借鉴成人模型:人工智能算法在儿童脑与肿瘤分割中的可迁移性
该回顾性队列研究在42例儿童患者中比较了成人训练的HD-BET脑提取与HD-GLIOMA肿瘤分割算法与专家手动分割的Dice系数,脑提取表现优异(平均Dice 0.99),对比增强肿瘤部分中等(平均Dice 0.67),非增强T2信号异常较弱(平均Dice 0.41);基于自动掩膜的MD值进行肿瘤分级的AUC为0.86,低于专家掩膜的AUC 1,但差异未达统计学显著(p=0.208)。
要点
- HD-BET在儿童脑提取中表现优异,平均Dice 0.99(范围0.85-1)。
- HD-GLIOMA对对比增强肿瘤分割中等,平均Dice 0.67(范围0-1);对非增强T2信号异常较弱,平均Dice 0.41。
- 基于自动掩膜MD值的肿瘤分级AUC为0.86,专家掩膜AUC为1,两者差异未达统计学显著(p=0.208)。
核心结论
- HD-BET可可靠提取儿童脑;HD-GLIOMA对对比增强肿瘤部分分割效果中等;未经优化的成人脑肿瘤AI算法用于儿童患者时,依临床场景可能获得可接受的结果。
研究设计与样本
- 回顾性队列研究,纳入42例患者(平均年龄7岁,9例小于2岁,26例男性),比较自动分割与专家掩膜,采用Dice系数评估相似性,多变量回归分析协变量影响,并探索基于扩散数据的自动肿瘤分类可行性。
诊断标准与关键数值
- 脑提取:平均Dice 0.99(范围0.85-1)。
- 对比增强肿瘤分割:平均Dice 0.67(范围0-1)。
- 非增强T2信号异常分割:平均Dice 0.41。
- 基于自动掩膜平均扩散率(MD)值的自动肿瘤分级AUC=0.86;基于专家掩膜MD值的AUC=1,两者比较p=0.208。
局限性
- 原文未报告局限性。
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)。
英文原文(摘要)· 点击展开
PURPOSE: AI brain tumour segmentation and brain extraction algorithms promise better diagnostic and follow-up of brain tumours in adults. The development of such tools for paediatric populations is restricted by limited training data but careful adaption of adult algorithms to paediatric population might be a solution. Here, we aim exploring the transferability of algorithms for brain (HD-BET) and tumour segmentation (HD-GLIOMA) in adults to paediatric imaging studies.
METHOD: In a retrospective cohort, we compared automated segmentation with expert masks. We used the dice coefficient for evaluating the similarity and multivariate regressions for the influence of covariates. We explored the feasibility of automatic tumor classification based on diffusion data.
RESULTS: In 42 patients (mean age 7 years, 9 below 2 years, 26 males), segmentation was excellent for brain extraction (mean dice 0.99, range 0.85-1), moderate for segmentation of contrast-enhancing tumours (mean dice 0.67, range 0-1), and weak for non-enhancing T2-signal abnormalities (mean dice 0.41). Precision was better for enhancing tumour parts (p < 0.001) and for malignant histology (p = 0.006 and p = 0.012) but independent from myelinisation as indicated by the age (p = 0.472). Automated tumour grading based on mean diffusivity (MD) values from automated masks was good (AUC = 0.86) but tended to be less accurate than MD values from expert masks (AUC = 1, p = 0.208).
CONCLUSION: HD-BET provides a reliable extraction of the paediatric brain. HD-GLIOMA works moderately for contrast-enhancing tumours parts. Without optimization, brain tumor AI algorithms trained on adults and used on paediatric patients may yield acceptable results depending on the clinical scenario.
引用
- Drai, Testud, Brun, et al. Borrowing strength from adults: Transferability of AI algorithms for paediatric brain and tumour segmentation. Eur J Radiol. 2022. PMID: 35405580. doi:10.1016/j.ejrad.2022.110291
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。