A deep learning model for detection of leukocytes under various interference factors

Abstract The accurate detection of leukocytes is the basis for the diagnosis of blood system diseases. However, current methods and instruments either fail to fully automate the identification process or have low performance. To improve the current status, we do need to develop more intelligent methods. In this paper, we investigate fulfilling high-performance automatic detection for leukocytes using a deep learning-based method. A complete working pipeline for building a leukocyte detector is presented, which includes data collection, model training, inference, and evaluation. We established a new leukocyte dataset that contains 6273 images (8595 leukocytes), considering nine common clinical interference factors. Based on the dataset, the performance evaluation of six mainstream detection models is carried out, and a more robust ensemble scheme is proposed. The mAP @IoU = 0.50:0.95 and mAR@IoU = 0.50:0.95 of the ensemble scheme on the test set are 0.853 and 0.922, respectively. The detection performance of poor-quality images is robust. For the first time, it is found that the ensemble scheme yields an accuracy of 98.84% for detecting incomplete leukocytes. In addition, we also compared the test results of different models and found multiple identical false detections of the models, then provided correct suggestions for the clinic.