An Analysis of Convolutional Neural Network—Random Forest for Liver Cancer CT Scan Images

Cancer is an ailment which results from uncontrolled cell growth in the corpus. Meanwhile, liver cancer constitutes one of five categories cancer leading to the highest number of deaths, it is a malignant tumor that starts in the liver. Based on WHO research in 2015, liver cancer is responsible for over 700,000 out of the 9 million deaths resulting from cancer. Although several studies have classified cancer, there are no symptoms to suggest the presence of liver cancer. Therefore, this study used different types of machine learning methods namely Convolutional Neural Networks and Random Forests to classify liver cancer. The Convolutional Neural Networks is a prevalent used with a broad length of application domains. Meanwhile, the Random Forests method has also been applied in the classification. The Convolutional Neural Networks method was utilized in the early stages of the convolution section, while Random Forests was applied in the classification section. A dataset in form of a CT scan image implemented in the algorithm were obtained from the liver patients CT scan. This study aims to evaluate both performance and accuracy and to determine whether the renewal method is more accurate than the Convolutional Neural Networks in classifying the CT scan dataset for liver cancer patients by evaluating the performance results using the Convolutional Neutral Network-Random Forest. The method is also expected to provide higher accuracy for future studies; hence, more database can provide better results for predicting and classifying different diseases.