An Enhanced Vision Transformer Model in Digital Twins Powered Internet of Medical Things for Pneumonia Diagnosis
Lumin Xing, Wenjian Liu, Xiaoliang Liu, Xin Li
Abstract
The computer-aided system and chest X-ray images play an important role in the diagnosis of pneumonia, which are the main way of pneumonia diagnosis. The traditional deep learning models have achieved some success in medical images, which captures the potential features of the image by continuously sliding the fixed convolution kernel. The disadvantage of this method is that it cannot effectively capture the long-distance dependencies in the image, and it does not have the ability of dynamic adaptive modeling. Next, the high-quality labeled data of chest X-ray images are very scarce. In order to achieve high-quality artificial intelligence diagnosis, a large number of high-quality annotated chest X-ray images are required. In this work, based on technologies such as Internet of Medical Things (IoMT) and Digital Twins, we built an intelligent IoMT platform for automatic diagnosis of pneumonia. For the digital twin of the lung, we propose an enhanced vision transformer model (EVTM) for analyzing chest X-ray images to determine whether the patient is infected with pneumonia. The EVTM model utilizes the vision transformer for training and inference on chest X-ray images. Then the EVTM model uses the variational autoencoder model for data augmentation, so that the amount of chest X-ray images meets the training requirements of the model. Finally, we conducted extensive experiments on the standard chest X-ray image dataset to verify the effectiveness of the EVTM model.