Breast cancer is one of the most common cancers in the world, and is a very dangerous cancer among all cancers which mainly affect the woman, and hence leads many researchers in medical field to work towards its early diagnosis. There are many ways to detect breast cancer like, Biopsy which is the commonly used diagnosis method. Computer-Aided Diagnosis (CAD) is a capable tool for the pathologists for diagnosing cancer, which points out the cancerous cells. The CAD system includes the pre-processing of feature extraction, and feature selection. In this research paper, deep learning neural network technique is used to recognize breast cancer from the BreakHis dataset. This method firstly applies the reshaping of data and normalizing the data. After completion of these methods, different algorithms are used. These methods contain data augmentation layers, dropout layers, and dense layers. For this purpose, the dataset is split up into two groups, one containing 70% for training phase and the other containing 30% for testing phase for the images present in the BreakHis dataset of different magnification factors (40x, 100x, 200x, 400x). The performance is given in the accuracy value of different methods. The result obtained for the method having one data augmentation and more dense layers, the accuracy value is found to be high after the completion of all classifications of benign and malignant The proposed method achieves a higher accuracy for the image extraction and feature selection, and significantly improves the overall accuracy of classification.
BreastNet: Design and Evaluation of a Deep Learning model for recognizing Breast Cancer from Images
2022-12-01
689523 byte
Conference paper
Electronic Resource
English
Recognizing occluded objects in SAR images
Tema Archive | 2001
|Recognizing occluded objects in SAR images
IEEE | 2001
|