Abstract
Lung medical image recognition plays an integral role in today's clinical diagnosis. This process involves finding valid recognition parts in lung CT images and classifying them as normal and abnormal. although many breakthroughs have been made in image classification and lung CT image recognition, current models are mostly for 2D lung images. This paper uses convolutional neural networks (CNN) to extract features for image recognition of infected and uninfected pneumonia virus. Based on the characteristics of the dataset, this paper introduces dynamic learning rate and defines an early stopping strategy to compensate for the traditional manual parameter adjustment methods which are time-consuming and consume a lot of computing power. By comparing with other methods, the model proposed in this paper is able to produce very competitive results in a shorter time, which is exciting.
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call