Among numerous artificial intelligence approaches, k-Nearest Neighbor algorithms, genetic algorithms, and artificial neural networks are considered as the most common and effective methods in classification problems in numerous studies. In the present study, the results of the implementation of a novel hybrid feature selection-classification model using the above mentioned methods are presented. The purpose is benefitting from the synergies obtained from combining these technologies for the development of classification models. Such a combination creates an opportunity to invest in the strength of each algorithm, and is an approach to make up for their deficiencies. To develop proposed model, with the aim of obtaining the best array of features, first, feature ranking techniques such as the Fisher's discriminant ratio and class separability criteria were used to prioritize features. Second, the obtained results that included arrays of the top-ranked features were used as the initial population of a genetic algorithm to produce optimum arrays of features. Third, using a modified k-Nearest Neighbor method as well as an improved method of backpropagation neural networks, the classification process was advanced based on optimum arrays of the features selected by genetic algorithms. The performance of the proposed model was compared with thirteen well-known classification models based on seven datasets. Furthermore, the statistical analysis was performed using the Friedman test followed by post-hoc tests. The experimental findings indicated that the novel proposed hybrid model resulted in significantly better classification performance compared with all 13 classification methods. Finally, the performance results of the proposed model was benchmarked against the best ones reported as the state-of-the-art classifiers in terms of classification accuracy for the same data sets. The substantial findings of the comprehensive comparative study revealed that performance of the proposed model in terms of classification accuracy is desirable, promising, and competitive to the existing state-of-the-art classification models.
Matched MeSH terms: Breast Neoplasms/classification
Breast cancer may be classified into luminal A, luminal B, HER2+/ER-, basal-like and normal-like subtypes based on gene expression profiling or immunohistochemical (IHC) characteristics. The main aim of the present study was to classify breast cancer into molecular subtypes based on immunohistochemistry findings and correlate the subtypes with clinicopathological factors. Two hundred and seventeen primary breast carcinomas tumor tissues were immunostained for ER, PR, HER2, CK5/6, EGFR, CK8/18, p53 and Ki67 using tissue microarray technique. All subtypes were significantly associated with Malay ethnic background (p=0.035) compared to other racial origins. The most common subtypes of breast cancers were luminal A and was significantly associated with low histological grade (p<0.000) and p53 negativity (p=0.003) compared to HER2+/ER-, basal-like and normal-like subtypes with high histological grade (p<0.000) and p53 positivity (p=0.003). Luminal B subtype had the smallest mean tumor size (p=0.009) and also the highest mean number of lymph nodes positive (p=0.032) compared to other subtypes. All markers except EGFR and Ki67 were significantly associated with the subtypes. The most common histological type was infiltrating ductal carcinoma, NOS. Majority of basal-like subtype showed comedo-type necrosis (68.8%) and infiltrative margin (81.3%). Our studies suggest that IHC can be used to identify the different subtypes of breast cancer and all subtypes were significantly associated with race, mean tumor size, mean number of lymph node positive, histological grade and all immunohistochemical markers except EGFR and Ki67.
Matched MeSH terms: Breast Neoplasms/classification*
OBJECTIVE: To examine the fine needle aspiration cytologic features of invasive lobular carcinoma of breast and to discuss problems that may occur in cytodiagnosis.
STUDY DESIGN: Fine needle aspiration cytologic smears from 21 cases of invasive lobular carcinoma (ILC) of breast were subjected to detailed cytomorphologic analysis. Features studied included pattern of cells, size of cells, nuclear placement, pleomorphism, presence of intracytoplasmic lumina (ICL) and signet ring cells.
RESULTS: Cellularity was generally moderate or high, and the pattern was predominantly or partly dissociated in 86% of cases. Rosettelike pattern was discerned in alveolar-type ILC. Cell size was usually small or intermediate, with nuclei placed eccentrically in most cases. ICLs with or without signet ring cells were present in 12 cases (57%).
CONCLUSION: A cytologic picture consisting of predominantly dissociated small or intermediate-sized tumor cells with eccentric nuclei, with some of the cells showing ICLs, is highly suggestive of ILC. Indian file pattern, another characteristic feature of ILC, is, however, focal and inconsistent. Variant patterns of ILC may show other cytologic features, such as rosettelike pattern (alveolar variant of ILC) or large cell pattern (pleomorphic variant of ILC) and may consequently be difficult to categorize on cytologic smears.
Matched MeSH terms: Breast Neoplasms/classification
Hormone receptor (HR) negative breast cancers are relatively more common in low-risk than high-risk countries and/or populations. However, the absolute variations between these different populations are not well established given the limited number of cancer registries with incidence rate data by breast cancer subtype. We, therefore, used two unique population-based resources with molecular data to compare incidence rates for the 'intrinsic' breast cancer subtypes between a low-risk Asian population in Malaysia and high-risk non-Hispanic white population in the National Cancer Institute's surveillance, epidemiology, and end results 18 registries database (SEER 18). The intrinsic breast cancer subtypes were recapitulated with the joint expression of the HRs (estrogen receptor and progesterone receptor) and human epidermal growth factor receptor-2 (HER2). Invasive breast cancer incidence rates overall were fivefold greater in SEER 18 than in Malaysia. The majority of breast cancers were HR-positive in SEER 18 and HR-negative in Malaysia. Notwithstanding the greater relative distribution for HR-negative cancers in Malaysia, there was a greater absolute risk for all subtypes in SEER 18; incidence rates were nearly 7-fold higher for HR-positive and 2-fold higher for HR-negative cancers in SEER 18. Despite the well-established relative breast cancer differences between low-risk and high-risk countries and/or populations, there was a greater absolute risk for HR-positive and HR-negative subtypes in the US than Malaysia. Additional analytical studies are sorely needed to determine the factors responsible for the elevated risk of all subtypes of breast cancer in high-risk countries like the United States.
Matched MeSH terms: Breast Neoplasms/classification*
The purpose of this study was to evaluate the diagnostic value of qualitative and semi-quantitative assessment of ultrasound elastography in differentiating between benign and malignant breast lesions. This prospective study was conducted in two tertiary medical centers. Consecutive B-mode ultrasound and real-time elastographic images were obtained for 67 malignant and 101 benign breast lesions in 168 women. Four experienced radiologists analyzed B-mode ultrasound alone and B-mode ultrasound combined with elastography independently. Conventional ultrasound findings were classified according to the American College of Radiology Breast Imaging Reporting and Data System classification. The elastographic assessment was based on qualitative and semi-quantitative parameters (i.e., strain pattern, width ratio, strain ratio). The sensitivity and specificity of combined elastography and conventional ultrasound were significantly higher than that of conventional ultrasound alone. The sensitivity, specificity, positive predictive value and negative predictive value was 97%, 61.4%, 62.5% and 96.8%, respectively, for conventional ultrasound and 100%, 93%, 99% and 90%, respectively, for combined technique. The semi-quantitative assessment with strain ratio and width ratio in elastography were the most useful parameters in differentiating between benign and malignant breast lesions. Cut-off point values for width ratio of more than 1.1 and strain ratio of more than 5.6 showed a high predictive value of malignancy with specificities of 84% and 76%, respectively (p
Matched MeSH terms: Breast Neoplasms/classification