AI 中文总结
针对现场图像损坏下疟疾细胞分类鲁棒性问题，提出MalariaCascade两阶段级联，以轻量哨兵筛选并升级至隔离干净输入的专家模型，在保持临床安全约束下显著优于单模型基线。
AI 中文摘要
基于薄血涂片显微镜的自动化疟疾诊断可显著减轻资源匮乏实验室的负担，但一个在干净实验室图像上最大化准确率的模型，一旦廉价的智能手机摄像头引入传感器噪声，就会严重失效。本文提出MalariaCascade，一种两阶段推理系统，其中轻量级MobileNetV2哨兵模型（2,225,153个参数）以低计算成本做出高置信度分类，并将不确定病例升级至EfficientNet-B3专家模型（10,697,769个参数），该专家模型无论输入帧如何损坏，都仅处理干净、标准化的图像。专家阶段的结构性隔离（而非学习到的鲁棒性）是核心机制。哨兵模型在安全评分目标下训练，该目标在优化精确率之前，对Recall(Parasitised)（≥0.95）和Recall(Uninfected)（≥0.40）设置显式下限，确保检查点通过构造满足临床安全约束。在NIH疟疾细胞图像数据集（27,558个细胞）上，级联在干净测试集上达到Accuracy=0.9736，Recall(Parasitised)=0.9570，Precision(Parasitised)=0.9912，F1=0.9738，AUROC=0.9955。在完全严重度的高斯传感器噪声下，级联的Recall(Parasitised)仅下降2.4个百分点（从0.9570降至0.9329），而平坦的单模型基线则崩溃63.0个百分点（从0.9584降至0.3281）。McNemar检验确认级联改进具有统计显著性（χ²=11.14，p=0.00085）。一项隔离结构特性的消融实验表明，在传感器噪声下，相对于专家模型也接收损坏输入的级联，将干净图像路由至专家模型带来了16.8个百分点的Recall(Parasitised)优势。
英文摘要
Automated malaria diagnosis from thin blood-smear microscopy could meaningfully reduce the burden on under-resourced laboratories, but a model that maximises accuracy on clean laboratory images fails badly the moment an inexpensive smartphone camera introduces sensor noise. This paper introduces MalariaCascade, a two-stage inference system in which a lightweight MobileNetV2 sentinel (2,225,153 parameters) makes confident classifications at low compute cost and escalates uncertain cases to an EfficientNet-B3 expert (10,697,769 parameters) that sees only clean, standardised images regardless of how corrupted the incoming frame is. Structural isolation of the expert stage, not learned robustness, is the mechanism. The sentinel is trained under a safety-score objective that places an explicit floor on Recall(Parasitised) (>=0.95) and Recall(Uninfected) (>=0.40) before precision is optimised, ensuring the checkpoint satisfies clinical safety constraints by construction. On the NIH Malaria Cell Images Dataset (27,558 cells), the cascade reaches Accuracy=0.9736, Recall(Parasitised)=0.9570, Precision(Parasitised)=0.9912, F1=0.9738, and AUROC=0.9955 on the clean test set. Under Gaussian sensor noise at full severity, cascade Recall(Parasitised) degrades by only 2.4 pp (0.9570 to 0.9329), while the flat single-model baseline collapses by 63.0 pp (0.9584 to 0.3281). McNemar's test confirms the cascade improvement is statistically significant (chi^2=11.14, p=0.00085). An ablation isolating the structural property shows that routing clean images to the expert is responsible for a 16.8 pp Recall(Parasitised) advantage under sensor noise relative to a cascade where the expert also sees corrupted inputs.
Comments28 pages, 6 figures, 8 tables. Code available at https://github.com/josehagbe3/MalariaVision
