Exploring Novel CNN Architectures for Deep Learning-Based Liver Tumor Occurrence Prediction

According to the authorized findings of,” Hepatocellular carcinoma (HCC)” a common in the human species, liver located, tumor,90 % of its occurrences demand any sort of support for predicting them in advance, if possible. For this purpose, Image Processing, Text Mining, and Deep learning techniques are handled with effective pre-processing methods in the recent research works to improve the results available so far. A majority of contribution by the researchers finds it challenging mostly in selecting suitable architectures for deep learning as most of the previous literature studies show the lacking of details on feasibility and justifiability in specifying the architectures. In this article, it is aimed at the primary goal of such studies as the first attempt on varying architectures in terms of profile and position of layers and observing the performance for recognizing the liver Tumor from the Image-data sets. The CNN architecture of neural network like AlexNet, VGGnet, ResNet, LeNet are standard architectures, based on this liver tumor dataset is implemented in customized architecture and its produced best performance with accuracy of 92.4 %. And It is compared with MeanderHandPD, Breast cancer and Liver Tumor Datasets, the proposed architecture shows higher performance for Liver Tumor images.