End-to-end improved convolutional neural network model for breast cancer detection using mammographic data

Abstract

Any disease is curable if it is diagnosed at the early stages with the help of a little human effort. The disease breast cancer is the second leading cause of death among women after lung cancer. Mammography is one of the most mainstream clinical imaging modalities that are utilized for early recognition of breast cancer. Early breast cancer detection helps to alleviate unnecessary treatments as well as saving women’s lives. The speedy development in deep learning and some of the strategies of machine learning have invigorated abundant enthusiasm for their application to clinical imaging issues. This paper presents an improved convolutional neural network (CNN) model that consists of three convolutional layers where the starting layer searches for low-level features and the ending layer searches for high-level features. Two activation functions, that is, the Rectified Linear Unit and sigmoid functions, are utilized for the detection of breast cancer using digitized film mammograms from the Digital Database for Screening Mammography. The proposed convolutional neural system for identifying breast malignancy on mammogram imaging achieved praiseworthy execution on examination with prior models. The experimentation found that the model achieved has a true positive rate of 99% (accuracy = 97.20%, precision = 99%, true negative rate = 96%, F-score = 0.99, balanced classification rate = 0.975, Youden’s index = 0.95). The proposed improved CNN model can be used as a second opinion of doctors to detect breast cancer.

A novel bi-modal extended Huber loss function based refined mask RCNN approach for automatic multi instance detection and localization of breast cancer

“…Wu et al 26 2019 DCNN CBIS-DDSM AUC-0.895 Wang et al 27 2020 3D CNN ABUS SE-95% Min et al 28 2020 Mask R-CNN Mammogram TPR-0.90 and DCI-0.88. Kumar et al 29 2020 CNN Mammogram Ac-97.20%, PC-99%, TNR-96 and F-score-0.99. Ragab et al 30 2021 Multi DCNN Mammogram AC-76.01%.…”

“…The values of true positive rate (TPR) and average dice similarity index (DSI) for segmentation are calculated. Kumar et al 29 have developed an end-to-end improved CNN to predict the breast cancer. The model uses three CNN layers where the initial layer determines the low features and the last layers obtains the high features respectively.…”

Boosting Breast Cancer Detection Using Convolutional Neural Network

“…e dataset, Kaggle 162 H&E, was used for the proposed system [28]. Kaggle 162 H&E was also used by many researchers for similar kind of study [26,30].…”

“…Via our neural net, however, we can then transfer these patterns down and begin to identify more complex characteristics as we get deeper. is property ensures that CNNs are very effective at detecting objects in images [26].…”

“…Several previous studies have proposed using AI as well as CNN for image detection and healthcare monitoring [1,18,20,25,29]. But, for a medical side solution, which is around 60 percent for all class detection, 75 percent for only mass class, and 100 percent for only calcification, the accuracy percentage is too poor [26]. With the exception of only calcification argument, the accuracy of all argument and mass only argument can be further enhanced to obtain a better result [27].…”