Mammogram classification using VGG-16 architecture

The breast image x-ray is done to obtain the digital mammogram images acquired as input for the proposed model. Then the data augmentation technique is applied to a small dataset image to manipulate multiple sets of images. The breast cancer diagnosis includes preprocessing pre-processing the Deep C...

Full description

Saved in:
Bibliographic Details
Main Authors: Sivanantham, E., Epsiba, P., Gopi, B., Solainayagi, P, Umapathy, K., Kumar, S. Mohan
Format: Conference Proceeding
Language:English
Subjects:
Artificial neural networks
Datasets
Digital imaging
Freezing
Image acquisition
Image manipulation
Mammography
Mathematical models
Medical imaging
Parameters
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The breast image x-ray is done to obtain the digital mammogram images acquired as input for the proposed model. Then the data augmentation technique is applied to a small dataset image to manipulate multiple sets of images. The breast cancer diagnosis includes preprocessing pre-processing the Deep Convolutional Neural Network (DCNN) with Transfer Learning (T.L.) method aiding the healthcare center in automated digital medical imaging technology. Our experiment used a pre-trained VGG-16 model, which is fine-tuned by freezing some of the layers to avoid over-fitting because our dataset is minimal. The network image input shape with its pixel size is 224x224x3. The number of filters we employ doubles roughly at every step or through each stack of convolutional layers. The primary drawback was that the number of parameters to be learned was enormous. Hence, fine-tuning of the VGG-16 Net transfer model reduces the variable parameters in the CONV layers and improves computation time.
ISSN:0094-243X
1551-7616
DOI:10.1063/5.0111085