长尾医学诊断与关系意识的表示学习和代分类器校准
Li Pan1, Yupei Zhang2, Qiushi Yang3
1Department of Pathology, The University of Hong Kong, Hong Kong.
Computers in biology and medicine
|February 20, 2025
概括
长尾医学诊断 (LMD) 框架通过解决样本不平衡,改善了医学成像中的罕见疾病检测. 它提高了代表性不足的类别的分类器性能,从而导致更准确的计算机辅助诊断.
科学领域:
- 医疗成像医学成像
- 计算机辅助诊断 计算机辅助诊断
- 机器学习 机器学习
背景情况:
- 计算机辅助诊断 (CAD) 工具减轻了临床医生的工作量,但与不平衡的数据集作斗争,经常误诊罕见疾病.
- 医疗成像中长尾问题的现有方法面临着因有限的稀有类样本而带来的偏差学习和不良分类器校准的挑战.
研究的目的:
- 引入一种新的长尾医学诊断 (LMD) 框架,用于平衡的医学图像分类.
- 提高计算机辅助诊断系统在具有显著类不平衡的数据集上的性能.
主要方法:
- 开发了一个关系意识的表示学习 (RRL) 方案,通过利用数据增强来改善特征提取.
- 提出了一种使用预期最大化的代分类器校准 (ICC) 方案,以生成平衡的虚拟特征并完善分类器.
- 实施了两阶段的方法,重点是表示学习和对不平衡的医疗数据进行分类器校准.
主要成果:
- 在三个公开的长尾医学数据集上,LMD框架显示了与最先进的方法相比的显著改进.
- RRL和ICC方案有效地解决了偏见的表示学习和不充分的分类器校准问题.
- 在不平衡的医学图像数据集中,在划分少数 (罕见) 疾病类别方面取得了卓越的表现.
结论:
- 拟议的LMD框架提供了一个强大的解决方案,以在长尾数据分布存在的情况下,为平衡的医疗图像分类提供了强大的解决方案.
- 在计算机辅助诊断中,RRL和ICC的结合有效地减轻了多数类和少数类之间的绩效差异.
- 该框架有望提高诊断AI在临床实践中的可靠性和公平性.
更多相关视频
05:56Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application
Published on: April 14, 2023
2.3K
04:09Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma
Published on: October 10, 2018
8.2K
相关概念视频
01:15Receiver Operating Characteristic Plot
71
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...
71
02:39lncRNA - Long Non-coding RNAs
8.5K
In humans, more than 80% of the genome gets transcribed. However, only around 2% of the genome codes for proteins. The remaining part produces non-coding RNAs which includes ribosomal RNAs, transfer RNAs, telomerase RNAs, and regulatory RNAs, among other types. A large number of regulatory non-coding RNAs have been classified into two groups depending upon their length – small non-coding RNAs, such as microRNA, which are less than 200 nucleotides in length, and long non-coding RNA...
8.5K