TopIntroduction
Diagnostic imaging is a priceless tool in the field of medicines today. Magnetic resonance imaging(MRI), phase-contrast X-ray imaging (PCI), computed tomography (CT), X-rays, ultrasound scanner, digital mammography, and other imaging modalities provide an effective means for mapping the anatomy of a subject(Phamy, Xu, &Prince, 2001).Knowledge related to the anatomy of normal and diseased tissues has increased a lot because of these technologies. Computer Aided Imaging Diagnostics (CAID) techniques are being used to capture the patterns expressed by disease conditions in the images which can give certain clues to the doctor about the disease. Many such simplified imaging approach can be an inexpensive way to provide an early detection of certain diseases so that a warning message could be given to people who are at places where no sophisticated systems and trained doctors are available to deeply detect and confirm (Shen, Cheng, & Basu, 2010).
Before getting into diagnostics part of medical images, it is worth to have a glance at the features of a typical medical image in the digital format. It is a collection of pixels (point or dot) which contributes to the formation and presentation of that image. Each pixel is associated with some light intensity with a value. In medical images, traditionally, the value associated to each pixel represents the properties of cells within an organ. For example, pixel values represent the radiation absorption in X-ray imaging, acoustic pressure in ultrasound, and RF signal amplitude in MRI. However, new methods and techniques which are using phase contrast and other innovative methods to generate data with better resolution are under research. If a single data is collected (as in radiation absorption in X-ray) from each location or point in the image, then the image is called a scalar image. If more than one data is collected (as in dual-echo MRI), the image is called a vector or multi-channel image. Images can be represented in two-dimensional (2-D), three-dimensional (3-D) or in multi-dimensional space.
The methodology of dealing with medical images for advanced analysis is different from conventional visual verification analysis and validation. In the conventional method, identification of locations of anomalies is done by the radiologists. The radiologist uses apriori knowledge about the location, size, and shape of the structures in the medical image and matches that with clinical evidence as well as case history to make the diagnosis. In CAID, detection of the anomaly and recognition (classification) is done by the computer either in supervised (with help of doctors) or in unsupervised (without human intervention) domain. There are some standard steps to be followed for CAID. Those steps can be described as follows-