Digital PCR polymerase chain reaction technology is one of the main methods used in nucleic acid detection. With the development of microfluidic technology, droplet technology has the advantages of small size, high precision, and complete isolation between reaction chambers in nucleic acid detection. widely used in. The ddPCR experiment consists of four steps, namely, the preparation of samples and droplets to be detected, the amplification of target nucleic acids, the collection of fluorescent signals, and the analysis and processing of experimental data. Among them, the accuracy of fluorescence signal acquisition will directly affect the results of ddPCR experiments.
At present, the methods of fluorescence signal acquisition include microscopic image detection method and flow cytometry-based fluorescence detection method. Compared with the latter, the microscopic image detection method has less droplet breakage, higher precision and faster detection speed. . The characteristics of droplet fluorescence microscopy images of ddPCR experiments are composed of a large number of circular bright spots with similar sizes and different fluorescence intensities. The key to affecting the experimental results lies in the identification, classification and counting of droplets in the image, that is, calculating the total number of droplets. And read the fluorescent signal of each droplet to judge whether it is negative or positive. Image speckle detection mainly includes image denoising, enhancement of target speckle features, and speckle counting. Conventional image blob counting usually uses the image grayscale threshold segmentation method to binarize the image, and then calculate the number of areas corresponding to the target blobs. However, this single-threshold image binarization method is only applicable to the image background is uniform and the target point is single. In the detection of multiple targets, the error of this method will increase with the increase of the detection target types. Due to the large number of droplets, small size, tight arrangement and uneven fluorescence intensity in the experiment, the droplets will lead to Hard to count.
