Sistemas inteligentes para diagnósticos automatizados do câncer de mama

Abstract

Breast neoplasms are frequent cases among the various abnormalities related to the disordered growth of human cells. Breast cancer is one of the most common types in women and has a high mortality rate. For this reason, early diagnosis becomes essential to combat the disease. Using the WDBC database and WEKA Machine Learning platform, a system for automated breast cancer diagnosis has been proposed. Five different classification algorithms have been configured and applied to obtain a reliable system to support breast cancer diagnosis. The database was properly treated and filtered before the algorithms were applied and the result obtained could be analyzed by the average system accuracy: 15% higher than the current diagnostic methodology (BI-RADS). The breastfeeding technique has shown promise and has the potential to dramatically improve the level of assertiveness in the diagnosis of breast cancer. It was necessary to replicate the experiment and analyze its results by comparing performance levels by applying new databases in order to find the most appropriate algorithm and configuration, ensuring that the system became increasingly robust and reliable.