在有限的乳腺造影数据集中,培训和测试集之间的分类性能偏差在有限的乳腺造影数据集中
Rui Hou1,2, Joseph Y Lo2, Jeffrey R Marks3
1Department of Artificial Intelligence, Beijing University of Posts and Telecommunications, Beijing, China.
PloS one
|February 7, 2024
概括
哺乳镜放射学研究中的数据分割可能会导致性能偏差. 在有限的数据集上训练的模型可能会产生不可靠的结论,强调需要谨慎的测试集选择策略.
科学领域:
- 医学成像分析分析 医学成像分析
- 无线电学和机器学习
背景情况:
- 乳房镜中的放射学研究对于癌症诊断至关重要.
- 医学成像中的有限数据集大小可以引入偏差.
- 来自数据采样的性能偏差影响了模型的可靠性.
研究的目的:
- 为了评估乳房扫描中的性能偏差,由于数据分裂而导致的放射学.
- 评估培训和测试集选择对模型结果的影响.
- 调查在有限的医学成像数据集中减轻偏差的策略.
主要方法:
- 利用700名女性的乳房影像检测管道癌在现场升级.
- 重复将数据分成训练 (n=400) 和测试 (n=300) 集40次.
- 采用了具有放射学和临床特征的后勤回归和支向量机器.
主要成果:
- 在数据分割中观察到曲线下面积 (AUC) 的显著变化.
- 确定了性能权衡:提高训练准确性,有时降低测试准确性.
- 交叉验证减少了变异性,但需要超过500个案例进行代表性估计.
结论:
- 数据采样偏差可能导致乳房扫描放射学中不适当的结论.
- 在非代表性子集上训练的模型可能无法很好地概括.
- 开发最佳的测试集选择对于可靠的医学成像研究至关重要.
相关概念视频
Receiver Operating Characteristic Plot
179
A ROC (Receiver Operating Characteristic) plot is a graphical tool used to assess the performance of a binary classification model by illustrating the trade-off between sensitivity (true positive rate) and specificity (false positive rate). By plotting sensitivity against 1 - specificity across various threshold settings, the ROC curve shows how well the model distinguishes between classes, with a curve closer to the top-left corner indicating a more accurate model. The area under the ROC curve...
179
Multiple Comparison Tests
3.9K
Multiple comparison test, abbreviated as MCT, is a post hoc analysis generally performed after comparing multiple samples with one or more tests. An MCT will help identify a significantly different sample among multiple samples or a factor among multiple factors.
It would be easy to compare two samples using a significance alpha level of 0.05. In other words, there is only one sample pair to be compared. However, it would be difficult to identify a significantly different sample if the number...
It would be easy to compare two samples using a significance alpha level of 0.05. In other words, there is only one sample pair to be compared. However, it would be difficult to identify a significantly different sample if the number...
3.9K


