Diagnosis of Pneumonia from Chest X-ray Images Using YOLO Deep Learning

Yanchun Xie^1*Binbin Zhu^2*Bin Zhao³Yang Jiang³

• ¹Department of Orthopaedics, General Hospital of Northern Theater Command, Shenyang, China
• ²The Fourth People's Hospital of Shenyang, Shenyang, Liaoning Province, China
• ³Northeastern University, Shenyang, Liaoning Province, China

Early and accurate diagnosis of pneumonia is crucial to improve cure rates and reduce mortality.Traditional chest X-ray analysis relies on physician experience, which can lead to subjectivity and misdiagnosis. To address this, we propose a novel pneumonia diagnosis method using the Fast-YOLO deep learning network that we introduced. First, we constructed a pneumonia dataset containing five categories and applied image enhancement techniques to increase data diversity and improve the model's generalization ability. Next, the YOLOv11 network structure was redesigned to accommodate the complex features of pneumonia X-ray images. By integrating the C3k2 module, DCNv2, and DynamicConv, the Fast-YOLO network effectively enhanced feature representation and reduced computational complexity (FPS increased from 53 to 120). Experimental results subsequently show that our method outperforms other commonly used detection models in terms of accuracy, recall, and mAP, offering better real-time detection capability and clinical application potential.X-ray images, providing clinicians with an efficient and accurate auxiliary tool. This helps accelerate the diagnosis process and improve diagnostic accuracy. Related program will be open source in the future: https://github.com/Zhaobin7/Fast-Yolo. 2 Pneumonia Dataset and Evaluation Metrics 2.1 Pneumonia Dataset MIMIC-CXR (MIMIC Chest X-ray) is an open-source chest X-ray dataset, and the dataset is designed to provide data support for medical image analysis, disease prediction, and the development of automated diagnostic systems. It contains a large number of accurately annotated chest X-ray images along with corresponding pathology reports, covering a variety of pulmonary diseases, with pneumonia being one of the key categories. The MIMIC-CXR dataset includes over 200,000 chest Xray images, representing a diverse range of chest diseases, such as pneumonia, tuberculosis, and pneumothorax, with the majority of the images sourced from hospitalized patients. All images are stored in DICOM (Digital Imaging and Communications in Medicine) format, ensuring highresolution and clear image quality [13][14][15] . However, despite the detailed disease labels provided in the pathology reports, the reports are written by different radiologists, which may introduce subjectivity and label inconsistency. This variance can potentially affect the training performance of disease