新疆维吾尔自治区高发病哈萨克族食管癌计算机辅助诊断方法的研究

Esophageal cancer ,which has the high morbidity and mortality, accounted for 2% among all malignant tumors.It is a common malignant tumor attacked the esophageal mucosa epithelium and esophageal gland epithelium. Kazakh nation in Xinjiang Uygur Autonomous Region is the nation with the highest incidence of esophageal cancer in the world. Census and screening of the high incidence of high-risk groups (precancerous lesions and early cancer) is one of the key issues directly related to the ability of reducing the morbidity and mortality.Based on the medical image features of kazak nation's esophageal disease, this project will plan to focused research the methods and technologies of computer-aided diagnosis for kazak nation's esophageal cancer, based on our previous researches, with the clinicial support of the six affiliated hospitals of Xinjiang medical university and other hospitals in autonomous, combining wih the experts' experience, using computer vision, image processing, visual perception principles and other related technologies, combining content-based medical image retrieval technology with the computer aided diagnosis technique. Providing the decision-making for clinicians and the basis for the earily detection, definite diagnosis and the treatment of esophageal cancer,improving the credibility of diagnosis and the flexibility and performance of the diagnostic procedures. Furthermore, it is benefit to the Autonomous Region's economic construction and the development of medical and health services.

食管癌是食管粘膜上皮及食管腺上皮发生的常见恶性肿瘤，占所有恶性肿瘤的2%，其发病率和死亡率极高。新疆维吾尔自治区哈萨克族是世界上食管癌发病率最高的民族。高发区高危人群(癌前病变和早期癌患者)的普查和筛选，是直接关系到能否降低食管癌发病率和死亡率的关键问题之一。本项目根据哈萨克族食管癌疾病的医学图像特点，利用课题组已有的研究基础，借助我校六所附属医院及自治区各大医院所提供的数据、临床方面的支持，结合食管癌权威专家的经验，采用计算机视觉、图像处理和视觉感知原理等相关技术，将基于内容的医学图像检索技术与计算机辅助诊断技术融为一体，重点研究适于新疆维吾尔自治区高发病哈萨克族食管癌计算机辅助诊断的技术、方法及算法，为临床医师提供决策上的参考，提高食管癌诊断的可信度，为其早期发现、确诊、治疗提供依据，从而提高族食管癌诊断操作的灵活性和整体性能，这对自治区经济建设与卫生医疗事业的发展具有非常重要的意义。

本项目针对哈萨克族食管癌X射线医学图像，在国内外计算机辅助诊断技术研究成果的基础上，充分利用课题组成员在图像处理、模式识别、软件开发及数据统计等方面积累的经验和技术，并结合我校各附属医院及自治区各大医院所提供的数据、临床方面的支持，根据哈萨克族食管癌疾病的医学图像特点，采用计算机视觉、图像处理和视觉感知原理等相关技术，对哈萨克族食管癌X射线医学图像的预处理、分割、特征提取和选择、分类识别等关键技术做了大量的实验研究，提出了适用于哈萨克族食管癌X射线医学图像的预处理方法和分割方法，提取了灰度、纹理、形状等底层视觉特征，建立了哈萨克族食管癌图像数据库及图像特征数据库，对比了哈萨克族食管癌X射线医学图像的分类识别算法，最终开发了基于C/S模式的哈萨克族食管癌医学图像计算机辅助诊断系统。本项目的研究有助于为临床医师提供理解和决策上的参考，提高哈萨克族食管癌诊断的可信度和准确率，从而提高新疆哈萨克族食管癌诊断操作的灵活性和整体性能，为食管癌的早期发现、确诊、治疗提供依据，这对自治区经济建设与卫生医疗事业的发展具有非常重要的意义。