Design of Cancer Detection System Based on CNN Model and Virtual Reality with NLP Voice Output

Abstract

Nowadays, online medical consultation has become very popular. However, online consultation can only solve some minor diseases such as colds and coughs, but the detection of cancers needs to be improved. This project implements the design of a cancer detection system based on the CNN model and virtual reality. This project uses the CNN and regression models for training a mature network to detect prostate cancer, lung cancer and skin cancer. The patient who comes to check uploads the corresponding picture or blood test report to determine whether the patient has the corresponding condition. Then use the NLP model to voice output diagnosis results. As a result, the voice broadcasts whether the patient has cancer, making the humancomputer interface more friendly. After long enough training time, two CNN models and one regression model showed high scores. The experiment results to detect whether the patient has the corresponding cancer are efficient, owing to the accuracy of the test of the three models is above 95%. And when inputting the patient’s lung CT and skin details, the location of tuberculosis can be found quite accurately and whether the patient has skin cancer. Combined with virtual reality technology, it depicts models including wards, CT rooms, diagnosis rooms and supermarkets, successfully creating a friendly online hospital.