Optimisation analysis of pulmonary nodule diagnostic test based on deep belief nets

• Published in: IET image processing Vol. 14; no. 7; pp. 1227 - 1232
• Main Authors: Yang, Wei, Xia, Wenhua, Xie, Yuanliang, Mao, Shilong, Li, Rong
• Format: Journal Article
• Language: English
• Published: The Institution of Engineering and Technology 29.05.2020
• Subjects: actual detection process; belief networks; cancer; deep belief nets; deep belief network‐based diagnosis model; detection accuracy; detection effect; diagnostic effect; fault diagnosis; interference factors; learning (artificial intelligence); lung; lung cancer; medical image processing; missed diagnosis; model learning rate; multilayer nonlinear structure; nodular shape; optimisation analysis; patient diagnosis; previous clinical data; pulmonary nodule diagnostic test; pulmonary nodule phenomenon; pulmonary nodules; Special Section: Deep Learning for 3D Reconstruction; training effect; fault diagnosis; pulmonary nodules; detection effect; model learning rate; deep belief nets; lung cancer; training effect; diagnostic effect; detection accuracy; pulmonary nodule phenomenon; multilayer nonlinear structure; belief networks; learning (artificial intelligence); nodular shape; medical image processing; interference factors; previous clinical data; missed diagnosis; lung; deep belief network-based diagnosis model; optimisation analysis; actual detection process; cancer; pulmonary nodule diagnostic test; patient diagnosis
• ISSN: 1751-9659; 1751-9667; 1751-9667
• DOI: 10.1049/iet-ipr.2019.1022

At present, the rate of missed diagnosis of lung cancer is high. The reason is that the pulmonary nodule phenomenon cannot be effectively monitored due to various interference factors in the actual detection process. In order to improve the detection accuracy, this study combined with the actual situation to analyse the diversity of nodular shape and constructed a deep belief network-based diagnosis model for pulmonary nodules. At the same time, in order to improve the detection effect, this study sets the model to have multi-layer non-linear structure and analyses the previous clinical data to improve the model learning rate and training effect. In addition, in order to verify the performance of the model, the diagnostic effect of the model is studied by comparative experiments. The research shows that the model proposed in this study is higher than the traditional algorithm in detection accuracy, which can provide theoretical reference for subsequent related research.