Evaluation of an interpretable deep-learning model for the automated plan review of intensity-modulated radiation therapy - Takeaways - MDSpire

Evaluation of an interpretable deep-learning model for the automated plan review of intensity-modulated radiation therapy

  • By

  • Yuhan Fan

  • Jiawen Shang

  • Ke Zhang

  • Zhihui Hu

  • Zhiqiang Liu

  • Hui Yan

  • Peng Huang

  • July 15, 2026

Share

  • 1

    Intensity-modulated radiation therapy (IMRT) requires quality assurance due to its complexity and high risk.

  • 2

    An unsupervised deep-learning model using an autoencoder was developed to identify anomalies in IMRT treatment plans.

  • 3

    The feature perturbation interpretation (FPI) method was introduced to rank feature impacts on detected anomalies.

  • 4

    The autoencoder achieved an area under the curve (AUC) value of 0.98, outperforming classic detection models.

  • 5

    The combination of the autoencoder and FPI method provides a highly interpretable framework for automated plan review.

Original Source(s)

Evaluation of an interpretable deep-learning model for the automated plan review of intensity-modulated radiation therapy

  • by Yuhan Fan, Jiawen Shang, Ke Zhang, Zhihui Hu, Zhiqiang Liu, Hui Yan, Peng Huang

  • July 15, 2026

Related Content

Prognostic significance of extracapsular extension in patients with non-small cell lung cancer following neoadjuvant chemoimmunotherapy: a retrospective cohort study

A hybrid ViT-L/32–MaxViT-L architecture with adaptive gated fusion for multiclass gastrointestinal disease detection and multi-method post-hoc explainability

E3 ubiquitin ligase NKLAM/RNF19b suppresses myc-driven B cell lymphomagenesis in Eμ-myc mice