Lung Pathology Using Artificial Intelligence Analysis of Ultrasound Images: A Survey

This paper first studies and analyses the Artificial Intelligence algorithms that can potentially be used to identify various diseases from medical images and scans. Then researches the related work in this area to find a suitable Machine Learning model that can be used to identify Interstitial Syndrome in Lung Ultra Sound images. By researching similar applications of the latest Machine Learning algorithms, this team of researchers are adopting a Vision Transformer system developed on a Convolutional Neural Network-based algorithm. The later has proven to be efficient in the area of UltraSound image processing in general and more flexible to configure and parameterize in the training process. The Vision Transformer system is a segmentation-based model that divides the input image to smaller subset of images of the same size. By merging the adjacent small patches into bigger layers, bigger windows to perform self-attention on, the system extracts the points of interest out of these frames.