IEEE Access · 2023 · 60 citations · 31 references
ComputingComputer ScienceOncologyBiomedical ImagingComputer VisionDiagnostic ImagingDeep LearningImage Feature EngineeringImage OptimizationMedical Image AnalysisMedical ImagingArtificial Intelligence-based Nodule DetectionConvolutional Neural Network DetectionImage-based Tumor DetectionMedical Imaging OptimizationReceptive Field EncodingReceptive Field SizeRegion Of Interest DetectionSpatial Dimension Reduction
Lung cancer has become one of the malignant tumors with the highest morbidity and mortality rate worldwide. The early images of lung cancer are pulmonary nodules, and early detection and diagnosis can help reduce the incidence of lung cancer. However, due to the large differences in shape, size and location of pulmonary nodules in medical imaging, the detection of small nodules is very challenging. This paper proposes a new model, based on improved YOLOv5s, called YOLOv5-CASP. The proposed YOLOv5s improvements include: (i) incorporating Convolutional Block Attention Modules (CBAM) to suppress the interference features of the medical images through a channel dimension and spatial dimension, and to improve the detection accuracy of the model; (ii) optimizing the Spatial Pyramid Pooling - Fast (SPPF) module of YOLOv5s to Atrous Spatial Pyramid Pooling (ASPP) as to increase the model’s receptive field of images of different sizes and extract multi-scale contextual information for improving the detection performance on small lung nodules; and (iii) introducing a Contextual Transformer (CoT) module to optimize part of the CSPDarknet53 module of YOLOv5s in order to enhance the characteristics of the model while removing redundant operations extraction capacity. Experimental results conducted on two public datasets confirm that the proposed YOLOv5-CASP model outperforms the original YOLOv5s model and other five state-of-the-art models (Faster R-CNN, SSD, YOLOv4-Tiny, DETR-R50, Deformable DETR-R50), in terms of the mean average precision ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">mAP</i> ) and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">F1 score</i> , by achieving corresponding values of 72% and 0.740 on the LUNA16 dataset, and 79.44% and 0.766 on the X-Nodule dataset.
31
Deep Residual Learning for Image Recognition
Kaiming He, Xiangyu Zhang, Shaoqing Ren et al. · 2016 · 222.1K citations · Full text
DROPS (Schloss Dagstuhl – Leibniz Center for Informatics) · 2023 · 76.3K citations · Full text