BP神经网络在胸癌诊断中的应用研究

Research of Applying BP Neural Network into Breast Cancer Diagnosis

为了提高胸癌识别的识别精度,提出了应用反向传播网络(Back Propagation,BP)建立胸癌诊断。BP网络是一种典型的多层前馈型神经网络,采用有监督学习模式,利用均方误差和梯度下降来实现对网络连接权值的修正。应用BP网络建立的诊断模型是三层网络结构,输入层的9个神经元分别对应描述细胞的特征值;隐层设置4个神经元;输出层的一个单元是对该细胞诊断的类别。数据实验结果显示诊断模型具有较高的识别精度,表明BP网络是一种有效的诊断方法。

In order to improve the diagnosis accuracy, Back Propagation （BP） network was proposed to construct the breast cancer diagnosis model. BP network is a multi layer feed forward neutral network, which taking supervised learning model. The weight of network was adjusted by mean square difference and gradient descent. The diagnosis model is a three layer neural network： the nine neuron of input layer is the feature parameters of cell; four neurons were set in hidden layer and a neuron of output is the class of diagnosed cell. Date experiment shows that diagnosis model has a high distinguish accuracy and the BP algorithm is a valid method.