A mammogram is an X-ray picture of the breast. Doctors use a mammogram to look for early signs of breast cancer. The one shown above is called a mediolateral mammogram.
Pectoral muscles (colloquially referred to as "pecs") are the muscles that connect the front of the human chest with the bones of the upper arm and shoulder.
Canny Edge Detection is a popular edge detection algorithm. Using Canny Edge Detector solely for pectoral muscle segmentation produces quite unsatisfactory results. Thus, we use hough transform line detection.
The Hough transform is a technique which can be used to isolate features of a particular shape within an image. Because it requires that the desired features be specified in some parametric form, the classical Hough transform is most commonly used for the detection of regular curves such as lines, circles, ellipses, etc.
- The algorithm seems to produce satisfactory result on the images. However, not so much on the last image,
mammo_5.jpg. This is so because the pectoral region here can not simply be modelled by a single line. - We can repeatedly apply this algorithm on such images, and try to chisel away remaining parts of the pectoral muscle in each iteration.
- The parameters for the shortlisting lines have been chosen manually. However, they can be easily learned given a bigger dataset.
Give me a place to stand, and I shall move the earth. - Archimedes
Give me more data, and I shall learn the parameters. - Me :)
Please refer pectoral-segmentation.ipynb for the complete code. I have tried to explain each step as clearly as I could.
Resources and references
- Github repo by @anoo6527
- Assignment by Suven Consultants and Technology Pvt. Ltd.
- Paper: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6510623/
- Paper: https://core.ac.uk/download/pdf/82133766.pdf
- Tutorial on Youtube: Computer Vision Basics: Hough Transform | By Dr. Ry @Stemplicity
- Scikit-Image Hough Transform tutorial
- Science Direct article: https://www.sciencedirect.com/science/article/pii/S1361841518301129