| Objective: Lung cancer is the malignant tumor with the highest incidence and mortality in China.Smoking is one of the risk factors of lung cancer incidence,and there are still some problems to be solved,such as how long is the time interval between tobacco consumption and the incidence of lung cancer,or is the length of the time interval has any differences under different conditions.However,the smoking-related data collected through the high-risk assessment questionnaire for lung cancer is very limited and does not answer the above questions very well.Therefore,tobacco consumption and the incidence of lung cancer were analyzed globally in my study,focusing on the impact of tobacco consumption on the burden of lung cancer and its Latent-Effect.The onset of lung cancer is insidious,the early symptoms are not obvious,and the clinical stage at the time of consultation greatly affects the prognosis of patients.Therefore,the secondary prevention of lung cancer,its early detection,early diagnosis and early treatment is very important.Studies have shown that Low-dose Computed Tomography(LDCT)for lung cancer screening is an effective method for early diagnosis and early treatment of lung cancer.This study was based on the Cancer Screening Program in Urban China(Can SPUC)to screen lung cancer among Chinese urban residents who are assessed as being at high risk for lung cancer.The characteristics of the population that detected positive nodules and the imaging characteristics of the nodules were analyzed.The benign and malignant conditions of the positive nodules detected by the subjects were collected through follow-up tracking,so as to evaluate the effectiveness of LDCT in screening for lung cancer in high-risk populations of urban lung cancer.Based on this,the imaging indicators of pulmonary nodules were used to pass The Artificial neural networks(ANN)was used to construct predictive models of the malignancy of small pulmonary nodules,so as to facilitate the screening effect of lung cancer screening.It can also provide a basis for the diagnosis and treatment process of lung cancer screening baseline imaging examination.Method: Then data on lung cancer incidence from 1993 to 2012 in 30 countries around the world and tobacco consumption data from 1968 to 2012 in each country were collected as a new pooled database to explored the impact of tobacco exposure on global lung cancer,after analyzed then time interval from tobacco consumption to the lung cancer occurrence,explored the Latent-Effect of tobacco consumption on lung cancer incidence.Then selected the high-risk assessment of lung cancer and lung cancer screening data from September 2012 to September 2016 in the Can SPUC,analyzed the population characteristics of positive nodules and the imaging characteristics of nodules,and participates in lung cancer screening.Participation rates,population characteristics of positive nodules detected in lung cancer screening subjects,and nodular imaging characteristics were analyzed;the subjects were all Chinese urban residents without lung cancer-related symptoms,aged 40-69 years.In addition,by collecting the results of the follow-up of the Can SPUC from May 2017 to August 2018,the two screening outcome indicators of the Can SPUC "positive nodules" and "suspected lung cancer" were used for evaluation.Then,a back-propagation algorithm in ANN was used to establish prediction models for the malignancy of pulmonary nodules using the imaging indicators of lung cancer screening.By trying a combination of various inputs and parameters,a prediction model with the best prediction effect was selected.This study used SAS 9.4 software for statistical analysis and Matlab 2017 b to build ANN prediction models.Result: Focusing on the analysis of the impact of tobacco consumption on the incidence of lung cancer worldwide,the analysis of 45-year tobacco consumption and 20-year incidence of lung cancer in 30 countries worldwide has proved that tobacco consumption does have a Latent-Effect on the incidence of lung cancer.The Latent-Effect time in male was between 10 and 24 years,while in female it is about 25 to 29 years.A total of 88,286 of the 290,081 subjects who were assessed as high risk of lung cancer participated in lung cancer screening,accounting for 30.43% of the subjects who were assessed as being at high risk for lung cancer.The participation rate among women(39.52%)was higher than that of men(25.22%).Participants aged 40-44 years had the highest participation rate(33.75%),subjects aged 65-69 had the lowest participation rate(26.96%),and the participation rate of body mass index(BMI)was fat Higher(30.98%);higher education level(34.80%)or no smoking history(40.99%),the higher the participation rate was statistically significant.Among lung cancer screening subjects,2977 people were detected with positive nodules(3.37%),and 9635 people were detected with pulmonary nodules,but they were not rated as positive nodules by imaging physicians(10.91%),male The proportion of positive nodules(3.51%)was higher than that of women(3.20%);older(5.33%),leaner BMI(5.06%),lower education level(4.06%),and people with a history of smoking(2.99%)The proportion of positive nodules was high,and the difference was statistically significant.Analysis of the imaging characteristics of all detected pulmonary nodules found that the proportion of positive nodules detected was related to the size,density,quality,and length of nodules.The relationship between the aspect ratio,whether the nodule is calcified,the position of the lung lobe where the nodule is located,whether the nodule edge is clear,whether the pleura is stretched,and the distance between the nodule and the pleura are statistically significant.By evaluating the screening effect,It was found that when "positive nodule" was used as the outcome index of the screening test,the sensitivity was 75.41% and the specificity was 46.17%,and when "suspected lung cancer" was used as the outcome index of the screening test,the sensitivity was 60.66% and the specificity was 72.84%.Using artificial neural network method to build a predictive model of malignant degree of pulmonary nodules can improve the prediction effect.The effect of establishing a prediction model using multiple imaging indicators is the best.The sensitivity was 80.39%,the specificity was 92.00%,and the Jordan index was 0.72,the positive likelihood ratio was 10.05,the coincidence rate was89.64%,the Kappa value was 0.69,and the positive predictive value was 71.93%.Conclusion: Finally,analysis of global tobacco consumption and the incidence of lung cancer revealed that there was a Latent-Effect of tobacco consumption on the incidence of lung cancer.Latent-Effect of tobacco consumption on lung cancer in women are longer than in men.In countries with a high Human Development Index(HDI),the Latent-Effect time is longer,and this Latent-Effect is more pronounced in developed countries and in people with higher tobacco consumption.Analysis of lung cancer screening results as secondary prevention revealed that the characteristics of people with a high participation rate are: females,younger ages,non-standard BMI,higher education,better eating and living habits,and worse living environment quality(air pollution,exposure to kitchen fume,or occupational hazardous substances),physical and mental health abnormalities(psychological trauma or chronic respiratory disease),family history of lung cancer,etc.When using LDCT for lung cancer screening,males,older ages,lower education levels,and higher rates of positive nodules were detected by smokers;analysis of nodular characteristics of different types of positive nodules revealed that nodules in positive nodules Large diameters,long and wide diameters,and higher proportion of pleural traction;using suspicious lung cancer as the evaluation index of screening test outcomes The use of LDCT for lung cancer screening is better than using positive nodules as the screening test outcome indicators,but its accuracy needs to be improved.Compared with the prediction of benign and malignant pulmonary nodules based on the three imaging findings of nodule size,density,and whether it is an endoluminal nodule,the nodule malignancy prediction model constructed with multiple imaging features has higher accuracy and better prediction effect. |
PDF Full Text Request