Both breast ultrasonography (US) and scintimammography (SMM) shows limited diagnostic accuracy in the differential diagnosis of breast cancer. We investigated whether the diagnostic performance of computer-aided diagnosis (CAD) of US was improved by adding SMM results. We retrospectively reviewed 40 breast US images and corresponding SMM images from 40 patients who presented with breast masses (21 malignant and 19 benign tumors). The morphologic features of the breast lesions on US were extracted and quantitated using the automated CAD software program (So no Eye, CAD Impact, Inc., Seoul, Korea), which calculated the probability of malignancy based on the stored database. The quantitative data of SMM were obtained as the uptake ratio of lesion to contralateral normal breast on SMM SPECT. The diagnostic accuracy of CAD of US and SMM was determined using ROC curve analysis, respectively. The best discriminating function combining the CAD results of US and SMM uptake value was created using linear discriminant analysis and the diagnostic performance was compared to that using only one diagnostic modality. Both US and SMM showed a relatively good diagnostic accuracy (area under curve=0.831 and 0.846, respectively). Combining CAD results of US and SMM resulted in improved diagnostic accuracy (area under curve =0.860), but it was not statistically significant. The diagnostic performance of CAD of breast US in the differential diagnosis of the breast mass was not significantly improved by adding SMM. However, SMM SPECT may be complementary to CAD of US in differential diagnosis of breast cancer.
[1]
S. L. Stewart,et al.
Cancer mortality surveillance--United States, 1990-2000.
,
2004,
Morbidity and mortality weekly report. Surveillance summaries.
[2]
Ming-Kuei Hu,et al.
Visual pattern recognition by moment invariants
,
1962,
IRE Trans. Inf. Theory.
[3]
Yong Hoon Lee,et al.
An edge gradient enhancing adaptive order statistic filter
,
1986,
IEEE Trans. Acoust. Speech Signal Process..
[4]
Peter J. Ell,et al.
Nuclear medicine in clinical diagnosis and treatment
,
2004
.
[5]
D. Chen,et al.
Computer-aided diagnosis applied to US of solid breast nodules by using neural networks.
,
1999,
Radiology.
[6]
N. Otsu.
A threshold selection method from gray level histograms
,
1979
.
[7]
M. Giger,et al.
Computerized detection and classification of cancer on breast ultrasound.
,
2004,
Academic radiology.
[8]
D. Kopans.
The positive predictive value of mammography.
,
1992,
AJR. American journal of roentgenology.