Spectrochimica acta. Part A, Molecular and biomolecular spectroscopy · 2026
Deep learning-based multimodal fusion of vibrational spectroscopy and clinical data for aortic dissection diagnosis and prognosis.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Ma Guangyao, Yan Lei, Zhou Xuguang, Yang Hui, Chen Chen, Tao Jing, Liu Hao, Chen Cheng, Yang Yining
Аннотация
BACKGROUND: Aortic dissection and myocardial infarction are two life-threatening acute cardiovascular conditions with similar clinical presentations but fundamentally different treatment strategies. Misdiagnosis between these diseases can severely endanger patient survival, highlighting the urgent need for efficient and non-invasive differential diagnostic methods. Vibrational spectroscopy, characterized by its non-invasiveness, high throughput, and molecular sensitivity, offers a promising approach for early biochemical disease identification. However, single spectral modalities are limited in diagnostic performance due to insufficient complementary information.
RESULTS: To address these limitations, we propose MSFusion, a multimodal deep learning model that integrates infrared and Raman spectroscopy data for precise diagnosis and survival prognosis prediction of aortic dissection. MSFusion employs a three-branch multiscale convolutional architecture to extract hierarchical spectral representations, incorporates a cross-modal attention module to promote semantic synergy and complementary interaction between modalities, and utilizes a bidirectional multiscale fusion mechanism to align local and global features across multiple granularities.Experimental results demonstrate that MSFusion achieves 95.63% accuracy and 98.79% AUC in distinguishing aortic dissection from myocardial infarction, significantly outperforming unimodal and conventional fusion approaches. For survival prognosis, the model is further extended by incorporating clinical indicators to predict postoperative survival outcomes, achieving 81.82% accuracy and 70.00% AUC, showing its capability for effective clinical risk stratification.
SIGNIFICANCE: This study presents a multimodal deep learning framework that combines vibrational spectroscopy with clinical data for rapid and non-invasive diagnosis of aortic dissection. MSFusion effectively enhances diagnostic precision and prognostic reliability, providing a promising tool for intelligent cardiovascular disease assessment and clinical decision support.
RESULTS: To address these limitations, we propose MSFusion, a multimodal deep learning model that integrates infrared and Raman spectroscopy data for precise diagnosis and survival prognosis prediction of aortic dissection. MSFusion employs a three-branch multiscale convolutional architecture to extract hierarchical spectral representations, incorporates a cross-modal attention module to promote semantic synergy and complementary interaction between modalities, and utilizes a bidirectional multiscale fusion mechanism to align local and global features across multiple granularities.Experimental results demonstrate that MSFusion achieves 95.63% accuracy and 98.79% AUC in distinguishing aortic dissection from myocardial infarction, significantly outperforming unimodal and conventional fusion approaches. For survival prognosis, the model is further extended by incorporating clinical indicators to predict postoperative survival outcomes, achieving 81.82% accuracy and 70.00% AUC, showing its capability for effective clinical risk stratification.
SIGNIFICANCE: This study presents a multimodal deep learning framework that combines vibrational spectroscopy with clinical data for rapid and non-invasive diagnosis of aortic dissection. MSFusion effectively enhances diagnostic precision and prognostic reliability, providing a promising tool for intelligent cardiovascular disease assessment and clinical decision support.