A Distinctive Explainable Deep Learning Framework to Predict Pulmonary Function at Chest CT Scan
R. Renugadevi, M. Rajeswari, Anitha Roy, R. Rajalakshmi, G. Ramkumar
Abstract
Monitoring lung cancer patients' pulmonary function properly, particularly previous to surgery, is crucial from a clinical standpoint. This information can aid doctors in keeping tabs on their patients before and after operations, estimating how their lungs will respond to the procedure, and devising strategies to speed up their recoveries. The purpose of this study is to compare CT pulmonary involvement with PFT results in very ill COVID-19 patients who have been discharged from the hospital. We utilized a program to mechanically define ROIs across the pulmonary bronchi, lobes, as well as whole lung. Following that, we built a model using Multilayer Perceptron to evaluate pulmonary function based on radiomics characteristics. To evaluate the efficacy of the model, the findings of the machine learning were analyze to the gold standard of medical guidelines for lung function. Our findings demonstrated the machine learning model's efficacy in making FC and MVC forecasts.