人工神经网络在乳腺癌诊断中的应用

Study on Breast Cancer Diagnosis Based on Artificial Neural Network

研究提高乳腺癌诊断准确率问题,对病灶进行医学影像分析会诊结果可靠。因BP神经网络在乳腺癌诊断中,由于权值和学习率固定,导致收敛速度慢、易陷入局极小,使乳腺癌的诊断的准确率低。为了提高乳腺癌诊断的准确率,提出了一种改进BP神经网络的乳腺癌诊断模型。采用附加动量对BP神经网络权值进行调整,加快了网络的收敛速度,然后对学习率进行自适应调整,减少了迭代次数,最后利用改进的BP神经网络模型对乳腺癌进行诊断。仿真结果表明,改进BP神经网减少迭代次数,加快学习速度,提高乳腺癌诊断准确率,很好的解决传统BP神经算法在乳腺癌诊断中中存在的缺陷,是一种有效的乳腺癌的辅助诊断工具。

The traditional BP neural network has slow convergence speed and is easily to fall into local optimum which will affect the accuracy of the diagnosis of breast cancer.In order to enhance the correctness of breast cancer diagnosis,a breast cancer diagnosis model is put forward based on an improved BP neural network.First,this model used additional momentum to adjust the BP neural network weights and speed up the network convergence speed;then the adaptive rate was dynamically adjusted to enhance the network learning speed and reduce the iteration;finally,using the improved BP neural network to diagnosis the breast cancer.Compared with the traditional BP algorithm,experimental results show that the improved BP neural network not only reduces the learning and fasten the learning speed,but also improves the breast cancer diagnosis accuracy.This method can effectively overcome traditional BP algorithm defects.