Anatomical-Contextual YOLOv8-YOLOv12 Framework for Pulmonary Nodule Detection in CT: Multi-Organ Learning and Cross-Dataset Validation.
Authors
Affiliations (4)
Affiliations (4)
- Cyber-Physical Systems Laboratory, Centro de Investigación e Innovación Tecnológica (CIITEC), Instituto Politécnico Nacional (IPN), Cerrada Cecati s/n Col. Sta. Catarina, Azcapotzalco, Ciudad de México C.P. 02250, Mexico.
- Centro de Investigación en Ciencia Aplicada y Tecnología Avanzada, Unidad Legaria (CICATA Legaria), Instituto Politécnico Nacional (IPN), Av. Legaria No. 694 Col. Irrigación, Miguel Hidalgo, Ciudad de México C.P. 11500, Mexico.
- CAPROM Applied Science to Mexican Problems, Venustiano Carranza 90, Azcapotzalco, Ciudad de México C.P. 02440, Mexico.
- UMAE Hospital de Oncología, Centro Médico Nacional Siglo XXI, IMSS, A. Cuauhtemoc 330, Col. Doctores, Alcaldia Cuauhtemoc, Ciudad de México C.P. 06720, Mexico.
Abstract
<b>Background/Objectives:</b> Lung cancer outcomes depend strongly on early detection, highlighting the need for accurate, robust, and interpretable methods for identifying pulmonary nodules on computed tomography (CT). This study proposes a multi-organ anatomical-contextual framework based on YOLOv8-YOLOv12, in which supporting thoracic structures are incorporated as spatial references during training while the pulmonary nodule is retained as a single invariant pathological class. <b>Methods:</b> The models were trained and internally validated on 5000 2D axial CT slices from 320 patients in the LUNG-PET-CT-DX dataset and externally evaluated on an independent LIDC-IDRI subset comprising 700 axial slices from 70 patients, without retraining or fine-tuning. A controlled ablation study comparing Nodule-only with Nodule + Anatomical Context was performed. <b>Results:</b> The anatomical-contextual scheme showed consistent improvements across all five architectures, with mean absolute gains of +0.078 in Precision, +0.122 in Recall, +0.101 in F1-score, +0.079 in mAP@50, and +0.190 in mAP@50-95. During internal validation, YOLOv10m achieved the highest Precision (0.979) and mAP@50 (0.987), as well as the shortest inference time (9.25 ms), whereas YOLOv11m achieved the highest Recall (0.971) and F1-score (0.962). During external evaluation, YOLOv10m retained the highest Precision (0.924) and mAP@50 (0.956), while YOLOv8m achieved the highest Recall (0.930) and mAP@50-95 (0.845); both models achieved an F1-score of 0.922. <b>Conclusions:</b> These findings provide experimental evidence that anatomical-contextual learning improves pulmonary nodule detection while maintaining competitive performance across datasets, supporting its potential as an anatomically informed and interpretable strategy for AI-assisted thoracic CT analysis.