Scholarlink

Background <p>Mantle cell lymphoma (MCL) is a rare, biologically heterogeneous B-cell malignancy with highly variable outcomes. Existing prognostic tools are suboptimal. We developed an interpretable deep learning framework integrating baseline [<sup>18</sup>F]FDG PET/CT and electronic health record (EHR) data for individualized risk stratification.</p> Methods <p>In this multicenter study, 187 treatment-naïve MCL patients were analyzed. A mixture-of-experts (MoE) fusion network integrated multimodal representations from PET/CT and EHR data. Expert modules comprising vision encoders, radiomics extractors, and a medical language model were integrated through an attention-based gating mechanism to construct multimodal radiomic signatures (R-signatures) predictive of progression-free survival (PFS) and overall survival (OS). R-signatures were validated and incorporated with clinical and metabolic factors into multiparametric models. Deep learning model interpretability was evaluated using attention visualization, expert-level contributions and pathologic correlation.</p> Results <p>R-signatures robustly discriminated relapse (AUC = 0.893 training, 0.755 validation) and death (AUC = 0.804 and 0.844), and independently predicted adverse outcomes (PFS: HR = 27.70, <i>P</i> &lt; 0.001; OS: HR = 6.86, <i>P</i> = 0.001). Multiparametric models integrating R-signatures with total lesion glycolysis, β2-microglobulin, WBC, and Ki-67 outperformed conventional indices (C-indices: PFS 0.892 training, 0.781 validation; OS 0.877 training, 0.862 validation). Time-dependent ROC analyses consistently showed AUCs approaching or exceeding 0.800. Calibration and decision curve analyses confirmed excellent agreement and superior clinical net benefit. Attention maps localized high-weighted regions to hypermetabolic tumor areas, with higher R-signature values in blastoid and pleomorphic variants versus classical histology (<i>P</i> = 0.028 and <i>P</i> = 0.010).</p> Conclusions <p>This interpretable PET/CT-EHR fusion framework substantially improves prognostic precision in MCL, providing a noninvasive, clinically translatable tool for risk-adapted management.</p>

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

Interpretable multimodal PET/CT-EHR fusion via mixture-of-experts for prognostic stratification in mantle cell lymphoma: a multicenter study

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月16日

查看全文
  • Chong Jiang,
  • Zitong Zhang,
  • Zekun Jiang,
  • Chongyang Ding,
  • Yue Teng,
  • Limin Gao,
  • Ming Jiang,
  • Linhao Qu,
  • Rong Tian

摘要

Background

Mantle cell lymphoma (MCL) is a rare, biologically heterogeneous B-cell malignancy with highly variable outcomes. Existing prognostic tools are suboptimal. We developed an interpretable deep learning framework integrating baseline [18F]FDG PET/CT and electronic health record (EHR) data for individualized risk stratification.

Methods

In this multicenter study, 187 treatment-naïve MCL patients were analyzed. A mixture-of-experts (MoE) fusion network integrated multimodal representations from PET/CT and EHR data. Expert modules comprising vision encoders, radiomics extractors, and a medical language model were integrated through an attention-based gating mechanism to construct multimodal radiomic signatures (R-signatures) predictive of progression-free survival (PFS) and overall survival (OS). R-signatures were validated and incorporated with clinical and metabolic factors into multiparametric models. Deep learning model interpretability was evaluated using attention visualization, expert-level contributions and pathologic correlation.

Results

R-signatures robustly discriminated relapse (AUC = 0.893 training, 0.755 validation) and death (AUC = 0.804 and 0.844), and independently predicted adverse outcomes (PFS: HR = 27.70, P < 0.001; OS: HR = 6.86, P = 0.001). Multiparametric models integrating R-signatures with total lesion glycolysis, β2-microglobulin, WBC, and Ki-67 outperformed conventional indices (C-indices: PFS 0.892 training, 0.781 validation; OS 0.877 training, 0.862 validation). Time-dependent ROC analyses consistently showed AUCs approaching or exceeding 0.800. Calibration and decision curve analyses confirmed excellent agreement and superior clinical net benefit. Attention maps localized high-weighted regions to hypermetabolic tumor areas, with higher R-signature values in blastoid and pleomorphic variants versus classical histology (P = 0.028 and P = 0.010).

Conclusions

This interpretable PET/CT-EHR fusion framework substantially improves prognostic precision in MCL, providing a noninvasive, clinically translatable tool for risk-adapted management.