Optimized DL Approach for Microcalcification Analysis in Digital Breast Tomosynthesis
K. Jasna, S. Albert Jerome · International Journal of Electrical and Electronics Engineering · 2024
Breast cancer is the primary cause of cancer-related deaths in women globally, and increased survival rates and effective treatment depend on early identification. One of the earliest indications of breast cancer is frequently Microcalcifications (MC), making their accurate classification vital in breast cancer diagnosis. In order to classify MCs using data from Digital Breast Tomosynthesis (DBT), this study suggests an effective Deep Learning (DL) algorithm. Stacked Long Short-Term Memory (LSTM) with Bayesian Optimization was developed and evaluated on the DBT slice. Various features associated with MCs are extracted from the DBT slices with the help of image processing techniques. Extensive Exploratory Data Analysis (EDA) was performed on the DBT dataset to extract key features and identify patterns related to MC classification. Preprocessing techniques were applied to enhance image quality and remove noise. The proposed LSTM model achieved an accuracy of 97.27% and demonstrated superior classification performance, achieving better accuracy than traditional methods. The approach improves breast cancer detection speed and accuracy by automating the categorization process, which presents a possible path for early diagnosis and better patient outcomes. These results advance DL methods in medicine, especially in terms of better radiographic imaging analysis for breast cancer screening.