Performance Analysis of YOLOv11-m and Related Architectures in Pediatric X-Ray Fracture Detection
Artificial Intelligence and Applications · 10.04.2026 · 4 (4) · 549-560
DOI: 10.47852/bonviewaia62027255 ↗Kurz gefasst
Fachartikel in der Zeitschrift Artificial Intelligence and Applications, online erschienen am 10.04.2026. Verfasst von Muhanad Abdul Elah Alkhalisy und 3 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Wrist fractures are one of the most frequent fractures in children that should be diagnosed properly and immediately to avoid any further complications. Regular radiographic analysis is time-consuming, relies on qualified radiologists, and is subject to human error. There has been a lot of buzz about the potential of deep learning technologies to automate medical image analysis in recent times. The algorithms in the YOLO series are sophisticated and refined, representing some of the best work in the field. The GRAZPEDWRI-DX pediatric wrist dataset is used in this study to address the variation in the YOLOv11 detection model’s inference time, recall, precision, and mean average precision (mAP). A medium-scaled variant of YOLOv5, YOLOv7, YOLOv8, YOLOv9, and YOLOv10 models were compared with YOLOv11-m. According to the experimental results, YOLOv11-m achieves the highest mAP@50–95 (0.569) for fracture detection, whereas YOLOv11-x attains the highest mAP@50–95 (0.424) across all category classes. An ablation study of YOLOv11 architectural elements was done to determine how the model optimized for medical imaging through the attention mechanism. On the other hand, the performance in fracture detection and inference time for YOLOv11 is superior to that of previous YOLO models, while it has lower computational complexity. Additionally, a performance comparison among recent YOLO-based models and the RT-DETR and Faster R-CNN models reveals variations in detection precision, recall, and accuracy, providing insight into the strengths and trade-offs of each approach. YOLOv11 could therefore make recurrent contributions and be of great value to clinical decision-making. Received: 18 August 2025 | Revised: 2 February 2026 | Accepted: 25 March 2026 Conflicts of Interest The authors declare that they have no conflicts of interest to this work. Data Availability Statement The data that support the findings of this study are openly available at https://doi.org/10.1038/s41597-022-01328-z, reference number [15]. Author Contribution Statement Muhanad Abdul Elah Alkhalisy: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Resources, Data curation, Writing – original draft, Visualization. Qusay Shihab Hamad: Validation, Investigation, Writing – review & editing, Visualization, Supervision, Project administration. Ali Retha Hasoon Khayeat: Validation, Investigation, Writing – review & editing. Shahrel Azmin Suandi: Writing – review & editing, Supervision.
Zitieren
APA
Alkhalisy, M. A. E., Hamad, Q. S., Khayeat, A. R. H., & Suandi, S. A. (2026). Performance Analysis of YOLOv11-m and Related Architectures in Pediatric X-Ray Fracture Detection. Artificial Intelligence and Applications, 4(4), 549-560. https://doi.org/10.47852/bonviewaia62027255
Vancouver
Alkhalisy MAE, Hamad QS, Khayeat ARH, Suandi SA. Performance Analysis of YOLOv11-m and Related Architectures in Pediatric X-Ray Fracture Detection. Artificial Intelligence and Applications. 2026;4(4):549-560. doi:10.47852/bonviewaia62027255
BibTeX
@article{Alkhalisy2026,
author = {Alkhalisy, Muhanad Abdul Elah and Hamad, Qusay Shihab and Khayeat, Ali Retha Hasoon and Suandi, Shahrel Azmin},
title = {Performance Analysis of YOLOv11-m and Related Architectures in Pediatric X-Ray Fracture Detection},
journal = {Artificial Intelligence and Applications},
year = {2026},
volume = {4},
number = {4},
pages = {549--560},
doi = {10.47852/bonviewaia62027255}
}
EUVIMED