摘要
宫颈癌是全球第二高发的女性癌症,但是如果及时发现,其治愈率几乎为100%。阴道镜检查是临床筛查宫颈上皮内瘤变(CIN)和早期宫颈癌的重要步骤之一,直接影响患者的诊断方案。然而,这种方法取决于阴道镜检查者的观察。本文建立了宫颈图像的数据集,并提出了一种基于Efficientnet的宫颈图像分类的方法。实验结果表明,该模型取得了比经典深度学习方法更好的分类性能,其分类结果准确率可达90.56%。
Cervical cancer is the second most common female cancer in the world,but if detected in time,the cure rate is almost 100%.Colposcopy is one of the important steps for clinical screening of cervical intraepithelial neoplasia(CIN)and early cervical cancer,which directly affects the patients diagnosis plan.However,this method depends on the observation of the colposcopy examiner.In this paper,a dataset of cervical images is established and a method of cervical image classification based on Efficientnet is proposed.Experimental results show that the model achieves better classification performance than classic deep learning methods.The accuracy of the classification results can reach 90.56%.
作者
巫毅
吴钢华
乔政杰
蒋天豪
WU Yi;WU Ganghua;QIAO Zhengjie;JIANG Tianhao
出处
《计量与测试技术》
2021年第12期62-65,共4页
Metrology & Measurement Technique
