Indeterminate fine-needle aspiration of the breast

Application of decision tree-based ensemble learning in the classification of breast cancer

As a common screening and diagnostic tool, Fine Needle Aspiration Biopsy (FNAB) of the suspicious breast lumps can be used to distinguish between malignant and benign breast cytology. In this study, we first review published works on the classification of breast cancer where the machine learning and data mining algorithms have been applied by using the Wisconsin Breast Cancer Database (WBCD). This work then introduces useful new tools, based on Random Forest (RF) and Extremely Randomized Trees or Extra Trees (ET) algorithms to classify breast cancer. The RF and ET strategies use the decision trees as proper classifiers to attain the ultimate classification. The RF and ET approaches include four main stages: input identification, determination of the optimal number of trees, voting analysis, and final decision. The models implemented in this research consider important factors such as uniformity of cell size, bland chromatin, mitoses, and clump thickness as the input parameters. According to the statistical analysis, the proposed methods are able to classify the type of breast cancer accurately. The error analysis results reveal that the designed RF and ET models offer easy-to-use outcomes and the highest diagnostic performance, compared to previous tools/models in the literature for the WBCD classification. The highest and lowest magnitudes of relative importance are attributed to the uniformity of cell size and mitoses among the factors. It is expected that the RF and ET algorithms play an important role in medicine and health systems for screening and diagnosis in the near future.

Implementation of DNA cytometric measurements in fine-needle aspiration biopsy diagnostics of breast disease

BACKGROUND

For this study, the diagnostic protocol included fine-needle aspiration biopsy (FNAB) and DNA cytometric measurements. The objective of the study was to improve the diagnostic sensitivity of FNAB.

METHODS

Sixty-eight FNAB samples were stained using the Feulgen stain, and DNA histograms were produced by selecting the nuclei of cell groups and free cells separately.

RESULTS

Among 28 samples that were classified as definitely benign (n = 17 samples) or atypical but benign (n = 11 samples), DNA cytometry generally showed diploid/peridiploid peaks. Three of those samples were associated later with carcinoma. In those samples DNA cytometry did not improve sensitivity. Among moderately atypical samples (n = 17 samples), 8 samples were diagnosed later as carcinoma. DNA cytometry helped the diagnostic procedure in 62.5% of those samples. Among 21 samples that were classified as highly suspicious (n = 10 samples) or definitely malignant (n = 11 samples), DNA cytometry generally showed atypical DNA histograms (20 of 21 histograms), and DNA cytometry supported the diagnosis of carcinoma in 95.2%. Histograms that were based on free cells frequently were more abnormal compared with histograms that were based on cell groups. Histologically verified benign lesions also could show abnormalities in DNA histograms. Accepting wider gates than are used normally for primarily Feulgen stained samples on these restained samples resulted in improved specificity, efficiency, and predictive values.

CONCLUSIONS

DNA cytometry has a potential to support the differential diagnosis of breast lesions, and sampling of free cells increases sensitivity. Benign breast lesions (fibrocystic disease, fibroadenoma) included DNA-cytometrically abnormal cell clones and showed tendencies toward polyploidy, which should be included in the diagnostic criteria.

Image analysis of irregularity of cluster shape in cytological diagnosis of breast tumors: Cluster analysis with 2D-fractal dimension

To establish diagnostic criteria using comparison of cell cluster shapes, between benign and malignant tumors, breast tumors demonstrating weak cellular atypia in low grade invasive ductal carcinoma (IDC) were compared. Fine-needle aspiration (FNA) specimens of breast tumors were obtained from 37 patients. Among these, 16 were histologically diagnosed as IDC low-grade and the other 21 as benign fibroadenoma (FA). For evaluation, we examined 740 clusters from these 37 FNA specimens. Nine image morphometric parameters were studied, including the cluster area, circumference, maximal length, maximal breadth, ratio of length to breadth, cluster roundness, cluster size, and the edge and distribution image fractal dimensions for cluster analysis. We evaluated the irregularity in cell cluster shape using fractal dimension analysis, and determined the correlation to cluster size. The irregularity in the IDC cluster shape was higher than that in the FA cluster shape. However, six cases (28.5%) of 21 FA clusters showed high fractal dimensions similar to those for IDC. The clusters were classified by cluster analysis into three types: IDC clusters, FA with irregular cluster shape, and FA with no irregular clusters. The average cell cluster area of the FA with irregular shape was found to be about three times larger than that of IDC clusters. When the differential diagnosis between IDC and FA is difficult, it is important to focus on irregularities in the shape and on overall size of the cell clusters. For accurate diagnosis, the cell cluster shape is as important as the individual cellular atypia.