Introduction - kovimesterr/SSIP2013 GitHub Wiki

Breast cancer is the type of cancer with highest incidence rates in women, accounting for 26% of cancer occurrence. It is the most common cause of cancer death among women 20–59 years old. Statistics indicate that 1 in 8 women will develop breast cancer during their lifetime [1].

Because there is currently no known cure for breast cancer, the most successful means for reducing the mortality rate are early detection and improved treatment methods. Thus, prevention and screening have become important health issues.

Screening looks for cancer before a person presents any symptom. It can help to find cancer at an early stage. When cancer is found earlier, the healing chances are greater. If a screening test result is abnormal, more diagnostic tests are necessary to define whether the finding is a cancer .

In the rest of this report, we start by a brief explanation of the breast anatomy followed by mammography technique. Then we will explain the need for proper mammogram enhancement algorithm for helping early detection of breast cancer. Then we detail some of this algorithms which are most common and finally we will be conclude with a performance evaluation and comparison of these methods on a mammogram dataset.

Breast Anatomy (What is breast density and how it impacts the risk for cancer):

the breast mainly have two types of tissue: the glandular tissue and adipose tissue (see Fig 3). Their ratio varies from patient to patient and according to age, hormonal status (menstruation or menopause), changes in weight of the patient, etc. For this reason, each breast has his own identical structure and it is impossible to identify a structure as a reference for further easy analysis.

For the diagnosis, doctors normally perform bilateral comparisons between left and right breast (see Fig 4) and regular tests (tests which are performed at different times, which are part of the medical history of the patient).

Breast anatomy

The dense breast tissue is comprised of less fat and more connective tissue (glandular tissue). Increased breast density is a strong independent risk factor for developing breast cancer. The more glandular tissue, the greater the density and the higher the risk of breast cancer. (women with dense breasts are at a 4-6 times higher risk of developing breast cancer than women who do not) [2].

Unfortunately, in women with dense breast tissue, mammograms are not as effective in detecting the cancer. This is because both breast tissue and breast cancer will appear white on a mammogram (thus tumors are often hidden behind the dense tissue) and the lack of contrast makes identification of the cancer more difficult.

An example of a bilateral comparison

This is why more than one third of breast cancers are missed by mammography in women with dense breasts. Fig 4 demonstrate this issue. By comparison, on the left there is an example of a breast cancer in a woman with non-dense breast tissue. The white tumor (circle) is easily visible against the dark gray and black non-dense tissue. However, the mammogram on the right shows a very dense breast and even though it too has a cancer in roughly the same location but it difficult to see [3].

The effect of breast density on the visibility of the cancer

Mammography

X-ray mammography is the most common technique used by radiologists in the screening and diagnosis of breast cancer in women. Although it is seen as the best examination technique for the early detection of breast cancer reducing mortality rates by up to 25%, their interpretation requires skill and experience by a trained radiologist.

The World Health Organization (WHO) has suggested that two components of early detection have been shown to improve cancer mortality [5]:

  • Education: to help people recognize early signs of cancer and seek prompt medical attention for symptoms.
  • Screening programs: to identify early cancer or pre-cancer before signs are recognizable, including mammography for breast cancer.

Like all X-rays, mammograms use doses of ionizing radiation to create images where the overall contrast displayed in the mammogram results from the x-ray attenuation as well as from the optical enhancement provided by the mammographic film. During the exam, the breast is compressed using a dedicated mammography unit between 2 plastic plates attached to the machine unit in order to spread the tissue apart. This squeezing is to ensure that there would be very little movement, thus the image is sharper and the examination could be done with a lower x-ray dose (see on the next Fig).

Mammography unit

Mammograms Analysis Difficulties:

The detection of either mass lesions or micro calcifications takes place in two steps. Firstly a pre- processing step is carried out in which the whole image is enhanced. Then the individual tumors are detected using different methods which include segmenting the tumors.

Tumor detection in mammograms through image processing is a difficult task due to the following reasons:

  1. intensity levels vary greatly across different regions in mammogram
  2. features for segmentation are hard to formulate
  3. subtle gray level variations across different parts of the image make the segmentation of tumor areas by gray level alone difficult
  4. tumors are not always obvious, especially where they are subtle or extremely subtle under the glandular tissues, which makes the task of interpretation difficult even for the radiologists themselves
  5. mammograms contain low signal to noise ratio (low contrast) and a complicated structured background.

In addition, mammography is 2D projection acquisition of a 3D structure, resulting in structure and tissue overlap. Thus, although breast compression is typically performed, normal anatomical structure, such as the parenchymal tissue, can combine with useful diagnostic information (e.g., a tumor) in such a way as to impede visualization and reduce lesion detectability.

Because of these limitations (spatial resolution, breast density and overlapping), discrimination between normal and abnormal tissue is poor.

The main obstacle lays in low contrast between normal and malignant glandular tissues and the noise in such images that makes it very difficult to segment them. Therefore, in digital mammogram there is a need for enhancing imaging before a reasonable interpretation and segmentation can be achieved.

The purpose of the image enhancement is to provide an automated tool to smoothing, deblurring, noise removing or, in the most case, gray level modification for an increase of contrast. The gray level modification is one of the most popular methods to perform image enhancement because it is simple in implementation and fast in computing. Because mammograms have limited contrast it may be hard to see an anomaly. An enhanced image could both help the specialist see different structures in the image and check each mammogram faster. Therefore, in mammogram there is a need for enhancing imaging before a reasonable segmentation can be achieved.

Types of image enhancement include, noise reduction, edge enhancement and contrast enhancement. In the next section, we present several algorithms which are suitable for either removing noise or enhancing the contrast.

References