Computer-Aided Identification of Microcalcifications using Matlab

1
Computer-Aided Identification of
Microcalcifications using Matlab

• BME 316
• Angela Bookwalter
• Kelly Braun

2
First--Some Facts

• 1 out of every 8 women will be diagnosed with
  breast cancer
• Family history plays a very important role in
  determining whether a women is at risk for breast
  cancer.
• Each year more than 180,000 new cases of breast
  cancer are diagnosed in the US. It is the leading
  cause of death for women between the ages of 35
  and 54.
• Mammograms find invasive breast cancer in about
  75 of the women who have breast cancer under the
  age of 50, mammograms can miss cancer

3
What Is Breast Cancer?

• Uncontrolled growths of abnormal cells
• Cancerous tumor that develops in the breast
• 80 of women find a lump in breast - most common
  symptom
• 20-25 of all breast cancers are associated with
  some form of microcalcifications

4
Mammograms

• special type of x-ray imaging using low dose
  x-ray high contrast, high resolution film
• Breast compression is necessary in order to image
  maximum amount of tissue, reduce x-ray scatter,
  and immobilize the breast.
• Best way to find breast cancer in its earliest
  stages-find microcalcification clusters

5
Microcalcifications

• Microscopic grains of calcium produced by the
  cells as the result of some benign or malignant
  process, such as the rest products of broken down
  cells, a cyst or milk.
• Calcifications produced by cancer cells are
  generally granular or irregular in shape and
  appear in clusters.
• Have much higher attenuation than surrounding
  tissue, absorb more radiation.

6
Finding Microcalcifications

• In some mammograms they are visible as nearly
  white spots on dark gray background and in others
  they are visible as brighter gray spots on a
  slightly darker gray background, depending on the
  grayscale.

7
Cluster of Microcalcifications

• This image of a breast shows a cluster of many
  microcalcifications

8
Branching Microcalcifications

• Left Breast
• This mammogram depicts irregular clustered
  branching calcifications in the sub-areolar area.
• This branching extends over 30 mm.
• Left LMO

9
Cancerous Tumor

• This mammogram depicts the appearance of a
  spiculated lesion in the upper outer quadrant of
  the right breast typical of carcinoma.
• The mass in this image measures 3 cm.
• This patient discovered a lump in her breast.
• Craniocaudal view

10
Digital Screening

• Image analysis technique used for the automatic
  detection, segmentation, numerical analysis and
  classification of microcalcifications
• Digital mammography is "The most fertile
  territory for major advances in the detection and
  diagnosis of minimal breast cancers."
• In a digital mammographic system the film/screen
  cassette is replaced by a phosphor screen which
  detects the x-ray photons leaving the breast. The
  phosphor screen converts the x-ray photons into
  light which is transferred through a fiber
  optical reducer to the CCD (Charged Coupled
  Device) detector. A 50mm by 50mm phosphor screen
  image is reduced to a 25mm by 25mm CCD light
  sensor. The matrix of a typical CCD is 1024 by
  1024 therefore each pixel can resolve 0.05mm.
  The CCD converts the incident light into a
  digitized analogue signal and this signal is
  displayed onto a computer monitor. Rather than
  reaching an optical density of 1.4, the exposure
  is terminated when a sufficient signal to noise
  ratio is achieved.
• This would reduce the discomfort to the patient
  as it would reduce the procedure time
• Digital imaging of the breast produces a much
  wider and more linear dynamic range and is
  therefore more ideal for radiological dense
  breast types.
• http//www.ozemail.com.au/glensan/digmam.htm

11
Objective

• Develop a program in Matlab which will find
  individual microcalcifications in a breast image
  and
• Display black and white image highlighting
  microcalcifications as white points.
• Count the microcalcifications.
• Pick out clusters.
• Give average distances between individual
  microcalcifications.

12
Matlab Programming

• Scans image
• Issues gray scale values of 1 or 0, 1 for the
  microcalcifications and 0 for background
• reads the image into an array based on 1s and
  0s
• finds clusters of 1s

13
Results
14
Discussion
15
Conclusion
16
fty