【作者】 王喆；

【导师】 路文焕；

【作者基本信息】 天津大学 ， 软件工程， 2018， 硕士

【摘要】 磁共振成像(MRI)是最常见的乳腺癌诊断后的评估手段,用以评估乳腺癌疾病发展的风险。乳腺MRI可以用来确定乳腺癌的分期。乳腺MRI主要用于观测乳腺癌诊断后,癌组织的大小和范围,从而帮助医师确定癌症的分期,以及化疗后癌组织的变化情况。本课题的研究目的是构建一个基于卷积神经网络的,能够自动将乳腺同一位置的四种模式的核磁共振成像图像融合起来,对乳腺癌病灶进行分类分割检查的模型。主要研究内容是使用卷积神经网络方法,对多参数磁共振成像的乳腺图像中肿瘤组织部分的分类与分割。本研究中的乳房的MRI数据是使用1.5T-MRI扫描仪,对67名受试者分别采用四种不同成像模式扫描而获得的。主要采用的四种不同的成像模式分别为:T1加权成像,T2加权成像,弥散加权的eTHRIVE序列成像,和动态对比度增强参数成像。研究模型主要由图像分类和图像分割两部分组成。其中,研究提出的用于乳腺癌分类诊断的四图融合骨干网络,克服了单模态图像检测的局限性,并模拟了临床医生及放射科医师的实际诊断过程,达到0.942的分类准确率。其次,研究提出的自动分割肿瘤组织的过程采用了优化的U-Net模型,使得图像分割结果得到了显著性的提升。四种模式融合的分类骨干网络和用于癌组织分割的改进的U-Net模型相组合,构建了基于神经网络的乳腺癌诊断模型,形成了能够应用于实际乳腺癌临床诊断工作的计算机辅助检测系统(CAD)。更多还原

【Abstract】 Magnetic resonance imaging(MRI)is the most common post-diagnostic assessment of breast cancer.It is used to assess the risk of breast cancer disease progression.Mammary MRI can be used to determine the stage of breast cancer.Mammary MRI is mainly used to observe the size and extent of cancer tissues after breast cancer diagnosis,and help physicians to determine the stage of breast cancer and the changes of cancer tissues after the chemotherapy.The purpose of this research is to construct a model based on convolutional neural network that can automatically merge the four modes of MRI images at the same location of the breast to classify and segment the lesions of breast cancer.Attempts were made for tumor classi cation and segmentation;using a multi-parametric Magnetic Resonance Imaging(MRI)method on breast tumors.MRI data of the breast were obtained from 67 subjects with a 1.5T-MRI scanner.Four imaging modes: were T1 weighted,T2 weighted,Di usion Weighted and eTHRIVE sequences,and dynamiccontrast-enhanced(DCE)-MRI parameters are acquired.The proposed four-mode linkage backbone in tumor classi cation,which overcomes the limitations of single-modality image detection and simulates actual diagnosis processes by clinicians,achieves the accuracy of 0.942.The proposed automatic segmentation approach is performed by a re ned U-Net architecture,and the result improved segmentation performance signi cantly.The combination of four-mode linkage classi cation backbone and improved segmentation network for breast cancer detection forms a detection model based on convolutional neural network,and give a computer-aided detection(CAD)system that corresponds to the actual clinical diagnosis work.更多还原