基于深度学习的甲状腺淋巴结转移癌病理诊断方法

Pathological Diagnostic Method of Thyroid Lymph Node Metastases Based on Deep Learning

为了自动化识别甲状腺淋巴结转移癌,提出了多任务中心的深度学习诊断方法。以卷积神经网络算法为基础构造了镜下结构观察任务中心和镜下细胞观察任务中心,以模拟完整的镜下诊断过程,而且各中心的观察因素呈多样性;诊断任务中心采用的是多因素下的直觉模糊集诊断方式,从而综合给出诊断结果。该方法完整地模拟了病理医生的显微镜下分析过程,实验结果表明了该方法的有效性,甲状腺淋巴结转移癌的识别率结果令人满意。

In order to automatically identify thyroid lymph node metastases,a deep leaning pathology diagnosis of multitask centers was proposed.The observation task center of structure under the microscope and the observation task center of cell under the microscope were constructed based on the convolution neural network algorithm,which can be used to simulate the complete diagnosis process,furthermore the observation factors of task center are various.The intuitionistic fuzzy set was used to give the comprehensive diagnosis results.The method simulates the whole diagnostic process of doctors,the experimental result shows that the method is effective and the recognition rate of thyroid lymph node metastases is satisfactory.