Enhancing Watershed Segmentation for Precise Mammogram Detection and Classification with the Application of Feedforward Neural Network Strategies

Mammography may detect breast cancer early. CAD systems increase tissue visualisation by processing medical images like digitally chest x-rays and digital mammograms. Digital mammograms are recommended for women over 40 without breast cancer symptoms. The CAD method mostly uses chest X-rays to diagnose issues early. Healthy breast tissue is protected by low-dose X-rays. Tracking and measuring breast cancer requires accurate computational tumour segmentation. The use of fully automated or CAD for breast cancer segmentation and categorisation has various obstacles. A hybrid two-stage segmentation approach is presented here. Since cancers are abundant under all settings, we first use a watershed transform to divide the picture into basins based on pixel density, which indicates tumour mass. Despite being sensitive to even tiny changes in image size, the number of areas is unreasonably large, and segment boundaries are typically inflexible. The second level set efficiently segments medical images because it naturally allows cavities, folds, splits, and merges. The division result is used as the curve's beginning point in the level-set procedure, simplifying and improving identification. The FNN model has the highest accuracy at 97.68%, showing that it is quite good at data classification. The FNN model has the greatest precision at 98.45%, outperforming both ATMF (93.68%) and MIAS (92.98%). Furthermore, the FNN model has an exceptional capacity to detect real negatives, as seen by its specificity level of 97.54%. This indicates that the model performs much better on the training data than on the test data.