Mask extraction from manually segmented MIAS mammograms

The first step in computer aided detection (CAD) in mammography relies on accurate image segmentation. Testing of CAD algorithms should include comparison with other proposed methods in order to show how a new method compares with ones presented before. Two most popular mammographic databases, which...

Full description

Saved in:
Bibliographic Details
Main Authors: Mustra, M., Grgic, M., Huzjan-Korunic, R.
Format: Conference Proceeding
Language:English
Subjects:
Breast
Computer Aided Detection
Design automation
Image Registration
Image segmentation
Manuals
Muscles
Region Growing
Testing
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The first step in computer aided detection (CAD) in mammography relies on accurate image segmentation. Testing of CAD algorithms should include comparison with other proposed methods in order to show how a new method compares with ones presented before. Two most popular mammographic databases, which are publicly available, consist of scanned films. This presents a segmentation challenge in order to achieve the best possible results. The major problem which arises in testing automatic segmentation algorithms is defining the ground truth. In this paper we present a method for automatic extraction of manually segmented breast segmentation masks from MIAS database. The manual segmentation process has been performed by the professional radiologist on printed images, which have later been digitized and breast pectoral muscle masks have been extracted. The extraction process differs for breast mask extraction and for pectoral muscle mask extraction because of different manually drawn segmentation line properties.
ISSN:1334-2630