Web based Liver Cancer CAD system for Deep Learning using Convolution Neural Networks

Abstract

Liver cancer can be broadly classified into two types. They are primary and secondary liver cancers. Hepatocellular Carcinoma (HCC) is the most common type of primary liver cancer. Detection of liver cancer from CT images is a sophisticated work for the radiologist. A web based Liver CT image processing for online processing without losing vital information is the objective of this paper. The volume and dimensions to be considered in liver CT images may also be reduced for efficient feature extraction and faster diagnosis of Liver cancer. The most common CT image file format DICOM or NifTi is converted to PNG file before the preprocessing step. The commonly used preprocessing techniques considered before introducing the 2D or 3D feature map to a CNN are also discussed here. The changes in gray levels are used to detect lesions in liver tissues with HCC. The imbalanced data set gives us poor classification accuracy. Data augmentation must be carried out using addition of noise or morphological operations in these CT images in case the sample size is less for the training dataset.