可解释AI提升多巴胺转运体SPECT分类的临床可接受度
本研究回顾性纳入1296例临床多巴胺转运体SPECT,以两名有经验阅片者的二分类视觉判读为金标准,用1008例训练定制卷积神经网络(CNN),余288例评估分类性能并检验逐层相关性传播(LRP)对CNN决策的解释能力。结果显示CNN总准确率、敏感性和特异性分别为95.8%、92.8%和98.7%,LRP可为每例SPECT提供易于解读的相关性图,总计算时间3秒。
要点
- CNN自动分类DAT-SPECT的总准确率95.8%、敏感性92.8%、特异性98.7%。
- LRP相关性图在每例DAT-SPECT中均易于解读,可解释CNN决策依据。
- 结果提示LRP可支持CNN分类用于临床常规,总计算时间3秒。
核心结论
- LRP能为单例DAT-SPECT的CNN分类决策提供解释,因此可推荐用于支持临床常规中的CNN分类。
研究设计与样本
- 回顾性研究,纳入1296例临床DAT-SPECT,以两名有经验阅片者的二分类视觉判读「正常」或「减低」为金标准。
- 用1008例随机选取的DAT-SPECT训练定制CNN,剩余288例用于评估分类性能并检验LRP解释能力。
主要发现
- CNN总体准确率95.8%、敏感性92.8%、特异性98.7%。
- LRP在每例DAT-SPECT中均提供易于解读的相关性图。
- 在所有「减低」DAT-SPECT中,黑质纹状体变性最重侧半球的壳核是CNN分类最相关的脑区。
- 部分误分类DAT-SPECT呈现更符合真实类别的「不一致」相关性图。
关键数值与推荐
- CNN总体准确率95.8%、敏感性92.8%、特异性98.7%。
- 训练集1008例,测试集288例,总样本1296例。
- LRP总计算时间3秒。
临床怎么用
- 可将LRP相关性图作为CNN辅助判读DAT-SPECT的解释工具,提升临床对自动分类结果的接受度。
- 对相关性图与预测类别「不一致」的病例应警惕误分类可能,但该用途尚待进一步研究。
局限性
- 为回顾性研究,金标准为两名阅片者的视觉二分类判读而非临床随访诊断。
- 「不一致」相关性图识别误分类病例的效用尚需进一步研究验证。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
PURPOSE: Deep convolutional neural networks (CNN) provide high accuracy for automatic classification of dopamine transporter (DAT) SPECT images. However, CNN are inherently black-box in nature lacking any kind of explanation for their decisions. This limits their acceptance for clinical use. This study tested layer-wise relevance propagation (LRP) to explain CNN-based classification of DAT-SPECT in patients with clinically uncertain parkinsonian syndromes. METHODS: The study retrospectively included 1296 clinical DAT-SPECT with visual binary interpretation as "normal" or "reduced" by two experienced readers as standard-of-truth. A custom-made CNN was trained with 1008 randomly selected DAT-SPECT. The remaining 288 DAT-SPECT were used to assess classification performance of the CNN and to test LRP for explanation of the CNN-based classification. RESULTS: Overall accuracy, sensitivity, and specificity of the CNN were 95.8%, 92.8%, and 98.7%, respectively. LRP provided relevance maps that were easy to interpret in each individual DAT-SPECT. In particular, the putamen in the hemisphere most affected by nigrostriatal degeneration was the most relevant brain region for CNN-based classification in all reduced DAT-SPECT. Some misclassified DAT-SPECT showed an "inconsistent" relevance map more typical for the true class label. CONCLUSION: LRP is useful to provide explanation of CNN-based decisions in individual DAT-SPECT and, therefore, can be recommended to support CNN-based classification of DAT-SPECT in clinical routine. Total computation time of 3 s is compatible with busy clinical workflow. The utility of "inconsistent" relevance maps to identify misclassified cases requires further investigation.
引用
- Nazari M, Kluge A, Apostolova I, et al. Explainable AI to improve acceptance of convolutional neural networks for automatic classification of dopamine transporter SPECT in the diagnosis of clinically uncertain parkinsonian syndromes. Eur J Nucl Med Mol Imaging. 2021. PMID: 34651223. doi:10.1007/s00259-021-05569-9
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。