基于医学影像肝癌检测方法的研究资料.docVIP

华北水利水电学院硕士学位论文 基于医学影像的肝癌检测方法研究 摘 要 近年来，随着科技的发展，医疗技术也在快速的发展。数字化医学图像采集设备不 断更新，CT、MR 等设备的普及，使得医院每天产生大量的医学影像数据，医生的工作量 越来越大，在紧张与高强度的工作压力下容易出现误诊和漏诊的现象。采集得到的影像 数据存在一定的问题，大量的边缘模糊现象，这给病理诊断和治疗带来了困难。计算机 在硬件方面和软件方面上的快速发展和完善给医学辅助诊断带来了新的契机。 本文对基于 CT 图像的肝癌检测技术进行了研究，根据放射科医生临床诊断经验及 肝癌 CT 图像的特点，提出了一个从图像预处理、图像分割、特征提取、到分类识别的 肝癌检测方法，实现了对肝癌的有效识别。通过图像分割、轮廓提取和轮廓校正，得到 了较为精确的病灶区边缘；提取了图像的纹理特征和形状特征；采取 BP 神经网络方法 设计了两个分类器进行肝癌识别，首先依据纹理特征区分正常肝组织和非正常肝组织， 其次利用形状特征区分了非正常肝组织中的囊肿、血管瘤和肝癌。实验结果表明，正常 肝的识别率为 90.00%，非正常肝的识别率为 85.00%，区分囊肿、血管瘤和肝癌的识别 率分别为 72.73%、50.00%和 83.33%，识别效果良好。 通过计算机肝癌辅助诊断技术的使用，减轻了医生的工作压力和劳动强度，提高了 医生的工作效率；在一定程度上提高了诊断的准确度；为医院节省了大量的开支，减轻 了病人的负担。因此，计算机辅助诊断研究具有重要的意义和应用价值。 关键词：图像处理；肝癌；检测方法；特征提取 I 华北水利水电学院硕士学位论文 THE RESEARCH OF LIVER CANCER DETECTION METHODS BASED ON MEDICAL IMAGE ABSTRACT In recent years, medical technology develops rapidly with the development of science technology. The coming out of digital medical image acquisition one after another and the university of CT, MR and other equipment make the large amount of image data per day and the workload of doctors. The image data collected always have some problems, such as a large number of blur edges and there are no visible lesions in ninety-five percent of the images. The problems presented above bring difficulty to pathological diagnosis and treatment. The rapid development and improvement in computer hardware and software have brought an opportunity for medical diagnosis. The combination of computer technology with medical image processing technology in diagnostic improves doctors’ efficiency and also improves the capabilities of lesion identification and diagnostic. The research in this paper is of great significance and value. The research of liver cancer detection methods based on clinical experience of radiologist and features of liver CT images. A method of liver cancer detection from the image preprocessing, image segmentation, feature extraction, to the classification has been proposed, which can identify liver cancer ef