During the past decade, computer-aided diagnosis (CAD) has been shown to be of clinical benefit in fields such as detection of microcalcifications and classification of masses in mammograms (Astley and Gilbert 2004). The concept of CAD is not unique to these fields; indeed, it is more important and beneficial for examinations in which a large quantity of images need to be interpreted rapidly for finding a lesion with low incidence, such as the detection of polyps in CT colonography (CTC) and the detection of lung nodules in thoracic CT scans. In its most general form, CAD can be defined as a diagnosis made by a radiologist who uses the output of a

H. Yoshida, PhD

Associate Professor, Department of Radiology, Massachusetts General Hospital and Harvard Medical School, 75 Blossom Court, Suite 220, Boston, MA 02114, USA

computerized scheme for automated image analysis as a diagnostic aid (Doi 2004). Conventionally, CAD acts as a "second reader," pointing out abnormalities to the radiologist that otherwise might have been missed. The final diagnosis is made by the radiologist. This definition emphasizes the intent of CAD to support rather than substitute for the human reader in the detection of polyps.

CAD for CTC typically refers to a computerized scheme for automated detection of polyps and masses in CTC data. It provides the locations of suspicious polyps and masses to radiologists. This offers a second opinion that has the potential to improve radiologists' detection performance and to reduce variability of the diagnostic accuracy among radiologists, without significantly increasing the reading time. Such a CAD scheme should be distinguished from semi-automated computer applications in radiology that automate only one of these components and depend on user interaction for the remaining tasks. A typical example is the 3D visualization of semi-automatically segmented organs (e.g., segmentation of the liver, endoluminal visualization of the colon and bronchus), or image processing of a part of an organ for generation of an image that is more easily interpreted by human readers (e.g., peripheral equalization of the breast in mammograms, digital subtraction bowel cleansing in virtual colonoscopy).

Despite its relatively short history, CAD is becoming a major area of investigation and developments in CTC. Rapid technical developments have established the fundamental CAD scheme for the detection of polyps during the last several years. Prototype CAD systems have been demonstrated at conferences (NAppi et al. 2005b; YosHiDA et al. 2004b) (Fig. 11.1) and commercial systems that implement the full CAD scheme or a part of it are becoming available in the market with names such as the Poly Enhanced View (Siemens Medical Solutions) and Colon Computer-Assisted Rader (MedicSight Inc.).

In the colon CAD workstation shown in Fig. 11.1, for example, the left and right images show the 2D

Fig. 11.1. Prototype colon CAD workstation

multiplanar reconstruction (MPR) views of the supine and prone scans of a patient, respectively, with the computer-extracted colonic wall superimposed. The bottom middle two images show the corresponding 3D endoluminal views of the colon. Polyps detected by CAD are shown as a list of icons on the middle row of the screen. By clicking on one of the icons, one can jump to the corresponding polyp on a 3D endoluminal view and/or an MPR view. The polyp (green) is displayed in both supine and prone views if it is found in the corresponding regions in these two views (see Sect. 11.5.2). CAD output is integrated into the 2D MPR and 3D endoluminal views by use of the coloring scheme that delineates the detected polyps and the normal structures in the colonic lumen (see Sect. 11.2).

The latest prototype CAD systems yield a clinically acceptable high sensitivity and a low false-positive rate (see Sect. 11.4), and they are becoming integrated into the 3D workstation for CTC examinations and thus into clinical workflow. However, some technical and clinical challenges still remain as open problems for CAD to become a clinical reality.

The remainder of this chapter describes the benefits of CAD, the fundamental CAD scheme, detection performance of CAD, pitfalls in CAD, and the current and future challenges in CAD.

Was this article helpful?

0 0
Managing Diverticular Disease

Managing Diverticular Disease

Stop The Pain. Manage Your Diverticular Disease And Live A Pain Free Life. No Pain, No Fear, Full Control Normal Life Again. Diverticular Disease can stop you from doing all the things you love. Seeing friends, playing with the kids... even trying to watch your favorite television shows.

Get My Free Ebook

Post a comment