Classification of Cervical Cancer Pathological Images Based on
Texture Features and Conditional Random Field
ZHANG Jiawei1，LI Chen1，HE Liangzi2，Chen Hao1
1. College of Medicine and Biological Information Engineering, Northeastern University, Shenyang Liaoning 110819, China丨
2. School of Biological Science and Medical Engineering, Beihang University, Beijing 100191, China
Abstract：Objective To solve the problem of low accuracy of traditional artificial diagnosis caused by subjective factors, conditional
random field was used to increase the classification accuracy of cervical cancer pathological images. Methods First, the image was
preprocessed by grayscale and meshing丨 secondly, the feature of GLCM was extracted to characterize the histopathological image
blocks of cervical cancert丨 then, the effective feature vector combination was selected as the texture feature for the one-potential and
the binary potential of the conditional random field丨 finally, using the resulting potential difference, the final image-level classification
results were predicted by the design conditional random field model. Results The accuracy rate was 82% on the samples of cervical
cancer histopathological images stained by immunohistochemical. Conclusion The result shows a good classification accuracy,
which shows computer-aided diagnosis can be widely used in the diagnosis of histopathological images in the future, and this method
can reduce the burden of doctors, help doctors to improve the work efficiency, and the accuracy of judgment.
张家伟1，李晨1，贺良子2，陈昊1. 基于纹理特征与条件随机场的宫颈癌病理图像分类研究[J]. 中国医疗设备, 2021, 36(8): 45-50.
ZHANG Jiawei1，LI Chen1，HE Liangzi2，Chen Hao1. Classification of Cervical Cancer Pathological Images Based on
Texture Features and Conditional Random Field. China Medical Devices, 2021, 36(8): 45-50.
