基于一种改进的LBP算法和超限学习机的肝硬化识别
发布时间：2018-11-23 18:33
【摘要】：肝硬化的计算机辅助诊断对肝脏疾病的早期治疗和诊断具有重要意义。针对B超图像中肝硬化病变区域边缘模糊和回声不均匀、尺度因素影响等问题,提出了改进的LBP算法并提取了相应的SLBP特征。该特征较传统的纹理特征更准确地描述了B超图像中肝硬化病变的特征,结合二维Gabor变换,解决了上述难题。鉴于传统的机器学习方法的训练时间较长,采用基于超限学习机的训练方法,并首次将其应用于肝硬化识别。实验结果表明,所提方法对测试集的分类准确率达到95.4%,在时间效率上较传统方法有很大提高。ROC曲线表明,提出的分类方法在准确率和泛化能力上均优于传统方法,有助于肝硬化的临床诊断。
[Abstract]:Computer aided diagnosis of liver cirrhosis is of great significance for early treatment and diagnosis of liver diseases. An improved LBP algorithm is proposed and the corresponding SLBP features are extracted in order to solve the problems of edge blur and non-uniformity of echo and the influence of scale factors in B-mode ultrasound images. This feature is more accurate than the traditional texture features in describing the liver cirrhosis in B-mode ultrasound images. The above problems are solved by combining two-dimensional Gabor transform. In view of the long training time of the traditional machine learning method, the training method based on the transcendental learning machine is adopted, and it is applied to the liver cirrhosis recognition for the first time. The experimental results show that the classification accuracy of the proposed method is 95.4, and the time efficiency of the proposed method is much higher than that of the traditional method. The ROC curve shows that the proposed classification method is superior to the traditional method in accuracy and generalization ability. It is helpful for the clinical diagnosis of liver cirrhosis.
【作者单位】： 青岛大学计算机科学技术学院;山东省数字医学与计算机辅助手术重点实验室;加州大学洛杉矶分校;
【基金】：国家自然科学基金项目:计算机辅助肝纤维化无创诊断(61303079);国家自然科学基金项目:空变运动模糊图像的盲复原变分模型及其快速算法(61305045)资助
【分类号】：R575.2;TP391.41
本文编号：2352378
[Abstract]:Computer aided diagnosis of liver cirrhosis is of great significance for early treatment and diagnosis of liver diseases. An improved LBP algorithm is proposed and the corresponding SLBP features are extracted in order to solve the problems of edge blur and non-uniformity of echo and the influence of scale factors in B-mode ultrasound images. This feature is more accurate than the traditional texture features in describing the liver cirrhosis in B-mode ultrasound images. The above problems are solved by combining two-dimensional Gabor transform. In view of the long training time of the traditional machine learning method, the training method based on the transcendental learning machine is adopted, and it is applied to the liver cirrhosis recognition for the first time. The experimental results show that the classification accuracy of the proposed method is 95.4, and the time efficiency of the proposed method is much higher than that of the traditional method. The ROC curve shows that the proposed classification method is superior to the traditional method in accuracy and generalization ability. It is helpful for the clinical diagnosis of liver cirrhosis.
【作者单位】： 青岛大学计算机科学技术学院;山东省数字医学与计算机辅助手术重点实验室;加州大学洛杉矶分校;
【基金】：国家自然科学基金项目:计算机辅助肝纤维化无创诊断(61303079);国家自然科学基金项目:空变运动模糊图像的盲复原变分模型及其快速算法(61305045)资助
【分类号】：R575.2;TP391.41
【相似文献】
相关期刊论文 前1条
1 刘金林;房颖;王涛;;肝硬化自发性细菌性腹膜炎患者血清LBP水平变化及意义[J];山东医药;2013年28期
，本文编号：2352378
本文链接：https://www.wllwen.com/yixuelunwen/xiaohjib/2352378.html
最近更新
教材专著
