Kong Guanqing, Wu Chuanfu, Zhang Zongqiu, Yin Chuansheng, Qin Dawei
Linyi People's Hospital, Linyi City, Shandong Province, China.
Linyi Key Laboratory of Health Data Science, Linyi City, Shandong Province, China.
Front Neuroinform. 2024 Jul 29;18:1403732. doi: 10.3389/fninf.2024.1403732. eCollection 2024.
Brain diseases, particularly the classification of gliomas and brain metastases and the prediction of HT in strokes, pose significant challenges in healthcare. Existing methods, relying predominantly on clinical data or imaging-based techniques such as radiomics, often fall short in achieving satisfactory classification accuracy. These methods fail to adequately capture the nuanced features crucial for accurate diagnosis, often hindered by noise and the inability to integrate information across various scales.
We propose a novel approach that mask attention mechanisms with multi-scale feature fusion for Multimodal brain disease classification tasks, termed , which aims to extract features highly relevant to the disease. The extracted features are then dimensionally reduced using Principal Component Analysis (PCA), followed by classification with a Support Vector Machine (SVM) to obtain the predictive results.
Our methodology underwent rigorous testing on multi-parametric MRI datasets for both brain tumors and strokes. The results demonstrate a significant improvement in addressing critical clinical challenges, including the classification of gliomas, brain metastases, and the prediction of hemorrhagic stroke transformations. Ablation studies further validate the effectiveness of our attention mechanism and feature fusion modules.
These findings underscore the potential of our approach to meet and exceed current clinical diagnostic demands, offering promising prospects for enhancing healthcare outcomes in the diagnosis and treatment of brain diseases.
脑部疾病，尤其是胶质瘤和脑转移瘤的分类以及中风中出血转化（HT）的预测，给医疗保健带来了重大挑战。现有方法主要依赖临床数据或基于成像的技术（如放射组学），在实现令人满意的分类准确率方面往往存在不足。这些方法未能充分捕捉对准确诊断至关重要的细微特征，常常受到噪声以及无法跨不同尺度整合信息的阻碍。
我们提出了一种新颖的方法，用于多模态脑部疾病分类任务，该方法采用多尺度特征融合的掩膜注意力机制，称为 ，旨在提取与疾病高度相关的特征。然后使用主成分分析（PCA）对提取的特征进行降维，接着使用支持向量机（SVM）进行分类以获得预测结果。
我们的方法在针对脑肿瘤和中风的多参数MRI数据集上进行了严格测试。结果表明，在应对关键临床挑战方面有显著改进，包括胶质瘤的分类、脑转移瘤以及出血性中风转化的预测。消融研究进一步验证了我们的注意力机制和特征融合模块的有效性。
这些发现强调了我们的方法满足并超越当前临床诊断需求的潜力，为改善脑部疾病诊断和治疗中的医疗保健结果提供了有前景的前景。
