医学图像纹理分析的方法及应用

Research on methods and application of texture analysis of medical images

医学图像中的纹理特征变化反映了机体的病理改变,因此,图像纹理分析对于疾病的鉴别诊断具有重要意义。目前,由于医学图像及其纹理的复杂性,尚未有一种通用的、适合各类医学图像的纹理分析方法。医学图像纹理分析中有4种常用的纹理特征提取方法,其每种纹理特征提取方法均存在优点和缺点。对不同医学影像(CT、MRI、B超及数字X线片)上肝部及肺部疾病的图像纹理分析表明,基于肝脏病变CT图像的计算机辅助诊断系统提取的纹理特征对图像内容有较好的区分性;通过纹理分析可对早期肺癌进行检测。

The changes on texture features of medical image often reflect the pathological changes in patients. Therefore, texture analysis of medical images is significant for the diseases diagnosis. Because of the complexity of the medical image and its texture characteristics, there wasn’t an universal and appropriate method for all kinds of medical images at present. There are four popular methods for feature extraction, each of them has different advantages and disadvantages. According to the texture analysis on the images of liver and lung diseases in different images （CT, MRI, ultrasound and digital radiograph）, the texture feature based on CT of liver disease which extracted by computer aided diagnosis system can distinguish images better. Texture analysis can be a measure for early lung cancer detection.