested that I extract patches from my b-mode images and sort them from less dense to denser. Looking through my images, I realized that I only care for certain region of the image. In other word, the region of interest in each b-mode image changes because the size of the region containing the breast scan varies in each slice. So this is taking longer than I expected !!
After collecting the patches, I need to sort them. I compute the average reflectivity of each image that allows for automated sorting of the images from less dense to denser. We know that the glandular (more dense) tissues have a higher reflectivity than fatty tissues.
more images
here
Woah, are these imaging techniques related to your project?
ReplyDeletehttp://ucsdnews.ucsd.edu/newsrel/science/04-11OpticalImaging.asp
David the images are acquired using a specially developed ultrasound imaging machine. And yes the imaging is what makes my computation more novel because there has been studies done by others using mammogram or MR or etc. However, we are using a Low-risk Volumetric Ultrasound Imaging which provides 3D images with no physiological hazards. It's cheaper than MRI and the image quality is comparable.
ReplyDeleteHi Hourieh,
ReplyDeleteI am working on shape analysis of mammography masses (though I plan to apply it on biopsy slides in near future), and to improve segmentation; I am looking into the breast density. Currently I have access to Matakina's commercial software, however it would be great to see how you do it.