|
Challenges and Solutions in CNN based Diagnostic Techniques on Breast Mammographic images: A Survey
|
|
| Author:
|
PRATHEEPKUMAR P, DR.V.MARY AMALA BAI, DR.GEETHA G NAIR
|
| Abstract:
|
Breast cancer is the public cancer type among female cancer patients. Mammography is one of the widely used diagnostic modalities in the timely detection of breast cancer. Even though the former statement is true, mammographic images are subjected to various data issues while doing deep leaning based classification. Class imbalance and insufficient datasets are the major bottle necks in the classification of any medical image. This study is an attempt to find various methodologies adopted by researchers to overcome these issues in the classification of mammogram images for breast cancer diagnosis. This study focuses on reviewing solutions for two major challenges viz. class imbalance, and insufficient sum of data samples in the training dataset that exists in various Convolutional Neural Network (CNN) based Mammogram classification. Transfer Learning, Data sampling, Data Augmentation are some of the successful methods implemented to overcome small dataset and class imbalance problems.
|
| Keyword:
|
class imbalance, small dataset, Convolutional Neural Network, classification, breast cancer diagnosis, mammogram
|
| EOI:
|
-
|
| DOI:
|
https://doi.org/10.31838/ijpr/2020.12.02.421
|
| Download:
|
Request For Article
|
|
|