automated image analysis techniques for screening of mammography images
DESCRIPTION
Automated Image Analysis Techniques for Screening of Mammography Images. Enda Molloy, Electronic Eng. Initial Presentation, 7/10/08. Outline. Background Project Overview Initial Work Project Schedule. Background. Breast cancer can be missed on mammograms for a number of reasons: - PowerPoint PPT PresentationTRANSCRIPT
ENDA MOLLOY, ELECTRONIC ENG.
INITIAL PRESENTATION, 7/10/08.
Automated Image Analysis Techniques for Screening of
Mammography Images
Outline
Background
Project Overview
Initial Work
Project Schedule
Background
Breast cancer can be missed on mammograms for a number of reasons:
Cancer blends into the background of glandular tissue and is missed at screening.
Breast tissue is simply too dense and cancer cannot be seen on the mammogram.
Human error, where the radiologist misinterprets the mammogram.
Project Overview
The project aims to investigate analysis techniques for the screening of mammography images, which may be used in automated screening of a large set of images.
This will be achieved by developing a system comprising of feature extraction and a classification architecture.
It is also planned to provide functionality for remote access to the data via a web browser.
Project Overview
Image from database of mammograms
MATLAB will be used to carry out image processing
Web server and database
Initial Work
Contrast Enhancement: • Contrast Limited Adaptive Histogram Equalisation
(CLAHE) algorithm separates images into contextual regions and histogram equalisation is applied to each. This evens out the used grey values and brings out hidden features in the image.
CLAHE Example
Applying CLAHE to an image in MATLAB:
Canny Edge Detector
1. Image is smoothed by Gaussian convolution.2. Compute the x and y derivatives of the
image using a 2-D first derivative operator.3. From x and y derivative, compute the edge
magnitude.4. Suppress non-maximum edges.5. Hysteresis process.
Canny Edge Detector
Project Schedule
Now – Oct 27th Continuing with research on basic image processing
techniques for feature extraction. Familiarise myself with MATLAB. Test the different processing techniques on a subset of
mammographic images.
Oct 27th – Nov 10th Add to work already done on feature extraction. Include techniques to reduce noise e.g. wavelet
analysis
Project Schedule
Nov 10th– Nov 28th Investigate and research available options for
classification techniques. Choose a suitable classification architecture for
screening. Build and test a basic system for screening of
mammograms.
Christmas Break Time will be used to catch up if I have fallen behind. Start research on MySQL.
Project Schedule
Jan 12th– Jan 26th Using MySQL develop a simple online database that
would allow a doctor remote access to the data.
Jan 26th– Feb 16th Work on a second classification architecture and
compare the results between this architecture and the previous one developed, in terms of performance and complexity.
Feb 16th– March 2nd Test and debug the overall system.
Questions