Early lung cancer detection and classification integrated equipment based on deep learning and system

Abstract

The invention relates to the field of image analysis, and particularly relates to an early lung cancer detection and classification integrated system based on deep learning. The system comprises a lung CT image acquisition module, a CT image preprocessing module, a lung image recognition module, a lung nodule measurement module and a lung cancer screening module. The lung image recognition moduledetects and recognizes lung nodules and lung nodule holes in the lung images based on a DSSD_ Xception model, and intercepts lung nodule areas; the lung nodule measurement module is used for measuringlung nodules and lung nodule hole sizes based on the length-width ratio of the connected component bounding rectangle, and outputting a measurement result; and the lung cancer screening module is used for screening early lung cancer based on a preset screening standard according to the measurement result of the lung nodule. According to the invention, the type and probability of early lung cancercan be automatically detected, and the type, size, position and malignant degree of pulmonary nodules can be accurately identified.

Description

technical field [0001] The invention relates to the field of image analysis, in particular to an integrated device and system for early lung cancer detection and classification based on deep learning. Background technique [0002] Lung cancer is a malignant tumor with the highest morbidity and mortality. Its early symptoms are not obvious, and advanced lung cancer is extremely difficult to cure. Therefore, early detection of lung cancer is the main means to prolong the survival of patients and reduce mortality. Early lung cancer is often manifested as pulmonary nodules, which are large in number, small in size, low in contrast, and easily confused with other tissues; currently, lung CT examinations are mainly used to diagnose pulmonary nodules. size and shape to judge its malignancy, while the number of CT images of a lung cancer patient is at the level of hundreds, and the doctor’s workload is heavy. The accuracy of the evaluation will be affected by subjective factors such...

AI Technical Summary

Problems solved by technology

Early lung cancer is often manifested as pulmonary nodules, which are large in number, small in size, low in contrast, and easily confused with other tissues; currently, lung CT examinations are mainly used to diagnose pulmonary nodules. size and shape to judge its malignancy, while the number of CT images of a lung cancer patient is at the level of hundreds, and the doctor’s workload is heavy. The accuracy of the evaluation will be affected by subjective factors such as the doctor’s experience, fatigue, and personal emotions. The resources of experts and doctors in different regions are unevenly distributed, and traditional diagnosis and treatment methods are prone to missed and misdiagnosed cases

[0003] In recent years, with the improvement of computer computing power and the stepwise growth of data volume, deep learning technology has developed rapidly and has been widely used in the medical field. A lot of research work has been done on the Internet, and its main technical steps include data preprocessing and detection of pulmonary nodules using convolutional networks, but these methods mainly focus on identifying lung nodule regions and non-pulmonary nodule regions. On the other hand, the type and probability of lung nodule canceration are closely related to the type, size, shape and other characteristics of lung nodules. The traditional detection convolutional network can only get the total probability of lung cancer. , it is impossible to distinguish the specific types of lung cancer, so doctors are often required to evaluate, and the integration of early lung cancer detection process has not been realized

Images