Summary

Techniques developed for computer classification of malignant and benign breast lesions can potentially be used clinically to help radiologists differentiate breast cancers and benign breast lesions. These techniques can be either based on human perceptual features provided by radiologists or based on computer-extracted features that can be obtained with a fully automated process. The identification of the important features, either perceptual or computer-extracted, is critically important for computer analysis of mammograms. Once

TABLE 2 Attending radiologists' assessments of likelihood of malignancy and their clinical recommendations on a malignant case with computerassessed likelihood of malignancy of 60%a

Attending radiologists

0 0

Post a comment