I want to segment the masses from the mammogram images to reduce the false positive. Can anybody suggest a good method or some code for this? I have used morphological operation with the watershed, but it gives too many false positives and displays the over segmentation problem.
It seems to me there are two distinct issues. Segmentation of a lump, and classification of it: is it benign or malignant.
For classification, just the size of it (result of segmentation) is not good enough. You will need to incorporate more features to improve your ROC curve.
Lambert Zijp thankyou for ur reply ..
currently im focusing on segmentation, how to extract the lump and taking benign and malignant as one class of masses and for classification just using the haralick features. Could you suggest some more method and features that I should try on?
I'm sorry Jamal, I don't know much about mammograms. There should be a vast amount of literature though.
Dear Jamal,
Did you get the best result? for segment the masses from the mammogram images.
Actually i was not able to find a way in super pixels , so i implemented curvelet transform
Dear Jamal,
please, can you send me the curvelet transform code, also the paper you published.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics