Artificial Intelligence-Guided Classification Algorithm For Cervical Cancer Special Staining Images

With the progress of medical technology and the rapid development of artificial intelligence technology,the interdisciplinary combination of technical means for traditional medical technology has had a great impact.Due to the popularity of medical imaging equipment,the number of medical images has been increasing explosively.How to effectively classify and interpret medical images in real time to reduce the labor force has become a widely discussed problem.In numerous cancer,cervical cancer as a gynecological malignant tumour of the high-risk sexual,morbidity and mortality all the year round in the forefront of various malignant diseases,however,the early detection and treatment can greatly reduce the mortality of the disease,so for the work of the early screening of the disease is very important,with the rapid development of artificial intelligence method combined with auxiliary cervical cancer screening is worth studying.Traditional cervical cancer screening smear obtained by cytology test,including pap smear test and liquid-based cytology smear test(TCT),is complex and tedious,which is not conducive to the progress of the screening work.In recent years,based on folate receptor-mediated special cervical staining method,methylene blue color reaction was used to realize rapid and accurate screening of cervical cancer and precancerous lesions.This article in view of the methylene blue cervical special dyeing liquid detection images,pioneering design a complete image acquisition mode,and create the data sets,put forward combining with the computer vision technology for cervical cancer screening study classification algorithm,the supervision of bypass the traditional way of TCT detection,use effect and the supervised learning algorithms improve classification accuracy,to meet the requirements of cervical disease screening work.In this paper,classic algorithms in machine learning,including KNN nearest neighbor algorithm,bayesian algorithm,SVM support vector machine algorithm,integrated learning algorithm,and classic deep learning methods,including AlexNet network,SqueezeNet network,ResNet network,are studied and summarized.This paper has carried on the comprehensive analysis and summary of the algorithms and,by means of experimental verification of each algorithm parameters optimization experiment was carried out,through a large number of experimental verification for each kind of algorithm on cervical cancer screening data set classification effect,and through the integration of multiple indicators to determine the most suitable for cervical cancer screening system of classification algorithm.In the final test data,the accuracy and recall rate can reach more than 90%.