DF-dRVFL: A novel deep feature based classifier for breast mass classification

Abstract Amongst all types of cancer, breast cancer has become one the most common cancers in UK threatening millions people’s health. Early detection plays a key role timely treatment for morbidity reduction. Compared to biopsy, which takes tissues from lesion further analysis, image-based methods are less time-consuming and pain-free though they hampered by lower accuracy due high false positivity rates. Nevertheless, mammography standard screening method its efficiency low cost with promising performance. Breast mass, as palpable symptom received wide attention community. As result, past decades have witnessed speeding development computer-aided systems that aimed at providing radiologists useful tools mass analysis based on mammograms. However, main issues these include require enough computational power large scale datasets. To solve issues, we developed novel classification system called DF-dRVFL. On public dataset DDSM more than 3500 images, our best model deep random vector functional link network showed results through five-cross validation an averaged AUC 0.93 average $$81.71\%$$ 81.71 % . sole learning methods, increased 0.38. state-of-the-art better performance considering number images evaluation overall accuracy.