基于LVQ神经网络的医学图像识别研究

以医学肝脏CT图为研究对象,设计了一种基于LVQ神经网络的医学图像识别方法。基本思想是首先确定图像ROI区域,并作灰度映射变换。接着提取其颜色、纹理和形状特征构成表征医学图像的特征矢量,最后将特征归一化后利用LVQ神经网络进行识别。通过与其他典型神经网络识别方法的实验比较,结果表明,设计的方法能取得更为理想的识别效果。

Medical Image Recognition Based on LVQ Neural Network

Liver CT images are the object for research,and a method for medical image recognition using LVQ neural network is proposed.Firstly,confirm the region of interest and make gray-scale mapping transformation.Secondly,extract color,texture and shape features to consist of feature vector that can represent the medical image accurately.Finally,the features are normalized and sent to LVQ neural network for classification.By comparison with other representative neural network method,the experimental result shows that our methods for liver CT image recognition can achieve a satisfactory effect.