Zhu et al. developed YOLOv8-BCD, an enhanced real-time deep learning framework designed for pulmonary nodule detection in computed tomography (CT) imaging. Lung cancer screening depends heavily on accurate detection of small nodules, yet traditional approaches are limited by low sensitivity and computational efficiency. YOLOv8-BCD improves upon the YOLOv8 architecture by integrating several key components: the BiFormer attention mechanism for enhanced feature extraction, Content-Aware ReAssembly of Features (CARAFE) for refined upsampling, and Depth-wise Over-Parameterized Depth-wise Convolution (DO-DConv) for stronger local feature representation. Additionally, Super-Resolution GAN (SRGAN) preprocessing was applied to improve CT image clarity and highlight subtle nodules. Using 550 images from the LUNA16 dataset, YOLOv8-BCD achieved an accuracy of 86.4% and a mean average precision (mAP0.5) of 88.3%, exceeding baseline YOLOv8 by 2.2% in accuracy and 4.5% in mAP. External validation on the TianChi dataset further demonstrated robustness, yielding an mAP0.5 of 83.8%, mAP0.5–0.95 of 43.9%, and inference speed of 98 frames per second. Ablation studies confirmed that each integrated module provided distinct advantages, and their combination resulted in the strongest overall performance. Comparative analysis against state-of-the-art models including SSD, Faster R-CNN, DETR, PP-YOLOE, YOLOv9, and YOLOv10 showed YOLOv8-BCD achieved the highest detection accuracy while maintaining real-time inference capability.