基于GIS与遥感的中国肺癌生态环境响应因子研究

发布时间：2018-04-08 17:11

本文选题：肺癌　切入点：神经网络　出处：《中国科学院大学(中国科学院遥感与数字地球研究所)》2017年硕士论文

【摘要】：近年来,随着经济的快速发展,人民生活水平的不断提高,饮食习惯、生活节奏和生活方式产生了极大的改变,同时快速工业化导致的环境污染问题也日益严重,突出表现为水、土壤、大气等环境要素的恶化,这些都对人类生存状态产生不利影响,直接表现为人群各种疾病的响应。自2004年,全国系统开展大范围的疾病普查,研究数据表明,我国人群疾病谱发生了重要改变,突出表现为以肺癌为代表的恶性肿瘤为已经成为主要的疾病形式。肺癌严重威胁到人类健康,由此造成的疾病负担更是受到各国政府的重视。肺癌高发的趋势使人们意识到解决癌症问题不能仅仅寄托在治疗技术的突破上,更应该在癌症预防方面加大力度。除了社会因素影响之外,肿瘤的发生已被证明与某些地理环境和环境污染有着密切的关系,因此筛选生态环境中与肺癌有关的响应因子对于肺癌的防治将起到至关重要的作用。GIS具有强大空间分析和数据管理功能,卫星遥感技术能够以相对低廉的成本,迅速的获取大范围、长时间序列的生态环境数据。本文尝试借助于遥感与GIS技术,综合运用多种分析方法,从生态环境的角度初步探索对肺癌发病率具有显著影响的因子,以期为肺癌的生态防控有意的借鉴。本文基于卫星遥感影像获取研究区的植被指数与大气污染物SO2、NO2、AOD,并采用神经网络模型反演出PM2.5数据。基于地面气象站点获取气候数据,包括相对湿度、降水、气温、气压,并采用克里金插值模型计算出研究区的各气候因子均值。从《中国肿瘤登记年报》中提取出各肿瘤监测点的肺癌统计数据,并从空间和时间的角度分析了我国肺癌发病的时空特征,基于最小二乘法分析2006与2009年间肺癌发病率与各影响因子变化值的关系。分别基于普通线性回归和空间回归模型的结果,选取适宜的回归模型,分析了2009年肺癌发病率与各生态环境影响因子的关系,筛选出具有显著相关性的因子。最后采用灰色关联度模型,分别分析了2006年与2009年各影响因子对肺癌发病率影响的程度。得出如下结论:(1)我国肺癌发病率男性高于女性,并呈逐年上升趋势,其构成比也逐年上升,肺癌危害程度在增大。从空间分布上看,肺癌发病率表现出较大的空间差异性。高发区主要分布在东北、北京、武汉、长三角以及成渝地区,低发区主要分布在中西部省份。男女肺癌发病率在空间分布上也存在一定的差异,男性高发区主要位于东北及长三角地区,女性高发区主要分布在东北地区。在肺癌发病趋势上,出现负增长的登记点主要分布在江苏、湖南和广东境内,增长率为正的登记点主要分布在东北、北京、长三角和中原地区,其中中部地区增长最为显著。在时间维度上,2006-2009年间男女肺癌发病率随着NO2、SO2、PM2.5变化的增大而增大,随着NDVI的降低而增大,其中肺癌发病率与SO2的变化相关性最大,其次是NDVI、PM2.5和NO2。(2)分析结果表明,空间误差模型适宜对各因子与肺癌发病率进行回归分析。回归结果显示肺癌发病率随着NO2(β=0.8,p0.05)、SO2(β=0.59,p0.01)、PM2.5(β=0.13,p0.01)、相对湿度(β=7.37,p0.05)的增大而增大,随着降水量(β=-0.62,p0.05)、NDVI(β=-0.18,p0.01)的增大而降低。气温与气压的回归分析显示暂不能认为两者与肺癌发病率的相关性具有统计学意义(p0.05)。(3)灰色关联度模型分析结果表明,各因子(NO2、SO2、PM2.5、相对湿度、降水量、NDVI)虽然在不同年份对肺癌发病率的影响程度有所差异,总体上看,植被指数NDVI对我国肿瘤登记地区的肺癌发病率的影响最大,相对湿度最低。大气污染物中,对男女肺癌发病率影响最大的因素是SO2,其次是NO2,PM2.5最小。降水量对男女肺癌发病率的影响相对较小。
[Abstract]:In recent years, with the rapid development of economy, the improvement of people's life, eating habits, rhythm of life and way of life has changed greatly, the problem of environmental pollution and rapid industrialization led to increasingly severe, especially for water, soil, air and other environmental deterioration factors, which have adverse effects on human the survival of the state, directly shows the responses of human diseases. Since 2004, to carry out a wide range of diseases among the national system, research data show that there was an important change in population spectrum of disease in our country, this is manifested by lung cancer as the representative of the malignant tumor has become the main form of the disease. Lung cancer is a serious threat to human health, thus the burden of disease caused more attention by the governments. The trend of high incidence of lung cancer so as to make people aware of not only solve the problem of cancer treatment lies in the technical breakthrough, Should be more in cancer prevention efforts. In addition to the influence of social factors, the occurrence of tumor has been proved to have close relationship with the geographical environment and environmental pollution, prevention and treatment of cancer related response factors so screening for lung cancer in the ecological environment will play a important role to.GIS has strong spatial analysis and data the management function of satellite remote sensing technology can be at a relatively low cost, rapid access to a wide range of ecological environment data, long time series. This paper attempts by remote sensing and GIS technology, comprehensive use of various analysis methods, from the perspective of the ecological environment factors of preliminary exploration has significant effect on the incidence of lung cancer, in order to ecological control for lung cancer to learn. SO2, vegetation index and atmospheric pollutants in the study area of satellite remote sensing image based on NO2, AOD, and the neural network model The performance of PM2.5 data access. Based on Meteorological climate data, including precipitation, relative humidity, air temperature, air pressure, and calculate the mean climate factors in the study area using Kriging interpolation model. From the Chinese < tumor registry extract lung cancer statistics the tumor monitoring points of the annual report ", and from the Perspective of time and space analysis of the spatial and temporal characteristics of Chinese lung cancer, analysis of the relationship between 2006 and 2009 years, the incidence of lung cancer and the influence factor of variation based on least square method. Based on the general linear regression and spatial regression model results respectively, selected proper regression model should be analyzed in 2009, the incidence of lung cancer and the ecological factors relationship. Selected factors with significant correlation. Finally, by using the grey correlation model, respectively analyzes the extent of 2006 and the influence factors on the incidence of lung cancer in 2009. As a Conclusion: (1) in China, the incidence of lung cancer in men than women, and increasing year by year, the proportion is increasing, the degree of harm in lung cancer increased. From the view of spatial distribution, the incidence of lung cancer showed a larger difference of space. High incidence area mainly distributed in the northeast, Beijing, Wuhan, and the Yangtze River Delta Chengdu Chongqing region, the low incidence area mainly distributed in the central and western provinces. The incidence of lung cancer and there are some differences in the spatial distribution, the high incidence area of men mainly located in the northeast and the Yangtze River Delta region, the high incidence area of women is mainly distributed in the northeast region. The incidence trend of lung cancer, negative growth in the registration points are mainly distributed in Jiangsu, Hunan and in Guangdong, growth rate is the registration point is mainly distributed in the northeast, Beijing, Yangtze River Delta and the Central Plains region, the central region of the most significant growth. In the dimension of time, 2006-2009 years among men and women with the incidence of lung cancer NO2, SO2, PM2.5 increased change increases, and increases with the decrease of NDVI, the change of correlation between the incidence of lung cancer and SO2 was the biggest, followed by NDVI, PM2.5 and NO2. (2) analysis results show that the spatial error model suitable for each factor and the incidence rate of lung cancer by regression analysis. The regression results show the incidence of lung cancer with the rate of NO2 (beta =0.8, P0.05), SO2 (beta =0.59, P0.01), PM2.5 (beta =0.13, P0.01), relative humidity (beta =7.37, P0.05) increases with the amount of precipitation (beta =-0.62, P0.05), NDVI (beta =-0.18, P0.01) increased and decreased temperature regression. With the pressure of the analysis showed that can not be considered statistically significant correlation with the incidence of lung cancer (P0.05). (3) the grey correlation model analysis showed that the factors (NO2, SO2, PM2.5, relative humidity, precipitation, NDVI) although in different years have different degree of influence on the incidence of lung cancer. Look on the whole, The vegetation index NDVI has the largest impact on the incidence of lung cancer in China's cancer registries, and the lowest relative humidity. The most important factor in the incidence of lung cancer is SO2, followed by NO2 and PM2.5. The impact of precipitation on the incidence of lung cancer in men and women is relatively small.

【学位授予单位】：中国科学院大学(中国科学院遥感与数字地球研究所)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P208;R734.2;TP79

【参考文献】