Basal lamina visualization using color image processing and pattern recognition.

In histologic assessment, the absence of basal lamina is a useful feature for distinguishing invasive malignancy from benign and in situ lesions. As this feature is not possible to assess in routine H&E sections, pathologists have instead relied on histochemical and immunohistochemical stains to show components of the basal lamina such as laminin or type IV collagen. Standard image-processing software with the necessary image-processing toolbox (Matlab v5, Mathworks, Natick, MA) was used in a unique combination of color image processing and pattern recognition techniques to accentuate the collagenous stroma surrounding glands, which approximates basal lamina, in a series of benign, in situ, and invasive breast proliferations. Distinct differences in pattern were found between benign and invasive lesions, and also between in situ and malignant lesions, corresponding to that observed with type IV collagen immunostaining. Compared with immunostaining, this computer-generated method had a sensitivity of 0.96, specificity of 0.89, positive predictive value of 0.92, negative predictive value of 0.89, positive likelihood ratio of 9.1, and negative likelihood ratio of 0.042. Digital image processing serves as a less expensive and faster way of visualizing basal lamina and represents a useful adjunct to identify invasive malignancy in routinely stained sections. In addition, digital visualization of basal lamina is readily amenable to quantitative assessment, and the method provides a basis for the development of computer-based cancer diagnosis.

An automated diagnostic system for tubular carcinoma of the breast--an overview of approach and considerations.

A computer-based automated histopathology recognition system was developed to distinguish benign from malignant lesions. Tubular carcinoma of the breast, which has several reactive and neoplastic mimics, was selected as a model. Archival stained tumour sections from the United Kingdom National External Quality Assurance Scheme for breast pathology and supplementary material from external pathologists formed the study population. A diagnostic process similar to that employed by the histopathologist was adopted, viz, low-power feature extraction and analysis by cluster/glandular groupings followed by high-power confirmation. To circumvent problems of stain variability, greyscale quantisation of images was achieved through Karhunen-Loeve transformation with results suggesting that histological stains provide information primarily through contrast and not colour. Mean nearest neighbour and variance of cell nuclei distances were found to be 100% effective in distinguishing images which contained diffuse tumour, and no clustering. Gaussian smoothing followed by minimum variance quantisation allowed segmentation of gland clusters. Perona-Malik nonlinear diffusion filter employed prior to intensity thresholding and morphological filtering was 92% (7330/7973) effective in segmenting individual glands. In a set of 62 benign and 52 malignant gland clusters, the features found to discriminate tubular carcinoma from benign conditions included > 20% of glands with sharp-angled edge, cluster area > 150,000 pixels, ratio total gland area:total cluster area < 0.14, > 60 glands per cluster and the ratio average malignant gland area:benign gland area < 0.5. Suspicious clusters were subjected to high-power feature analysis for nuclear morphology, nucleoli detection and basement membrane assessment. Watershed thresholding achieved nuclear segmentation and nuclear area > 1.3x mean benign nuclear area was found to have a malignant likelihood ratio of 14.5. Progressive thresholding was used to detect nucleoli. Basement membrane was accentuated by colour segmentation and demonstrated 0.96 sensitivity, 0.89 specificity and 0.92 positive predictive value for distinguishing malignancy.