Lung Function Decline Predicting Using Improved EfficientNet

Abstract

Pulmonary fibrosis (pf) is a common outcome of various lung diseases, with scarring of lung tissue as the main manifestation, and if the scope of involvement is extensive, it leads to a reduction in lung volume, a significant decrease in lung function, and a serious impact on the quality of life of patients. Using CT scanning to examine high-risk people is an effective means to find early lung cancer. With the development of technology, computer-aided diagnosis plays a very important role in the cancer diagnosis. We state the related work and proposed an improved EfficientNet. We choose Laplace Log Likelihood as our experiment metrics. The higher Laplace Log Likelihood is, the better performance the model will gain. We can see the result that our improved EfficienNet model owns the best performance with -6.89 Laplace Log Likelihood, which is 0.23, 0.22, 0.15 and higher than Resnet34, Resnet50 and EfficientNet respectively.