Biomedical Engineering Reference
In-Depth Information
Figure 1.
General ultrasound image analysis architecture.
3.1. Typical Setup
Ultrasound has evolved gradually from a limited modality to a sophisticated
tool for diagnosis of breast lesions. Its criteria for lesion differentiation include
margin irregularity, shadowing, microlobulation, echogenicity, and shape (see
Stavros et al. [27]). Palpation is widely used in a screening procedure before
a breast imaging study because pathological changes in tissue are generally cor-
related with changes in elasticity (see Ophir et al. [28, 29]). The tissue elasticity
of a lesion could also be a useful diagnosis criterion in the United States. After
compression of the probe, a soft lesion will be flattened more than a hard one.
Many cancers appear as extremely hard lesions, which is the result of increased
stromal density. Fibroadenoma typically has a heterogeneous stiffness; cancers
are uniformly stiffer than surrounding tissue.
In the past, an attending physician was the most important personage for pa-
tients. When ultrasound images were obtained, the physician would diagnose
and analyze the ultrasound data himself, which was very time-consuming work.
For this reason, computer-aided diagnosis (CAD) systems were developed to aid
physicians in diagnosing ultrasonic abnormalities and possible breast cancers.
With the assistance of CAD, physicians can efficiently detect such lesions as spic-
ulations and discriminate more accurately between benign and malignant tumors
(see Horsch et al. [30] and Chen et al. [31]). The purpose of this chapter is to
