Breast Cancer Detection using Convolutional Autoencoder with Hybrid Deep Learning Model

The most deadly cancer among women in world is Breast cancer (BC). The early identification of malignancy helps in the disease diagnosis and it can help strongly to enhance the survival rate. With the rapid development of modern medical science and technology, medical image classification has become a more and more challenging problem. However, in most traditional classification methods, image feature extraction is difficult, and the accuracy of classifier needs to be improved. Therefore, this paper proposes a high-accuracy medical image classification method based on Deep Learning (DL) which is called Convolutional Neural Network (CNN). This research focused to create a hybrid DL model with a single test that subjected at inference and even adopted VGG16 as Autoencoder for Transfer Learning (TL) that performs an image analysis task such as segmentation and even set as an adaptor for pre training the model. The VGG16 is used to train from the source dataset and perform as the adaptors that have been optimized at the testing stage using a single test subject for effective computation. Therefore, this study has been used CNN with Bi-Long Short Term Memory (Bi-LSTM) method to extract features from Ultrasound Images of Breast for cancer detection database that involves images to benign as well as malignant breast tumors for performing analysis of the unsupervised images. The evaluated results showed that accuracy of VGG16 with CNN-Bi-LSTM has high accuracy as 98.24% indicates hybrid DL with VGG16 models have appropriate in detection and classification of the breast cancers precisely.