COLOR DOPPLER SONOGRAPHY IN BREAST-TUMORS - COMPUTER-AIDED QUANTITATIVE ASSESSMENT

Purpose: Quantification of color Doppler findings in breast tumors usi ng a computer-assisted image analysis. Methods: Color Doppler images a re quantitatively characterized by the color pixel density and the flo w velocities corresponding to the encoding colors (''color values''). The results in 44 patients with histologically proven breast carcinoma and 49 patients with benign lesions were compared with conventional d uplex measurements. Results: With a scale maximum of 6 cm/s, the color pixel density was 4.8 +/- 3.7 % for carcinomas (mean +/- standard dev iation) and 1.5 +/- 2.5 % for benign lesions (p < 0.0001). The mean co lor value was 2.8 +/- 0.6 cm/s for carcinomas and 1.1 +/- 0.8 cm/s for benign changes. Peak systolic flow velocity as determined with duplex Doppler was 23.1 +/- 10.3 cm/s in carcinomas and 16.7 +/- 8.3 cm/s in benign changes. With this scale maximum, the sensitivity for the diag nosis of breast cancer was 84 % with the mean color value, 59 % with t he color pixel density, and 55 % with the peak systolic flow velocity. The specificity was 67 % with the mean color value, 88 % with the col or pixel density, and 45 % with the peak systolic flow velocity. Pulsa tility index and resistance index wore not significantly different in benign and malignant lesions. Conclusion: Computer assisted image anal ysis may be a helpful tool to quantify color Doppler information. It a ppears superior to conventional Duplex Doppler in differentiating carc inomas and benign lesions in the breast.