FAD-YOLO: a lightweight feature-refined and task-aligned framework for AIS–MIA discrimination on pulmonary CT - Takeaways - MDSpire

FAD-YOLO: a lightweight feature-refined and task-aligned framework for AIS–MIA discrimination on pulmonary CT

  • By

  • Jinghui Chen

  • Tao Yang

  • Zhipeng Sun

  • Chengbin Ye

  • Lianxin Xie

  • Hongjia Zhao

  • June 25, 2026

  • 0 min

Share

  • 1

    FAD-YOLO is a novel detection framework designed to differentiate adenocarcinoma in situ (AIS) from minimally invasive adenocarcinoma (MIA) in pulmonary CT imaging.

  • 2

    The framework incorporates three improvements: a Feature Refinement Feed-forward Network, a dynamic upsampling module, and a task-aligned dynamic detection head.

  • 3

    FAD-YOLO achieved a precision of 93.8% and a recall of 93.4% on an internal test set, outperforming several existing models while reducing parameters by 18.7%.

  • 4

    On an external test set, FAD-YOLO achieved mAP@50 of 91.7% without fine-tuning, demonstrating its effectiveness across different datasets.

  • 5

    The framework balances accuracy and lightweight design, making it a potential tool for aiding radiologists in resource-constrained clinical settings.

Original Source(s)

FAD-YOLO: a lightweight feature-refined and task-aligned framework for AIS–MIA discrimination on pulmonary CT

  • by Jinghui Chen, Tao Yang, Zhipeng Sun, Chengbin Ye, Lianxin Xie, Hongjia Zhao

  • June 25, 2026

Related Content

Correction: MicroRNA-based integrated diagnosis and therapy for GBM: current status and advances

Case Report: First report of PIK3CA-related overgrowth spectrum revealed by refractory chylorrhea after breast-conserving surgery: failed thoracic-duct ligation

Evaluation of the CIB1R peptide derived from the cytoplasmic domain of neprilysin on cell migration in an in vitro model of lung cancer