relation: https://khub.utp.edu.my/scholars/15472/
title: Breast Cancer Detection by Hybrid Techniques based on Deep Learning Networks
creator: Azlan, N.A.N.
creator: Elamvazuthi, I.
creator: Tang, T.B.
creator: Lu, C.-K.
description: One of the cancers that gives high fatality rates in human life is breast cancer. The current method used to detect breast cancer needs radiologists, which makes it costly and time-consuming. A possible solution is to detect it early, which can be done by computer-aided diagnosis technologies. An end-to-end system that could automatically detect breast cancer is described in this paper. From the mammographic images, it was first undergone the pre-processing stages for noise elimination. The law's mask was then applied to the preprocessed image to filter out the secondary features further. The filtered image was segmented by the active contour to obtain the breast region before being fed into a deep convolutional neural network for feature extraction. Principle Component Analysis (PCA) technique was then applied to select the necessary features as input to the Support Vector Machine (SVM) for determining the class of cells (normal or abnormal). Lastly, k-fold cross-validation techniques were executed to validate the results and obtained the average reading for both training and testing datasets. The proposed system was tested on the Digital Database for Screening Mammography (DDSM) and Mammographic Image Analysis Society (MIAS) dataset, and attained 97.50, 96.67, 98.33, and 0.99 for accuracy, sensitivity, specificity, and area under curve, respectively. © 2021 IEEE.
publisher: Institute of Electrical and Electronics Engineers Inc.
date: 2021
type: Conference or Workshop Item
type: PeerReviewed
identifier:   Azlan, N.A.N. and Elamvazuthi, I. and Tang, T.B. and Lu, C.-K.  (2021) Breast Cancer Detection by Hybrid Techniques based on Deep Learning Networks.  In: UNSPECIFIED.     
relation: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85124147072&doi=10.1109%2fICIAS49414.2021.9642670&partnerID=40&md5=38a8d6e7bab51ac44be03a5ec0bb7256
relation: 10.1109/ICIAS49414.2021.9642670
identifier: 10.1109/ICIAS49414.2021.9642670