揭秘深度学习在中医舌诊中的成功秘诀，给出可复现的设计指南。
What Makes Deep Learning Work for Traditional Chinese Medicine Tongue Diagnosis? A Comprehensive Ablation Study
- 系统消融实验验证模型架构、损失函数等关键设计因素。
- 最佳模型在1.1万样本上达加权F1 0.7761，数据量提升带来20.6%性能增益。
- 提出6条通用设计原则，尤其强调克制的数据增强与标签维度控制。
深度学习在中医舌诊自动化中展现潜力，但其设计空间仍待探索。本研究在TongueDx2（5,109张图像，976例专家标注）和合并数据集（11,101样本）上，通过严格五折交叉验证，系统评估了六种主干网络、四种损失函数、五种增强策略与六种训练策略。最佳976样本模型采用ConvNeXt-Tiny，在受限增强与弱组集成下达到加权F1 0.6625；最佳11,101样本模型加权F1达0.7761。六项核心设计原则浮现：(1) ConvNeXt-Tiny参数效率最优；(2) BCE显著优于Asymmetric Loss（+2.7%）；(3) 受控色彩增强至关重要；(4) 弱组集成替换概率平均提升+2.1%；(5) 数据量扩大带来+20.6%性能提升；(6) 标签维度从13扩至45导致灾难性崩溃（F1由0.78降至0.22）。这些原则可推广至多标签医学图像分类与类别不平衡场景。
原文摘要 · Abstract (English)
Deep learning has shown promise for automated tongue diagnosis in traditional Chinese medicine (TCM), yet the design space remains underexplored. We conducted a systematic ablation study spanning 20+ model versions under rigorous 5-fold cross-validation on TongueDx2 (5,109 images, 976 expert-annotated) and a merged dataset of 11,101 samples. We compared six backbone architectures, four loss functions, five augmentation strategies, and six training strategies. The best 976-sample model achieved weighted-F1 of 0.6625 using ConvNeXt-Tiny with restrained augmentation and weak-group ensemble, while the best 11,101-sample model reached weighted-F1 of 0.7761. Six key design principles emerged: (1) ConvNeXt-Tiny offers optimal parameter efficiency; (2) BCE substantially outperforms Asymmetric Loss (+2.7%); (3) restrained color augmentation is critical; (4) weak-group ensemble replacement (+2.1%) outperforms probability averaging; (5) data scaling yielded +20.6% improvement; (6) expanding from 13 to 45 label dimensions caused catastrophic collapse (0.78 to 0.22). These principles are generalizable to multi-label medical image classification with class imbalance.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。