Efficient and precise cell counting for RNAi screening of Orientia tsutsugamushi infection using deep learning techniques

•Scrub typhus, transmitted by Orientia tsutsugamushi through mite bites, shows nonspecific symptoms. New Orientia species suggest global presence. Prompt antibiotics are vital to prevent severe complications and lower mortality.•Our dataset originates from gene knockdown, immunofluorescence, and high-content screening methods for comprehensive scrub typhus gene analysis, enhancing disease understanding.•Addressing precise cell quantification requires adeptly utilizing efficient deep learning models, specifically object detection variants such as faster R-CNN, You Only Look Once (YOLO), and Adaptive Training Sample Selection (ATSS). These models balance speed and accuracy through streamlined architectures, minimized backbones, data augmentation, and transfer learning.•ATSS, a novel computer vision model, excels in rectifying foreground-background imbalance and is unexplored in biological studies. Utilizing ATSS with shallow backbones highlights high precision and rapid performance, surpassing the popular biological model, Mask R-CNN, in our study.•For biological evaluation, ATSS excels in absolute counting, while faster R-CNN is superior for relative counting. Object detection models with shallow backbones match Mask RCNN performance and outperform image processing in biological assessments. Acquiring fluorescent scrub typhus images obtained through RNA interference screening for the analysis of 60 different human genes and 18 control genes poses challenges due to nonuniform or clumped cells and variations in image quality, rendering image processing (IP) counting inadequate. This study addresses three key questions concerning the application of deep learning methods to this dataset. Firstly, it explores the potential for object detection (OD) models to replace instance segmentation (IS) models in cell counting, striking a balance between accuracy and computational efficiency. Object detection models, including Faster R-CNN, You Only Look Once (YOLO), and Adaptive Training Sample Selection (ATSS) with reduced backbone sizes, outperform the instance segmentation model (Mask Region-Based Convolutional Neural Network: Mask R-CNN, Cascade Mask-RCNN) with both deep and shallow backbones. Notably, ATSS with Resnet-50 achieves an impressive mean average precision of 0.884 in just 33.1 milliseconds. Secondly, reducing the feature extractor size enhances cell counting efficiency, with OD models featuring reduced backbones demonstrating improved performance and faste