The Role of Computer-Aided Detection in Diagnostic Medical Imaging

Only available on StudyMode
  • Download(s) : 286
  • Published : August 5, 2008
Open Document
Text Preview
The Role of Computer-Aided Detection in Diagnostic Medical Imaging

Innovation and Imaging
Methods of improving the acquisition, display and interpretation of diagnostic medical images have been in a constant state of innovation since the discovery of x-ray in eighteen ninety-five (1895). The biggest changes have occurred along with and largely because of incredible advancements in computer technology. Medical imaging developers have harnessed computer technologies to perform tasks that have helped shape the typical diagnostic imaging department into an indispensable part of the diagnostic team.

Computerized Tomography
As computer power and speed has increased so has its utility in diagnostic imaging. As an example, look at the evolution of Computerized Tomography (CT) technology from its inception in the early nineteen seventies (1970’s) to its present day incarnation. The acquisition time for a single slice on a typical first generation CT scanner was approximately four hundred and twenty seconds. Scanners of today can easily cover an area from the apex of the lungs to the tip of the toes in about twenty-five seconds. To clarify how dramatic this rapid change in CT technology is, let us compare it to advances in aviation. From the Wright brothers flight at Kitty Hawk to the flight of a modern space shuttle, there has been a speed increase from forty miles per hour to twenty-five thousand miles per hour. This change represents a six hundred and twenty-five times increase in speed. In comparison, CT speed gains represent nearly a one thousand times increase (Beason, 2005, October 5).

Computerized Diagnosis
The first computerized image analysis and interpretation system was conceived in the nineteen sixties (1960’s) in an attempt at complete automation of the radiographic exam. “These early studies displayed a considerable optimism regarding the capabilities of computers to generate complete diagnoses” (van Ginneken, 2001, p.1228). The high expectation of the computer’s capabilities dwindled over time due to the complexities of radiographic interpretation.

Computer-aided Detection
Lessons learned from those early experiments lead the way for current computer-assisted methodologies to develop. “Modern theories hold that the human reader makes the diagnosis based on the computer output….balanced against the radiologist’s interpretation, patient history and other factors” (Doi, 2005, p. S3). The name given to this new generation of systems is Computer-aided Detection (CAD). CAD solutions are currently in use throughout the diagnostic imaging enterprise and will expand into other areas as research and development continues. Computers will not replace the role of the radiologist as diagnostician as once feared. Rather, computer-aided detection (CAD) will compliment human interpretation of diagnostic medical data resulting in greater diagnostic accuracy and streamlined workflow.

How CAD Works

Pattern Recognition
Computer-aided Detection works by applying a computer algorithm to the analysis of an image or volume of image data. The algorithm is a set of instructions that tell the computer to look for patterns in the image. The system then confirms all known patterns and makes note of any abnormalities in the pattern that it may have encountered. In order to do this, the algorithm has to include all of the possible normal features that may be contained in the data. The algorithm scheme detects and tags abnormal features for further analysis by the radiologist. Lung nodule analysis utilizes this type of algorithm to detect abnormal nodules in volumetric CT chest exams (Wiemker, 2005, p. S46).

Signal Intensity Recognition
Other CAD schemes analyze and quantify changes in pixel intensity over time. These quantification schemes make cerebral blood perfusion studies and dynamic MR breast analysis possible. The algorithm measures pixel intensity across the duration of the scan and quantifies changes in...
tracking img