We found that interpretation with CAD-based analysis for enhancement kinetics significantly improved the discrimination of benign from malignant lesions compared with interpretation with manual analysis of enhancement kinetics. In our study initial enhancement was below 50% in 86% of benign lesions. These results are in keeping with the findings of Williams et al. [5
]. Other authors also described an increase in specificity of using CAD that is based on excluding lesions with low threshold enhancement [6
Difference in specificity between MR interpretation on an Advantage Workstation with manual assessment of enhancement kinetics and CAD may also partly be explained by the fact that CAD provides enhancement information for all pixels in a lesion rather than for a portion of a lesion measured by using manual region-of-interest placement [5
Using threshold enhancement alone, the sensitivity of CAD is high for both the 50% threshold and 100% threshold. There was only one malignant lesion that was false negative at CAD and did not demonstrate enhancement at the 50% and 100% thresholds. This lesion was described by three readers as a lobular-shaped mass with irregular margin and heterogeneous enhancement. One reader described this lesion as a round-shaped mass with regular margin and heterogeneous enhancement. Placing a manual ROI showed a malignant curve (Fig. ). Three out of the four readers classified this lesion as a BI-RADS 5 lesion, highly suggestive of malignancy, and one reader as a BI-RADS 4 lesion, suspicious. Histopathological evaluation revealed an invasive ductal carcinoma. There are some technical limitations to the CAD program that can cause negative enhancement at malignant lesions which underscore the importance of using CAD as a complement to but not as a replacement for the radiologist’s assessment [5
Fig. 3 Example of a lesion that was not colour-coded by CAD as possibly malignant, because the absolute pre-contrast signal intensity was determined to be “too low”. Manual analysis revealed a washout-type curve for this lesion, with positive (more ...)
Three benign lesions had positive enhancement at a 50% threshold and two of these lesions had positive enhancement at 100%. The lesion that had positive enhancement at 50% threshold and negative enhancement at 100% threshold consisted of normal tissue at histopathological evaluation. Two readers classified this lesion as a BI-RADS 3 lesion, probably benign (one resident and one radiologist), and the other two readers classified this lesion as a BI-RADS 2 lesion, benign. The two false-positive lesions with positive enhancement at the 50% and 100% thresholds revealed one lobular carcinoma in situ and one fibroadenoma. Two readers classified the fibroadenoma (Fig. ) as a BI-RADS 3 lesion, one as a BI-RADS 2 lesion and one as an BI-RADS 4 lesion. Lobular carcinoma in situ (Fig. ), a high-risk lesion, was twice classified as a BI-RADS 3 lesion and twice as a BI-RADS 4 lesion.
Fig. 4 Example of a false-positive finding. CAD analysis of this lesion (the lesion in red and yellow on the left image) indicated positive enhancement at the 50% and 100% thresholds. The image on the right is the same lesion in the same slice without the colour-coded (more ...)
Example of a false-positive finding. CAD analysis of this lesion (the yellow lesion in the right breast) indicated positive enhancement at the 50% and 100% thresholds. Pathology revealed a lobular carcinoma in situ, a high-risk lesion
In our study the readers were not blinded to the results of other imaging examinations such as mammography, ultrasound and previous MRI findings. All readers had access to patient history and clinical information, i.e. similar to the normal clinical setting. In the study by Kurz et al. who compared three different software systems in the evaluation of breast MRI, all three investigators were blinded to this information [10
Our study had several limitations. Time needed to fully assess one MRI examination of the breast was not objectified as in the study of Kurz et al. [10
]. They describe an average time needed to evaluate each exam of 7.0 min. Wiener et al. reported that most cases were interpreted with the use of an automatic post-processing program which generated colour parametric maps as well in less than 5 min [9
]. Although not objectified, the two breast radiologists reported shorter interpretation times with CAD assistance than with manual analysis.
Another limitation is that manual interpretation was not repeated by the two residents. Meinel et al. showed that inexperienced readers without the use of CAD assistance performed as expected for their level of experience and that these differences became less pronounced with the use of CAD assistance [11
]. In our study, no significant differences were found between the two experienced breast radiologists and the two residents together, or between any two readers separately with the use of CAD.
With respect to interpreting the very high sensitivities reported here for CAD-based analysis based solely on the thresholding of enhancement kinetics, it should be noted that a selection bias was introduced by only including data from patients with lesions proven by core or excision biopsy.
In summary, our findings suggest that CAD has the potential to improve the discrimination of benign from malignant breast lesions at 3.0-T MRI. Additionally CAD may decrease the heterogeneity of interpretation across radiologists of varying levels of experience in breast MR interpretation.