arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2302.10185cs.CVcs.AIcs.LG
收藏

基于不确定采样、标注冗余约束和数据冗余的脑肿瘤分割中的主动学习

Active Learning in Brain Tumor Segmentation with Uncertainty Sampling, Annotation Redundancy Restriction, and Data Initialization

Warren Alpert Medical School of Brown University(布朗大学沃伦·阿尔珀特医学院);Rhode Island Hospital(罗德岛医院);Perelman School of Medicine at the University of Pennsylvania(宾夕法尼亚大学佩雷尔曼医学院);Second Xiangya Hospital of Central South University(中南大学湘雅二医院);University of Virginia(弗吉尼亚大学);Johns Hopkins University(约翰霍普金斯大学);Central South University(中南大学);Brown University(布朗大学)
Daniel D Kim, Rajat S Chandra, Jian Peng, Jing Wu, Xue Feng, Michael Atalay, Chetan Bettegowda, Craig Jones, Haris Sair, Wei-hua Liao, Chengzhang Zhu, Beiji Zou… 展开作者

Daniel D Kim, Rajat S Chandra, Jian Peng, Jing Wu, Xue Feng, Michael Atalay, Chetan Bettegowda, Craig Jones, Haris Sair, Wei-hua Liao, Chengzhang Zhu, Beiji Zou, Li Yang, Anahita Fathi Kazerooni, Ali Nabavizadeh, Harrison X Bai, Zhicheng Jiao

摘要:

深度学习模型在医学3D成像中表现出了巨大的潜力,但其发展受到所需的大量注释数据的限制。主动学习(AL)通过在信息量最大的数据样本的子集上训练模型来解决这个问题,而不会影响性能。我们比较了不同的AL策略,并提出了一个框架,最大限度地减少了国家的最先进的性能所需的数据量。使用638个多机构脑肿瘤MRI图像来训练3D U-网络模型并比较AL策略。我们研究了不确定性采样,注释冗余限制,和初始数据集选择技术。不确定性估计技术,包括贝叶斯估计与脱落,自举,和利润率抽样进行了比较,随机查询。还考虑了通过去除待注释子集内的相似图像来避免注释冗余的策略。我们确定了实现与在完整数据集上训练的模型相似性能所需的最小数据量(α = 0.1)。还提出了一种基于方差的选择策略,使用放射组学来识别初始训练数据集。在训练和测试时具有丢失的贝叶斯近似显示出与全数据模型相似的结果,其中小于20%的训练数据(p=0.293)与随机查询相比,在56.5%的训练数据(p=0.814)处实现相似的性能。注释冗余限制技术在大约40%-50%的训练数据上实现了最先进的性能。放射组学数据集初始化具有较高的Dice,初始数据集大小为20和80张图像,但改进不显著。总之,我们研究了各种AL策略与辍学不确定性估计实现最先进的性能与最少的注释数据。

英文摘要:

Deep learning models have demonstrated great potential in medical 3D imaging, but their development is limited by the expensive, large volume of annotated data required. Active learning (AL) addresses this by training a model on a subset of the most informative data samples without compromising performance. We compared different AL strategies and propose a framework that minimizes the amount of data needed for state-of-the-art performance. 638 multi-institutional brain tumor MRI images were used to train a 3D U-net model and compare AL strategies. We investigated uncertainty sampling, annotation redundancy restriction, and initial dataset selection techniques. Uncertainty estimation techniques including Bayesian estimation with dropout, bootstrapping, and margins sampling were compared to random query. Strategies to avoid annotation redundancy by removing similar images within the to-be-annotated subset were considered as well. We determined the minimum amount of data necessary to achieve similar performance to the model trained on the full dataset (α = 0.1). A variance-based selection strategy using radiomics to identify the initial training dataset is also proposed. Bayesian approximation with dropout at training and testing showed similar results to that of the full data model with less than 20% of the training data (p=0.293) compared to random query achieving similar performance at 56.5% of the training data (p=0.814). Annotation redundancy restriction techniques achieved state-of-the-art performance at approximately 40%-50% of the training data. Radiomics dataset initialization had higher Dice with initial dataset sizes of 20 and 80 images, but improvements were not significant. In conclusion, we investigated various AL strategies with dropout uncertainty estimation achieving state-of-the-art performance with the least annotated data.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑