基于YOLO模型的宫颈细胞簇团智能识别方法

针对宫颈细胞簇团自动识别问题,本文提出了一种基于YOLO v2模型的智能识别方法。首先,针对宫颈细胞簇团识别任务的特点,采用resnet 50模型作为YOLO v2网络的基础特征提取模块。同时,提出了相应的数据扩增方法与YOLO v2网络的训练方案。同时,我们收集宫颈细胞液基涂片扫描图像,建立了宫颈细胞簇团图像数据集,并由细胞病理专家对其中的细胞簇团进行了标注。实验表明,本文方法能够有效完成宫颈细胞病变簇团的自动识别,在测试图像集中,针对细胞簇团识别的准确率为75.9%,召回率为86.3%;针对宫颈细胞图像识别的准确率为87.0%,召回率为86.7%。本文将深度学习技术引入到宫颈细胞辅助筛查领域,对于促进宫颈癌早期自动筛查系统的研究,具有重要意义。

Intelligent recognition method of cervical cell cluster based on YOLO model

Aiming to the automatic recognition of cervical cell cluster, a smart recognition method based on YOLO v2 model was proposed. At first, the model resnet50 was used as basic feature extraction module according to the characters of cervical cell cluster recognition task. Meanwhile, the related data amplification and training program of YOLO v2 network were also proposed. At the same time, we collect the scan image of cervical cell liquid base smear to build the cervical cell cluster image data set and the cell cluster was marked by cytopathic experts. The result shows that the automatic recognition of cervical cell cluster was effectively realized with this method. The accuracy rate of cervical cell cluster was 75.9% and the recall rate was 86.3%. The accuracy rate of cervical cell pathological identification was 87.0% and the recall rate was 86.7%. In this paper, deep learning technique was leaded into the cervical cell auxiliary screening field, it can promote the research of automatic auxiliary screening of early cervical cancer.