An Unsupervised HIV Lateral Flow Immunochromatography Detection Algorithm Based on K-Means++

As a convenient biochemical detection technology, colloidal gold strip lateral flow immunochromatography has been widely used in disease detection and diagnosis, food safety and other fields. In order to achieve high accuracy and efficient test strip detection, the artificial intelligence-based image classification method was applied to HIV colloidal gold test strip detection in this paper. An unsupervised HIV colloidal gold strip detection algorithm based on K-means++ and machine learning is proposed, and key technologies such as strip image preprocessing, K-means++ image segmentation method, data enhancement and K value selection are studied. Three image classifiers KNN(K nearest neighbor classification algorithm), SVC(support vector machine) and GaussianNB(Gaussian Bayes classifier) were used to compare the classification effect. Experiments show that the classification effect of the proposed algorithm is better than that of the deep learning yolox network. The classification accuracy of the unsupervised detection algorithm based on the combination of K-means++ and KNN can reach 94%, the sensitivity is 98%, and the specificity is 80%, which can well solve the misjudgment problem caused by the insignificant T-line of weak positive test strips.