Prediction of Predisposing Factors to Breast Tumours Using Artificial Neural Network
Mfoniso Udonkang, Eluwa A. Mokutima, Anietie M. Archibong, David Onwineng, Blessing Anku · African Journal of Pure and Applied Sciences · 2024
Breast tumour occurrences are increasing among women worldwide. The predisposing factors remain unclear because of their heterogeneity, hence the need for computer-aided approach in prediction. This study investigated the occurrence of breast tumours; connective tissue changes; estrogen receptor (ER), progesterone receptor (PR), human epidermal growth factor 2 (HER2), cytokeratin 7 (CK7), and Kiel-67 (Ki-67) expressions; and used artificial neural network (ANN) to predict the important predisposing factors. Data from ninety-six women aged 13-78 years and fifteen breast biopsies from the Histopathology Laboratory, University of Calabar Teaching Hospital were obtained. The formalin-fixed-paraffin-wax-embedded tissues were stained with hematoxylin and eosin (H&E), van Gieson, aqueous beetroot, and Colloidal iron-PAS. Ten malignant tissue blocks were immunohistochemically-stained for ER, PR, HER2, CK7, and Ki-67. Data were analyzed with Chi-square and ANN of Statistical Package of Social Sciences (SPSS) software. In the results, breast cancers were 35(36.5%) of the 96 womenand age was a predisposing factor (p=0.001). Beetroot demonstrated stroma hyalinization. The malignant tumours were mostly HER2+ 8(80%). Chi-square showed HER2+, CK7+, and Ki-67+-tissues had collagen and mucin depositions (p=0.001). Using ANN, mucin deposition, stroma hyalinization, CK7, HER2+, and age were the most important predisposing factors (p=0.006). Breast cancer is characterized mostly by acid mucin deposition and stroma hyalinization