Evaluation of elastosis in breast cancer.

The possibly prognostic value of elastosis in breast carcinoma has usually been visually evaluated, using subjective gradings whose variability has led to contradictory conclusions. We used a Magiscan 2 automated image analyzer to perform objective measurements on elastosis on a set of 52 slides that were also subjectively evaluated. The wide range of parameters measurable by the automated system seemed likely to successfully deal with the lack of homogeneity of elastosis that causes problems in subjective evaluations. The homogeneity parameters (percentage distributions of elastosis in each slide) were processed by factorial analysis. On the plane defined by the first two factors, the principal components analysis produced a clustering of cases that was, on the whole, in concordance with the subjective evaluation.