Scholarlink

In recent years, artificial intelligence methods have been widely adopted in patient-specific quality assurance (PSQA) research to predict gamma pass rates (GPR), aiming to enhance verification efficiency. This study proposes a Multimodel Cascade Feature Fusion Network (MCFF-Net), which integrates cross-modal feature extraction and fusion mechanisms to accurately predict GPR for radiotherapy plans, thereby supporting quality assurance in radiotherapy planning. Retrospectively, 133 patient plans utilizing intensity-modulated radiotherapy (IMRT) and volumetric modulated arc therapy (VMAT) techniques were collected. The data were divided into training and testing sets at an 8:2 ratio. The model demonstrated strong predictive performance on the test set, with mean absolute errors (MAE) of 2.793, 1.584, and 0.821, and root mean square errors (RMSE) of 3.582, 2.192, and 1.215 under the 2%/2mm, 3%/2mm, and 3%/3mm criteria, respectively. Additionally, the Pearson correlation coefficients under the 2%/2mm and 3%/2mm criteria were 0.670 and 0.669. The results indicate that the MCFF-Net model, through cross-modal data fusion, can precisely predict gamma pass rates for radiotherapy plan fields, providing effective technical support for clinical radiotherapy quality assurance.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Prediction of Patient Plan Quality Assurance Based on a Multimodal Cascade Feature Fusion Model

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Jun Zhang,
  • PeiSen Zhao,
  • YuQuan Wang,
  • HongXia Deng,
  • RuiJie Yang

摘要

In recent years, artificial intelligence methods have been widely adopted in patient-specific quality assurance (PSQA) research to predict gamma pass rates (GPR), aiming to enhance verification efficiency. This study proposes a Multimodel Cascade Feature Fusion Network (MCFF-Net), which integrates cross-modal feature extraction and fusion mechanisms to accurately predict GPR for radiotherapy plans, thereby supporting quality assurance in radiotherapy planning. Retrospectively, 133 patient plans utilizing intensity-modulated radiotherapy (IMRT) and volumetric modulated arc therapy (VMAT) techniques were collected. The data were divided into training and testing sets at an 8:2 ratio. The model demonstrated strong predictive performance on the test set, with mean absolute errors (MAE) of 2.793, 1.584, and 0.821, and root mean square errors (RMSE) of 3.582, 2.192, and 1.215 under the 2%/2mm, 3%/2mm, and 3%/3mm criteria, respectively. Additionally, the Pearson correlation coefficients under the 2%/2mm and 3%/2mm criteria were 0.670 and 0.669. The results indicate that the MCFF-Net model, through cross-modal data fusion, can precisely predict gamma pass rates for radiotherapy plan fields, providing effective technical support for clinical radiotherapy quality assurance.