Comparison of Manual and Computer–aided Breast Cancer Biopsy Grading

Prognostic factors, such as estrogen or progesterone visualized in breast cancer biopsy slides, are crucial in the determination of the prognosis and subsequent treatment of a patient. However, the assessment of prognostic factors in biopsy slides and the comparison of results is difficult. Both manual and computer-aided assessment schemes are affected by error inducing variabilities. Moreover, manual and computer-based assessment schemes are not standardized. Here, I present BASS (Biopsy Analysis Support System), a computer-aided system which automatically detects breast cancer nuclei candidates in a biopsy image, classifies each nucleus using a neural network classifier, and computes an overall grade for the biopsy. The performance of BASS without user interaction is compared to the performance of two medical experts who use the same manual grading scheme as the system. The results indicate that strengths and weaknesses of the experts and the system compliment each other.