Lung Nodule Detection in Medical Images Based on Improved YOLOv5s

Lung cancer has become one of the malignant tumors with highest morbidity and mortality rate worldwide. The early images lung are pulmonary nodules, detection diagnosis can help reduce incidence cancer. However, due to large differences in shape, size location nodules medical imaging, small is very challenging. This paper proposes a new model, based on improved YOLOv5s, called YOLOv5-CASP. proposed YOLOv5s improvements include: (i) incorporating Convolutional Block Attention Modules (CBAM) suppress interference features through channel dimension spatial dimension, improve accuracy model; (ii) optimizing Spatial Pyramid Pooling - Fast (SPPF) module Atrous (ASPP) as increase model’s receptive field different sizes extract multi-scale contextual information for improving performance nodules; (iii) introducing Contextual Transformer (CoT) optimize part CSPDarknet53 order enhance characteristics model while removing redundant operations extraction capacity. Experimental results conducted two public datasets confirm that YOLOv5-CASP outperforms original other five state-of-the-art models (Faster R-CNN, SSD, YOLOv4-Tiny, DETR-R50, Deformable DETR-R50), terms mean average precision ( mAP ) xmlns:xlink="http://www.w3.org/1999/xlink">F1 score , by achieving corresponding values 72% 0.740 LUNA16 dataset, 79.44% 0.766 X-Nodule dataset.