WOMT: Wasserstein Distribution Based Minimization of False Positives in Breast Tumor Classification Using Deep Learning

Abstract

Women’s life suffered and killed by invasive cancerous tumors is the most frequently highlighted header in many newsletters since 2010. There can be many. By their nature, invasive tumors spread from tissue to tissue and abduct themselves to cause new tumors. In many of the human biological parts, CT scan has given an objective approach to the early and successful detection of cancerous tumors. However, there were cases where diagnosis with CT scan images failed, resulting in many false positives. In economically backward countries, these false positives raised the notifying concern of women who cannot afford multiple diagnostic tests. Due to changes in biological metabolism, the growth of breast fat in women to considerably abnormal size is the main cause of false positives. In many of the images under study, this huge thick breast fat layer led to the rise of misclassification rate with false positives. Rendering societal help requires a precise mechanism that can reduce the false positives at an initial diagnosis. The proposed method introduces a novel constraint-based algorithm to classify a mammogram image as cancerous, aiming to reduce false positives. The proposed deep learning algorithm WOMT is trained with Wasserstein Distribution constraints that are derived from the mass transfer of cancerous patches to non-cancerous patches. The experimental simulations with a deep learning model trained with these constraints resulted in reduced false positives.

Keywords

• Bhattacharya's similarity
• CNN
• false positives
• optimal mass transfer
• Wasserstein distribution