In:
International Journal of Recent Technology and Engineering (IJRTE), Blue Eyes Intelligence Engineering and Sciences Engineering and Sciences Publication - BEIESP, Vol. 8, No. 4 ( 2019-11-30), p. 7451-7454
Abstract:
Human beings are affected by several diseases nowadays. All those diseases are healable with minimal amount of treatment when they are identified at its early stage. Several patients were not serious enough in diagnosing the disease initially, which makes the disease incurable for the patient lifelong. Hence in recent days number of death rates are getting increased. Cancer is the most dangerous diseases. Among several types of cancers women are mostly affected by breast cancer. In most of the developing and under developing countries breast cancer is the most prominent reason for women mortality. It is also curable when it is identified at its starting stage. During the later stages the cancer cells will be disseminated all over body hence it is difficult to remove it completely. Hence it has to be identified at its initial stage in order to give best treatment for the patient at right time. In this paper , Convolution Neural Network (CNN) a deep learning model is proposed for the investigation of breast cancer images for finding whether the person is affected by cancer or not. In the proposed work , features from images are extracted using convolution layers and then it is passed to the fully connected layer where it classifying the images as either malignant or benign. Experiments using standard benchmark datasets for the proposed CNN Model and standard Visual Geometry Group Network (VGGNet) has been conducted to measure its performances. From the results ,it is clear that CNN outperformed with the accuracy of 86.32% when compared to VGGNet which provides only 50% accuracy for the identification of breast cancer.
Type of Medium:
Online Resource
ISSN:
2277-3878
DOI:
10.35940/ijrte.2277-3878
DOI:
10.35940/ijrte.D5317.118419
Language:
Unknown
Publisher:
Blue Eyes Intelligence Engineering and Sciences Engineering and Sciences Publication - BEIESP
Publication Date:
2019
detail.hit.zdb_id:
2722057-6
Permalink
