Resolution-based distillation for efficient histology image classification.
Resolution-based distillation for efficient histology image classification.
复制标题
DOI：
10.1016/j.artmed.2021.102136
复制
发表时间：
2021-09
影响因子：
7.5
通讯作者：
Hassanpour S
中科院分区：
文献类型：
--
作者：
DiPalma J;Suriawinata AA;Tafe LJ;Torresani L;Hassanpour S
开发深度学习模型来分析组织学图像在计算上具有挑战性，因为图像的巨大尺寸会对计算管道的所有部分造成过大的压力。本文提出了一种新的基于深度学习的方法来提高组织学图像分类的计算效率。该方法在处理输入分辨率降低的图像时具有鲁棒性，并且可以使用有限的标记数据进行有效训练。此外，我们的方法可以在组织或幻灯片级别操作，从而消除了费力的贴片级别标记的需要。我们的方法使用知识蒸馏将知识从以高分辨率预训练的教师模型转移到以相当低的分辨率在相同图像上训练的学生模型。此外，为了解决缺乏大规模标记组织学图像数据集的问题，我们以自监督的方式执行知识蒸馏。我们用与乳糜泻、肺腺癌和肾细胞癌相关的三种不同的组织学图像数据集来评估我们的方法。我们在这些数据集上的结果表明，知识蒸馏和自我监督的结合允许学生模型接近并在某些情况下超过教师模型的分类精度，同时计算效率更高。此外，我们观察到，随着未标记数据集的大小增加，学生的分类性能也有所提高，这表明该方法有可能在更多未标记数据的情况下进一步扩展。我们的模型在准确性、f1评分、精度和召回率方面优于高分辨率的乳糜泻教师模型，而需要的计算量减少了4倍。对于肺腺癌，我们在1.25倍放大下的结果与教师模型在10倍放大下的结果相差不到1.5%，计算成本降低了64倍。我们的肾细胞癌模型在1.25倍放大率下的表现比教师模型在5倍放大率下的表现差不到1%，而计算量却减少了16倍。此外，我们的乳糜泻结果受益于使用更多未标记数据的额外性能缩放。在0.625倍放大倍率的情况下，使用未标记的数据比组织水平基线提高了4%的准确性。因此，我们的方法可以提高标准计算硬件和基础设施上数字病理学深度学习解决方案的可行性。
Developing deep learning models to analyze histology images has been computationally challenging, as the massive size of the images causes excessive strain on all parts of the computing pipeline. This paper proposes a novel deep learning-based methodology for improving the computational efficiency of histology image classification. The proposed approach is robust when used with images that have reduced input resolution, and it can be trained effectively with limited labeled data. Moreover, our approach operates at either the tissue- or slide-level, removing the need for laborious patch-level labeling. Our method uses knowledge distillation to transfer knowledge from a teacher model pre-trained at high resolution to a student model trained on the same images at a considerably lower resolution. Also, to address the lack of large-scale labeled histology image datasets, we perform the knowledge distillation in a self-supervised fashion. We evaluate our approach on three distinct histology image datasets associated with celiac disease, lung adenocarcinoma, and renal cell carcinoma. Our results on these datasets demonstrate that a combination of knowledge distillation and self-supervision allows the student model to approach and, in some cases, surpass the teacher model’s classification accuracy while being much more computationally efficient. Additionally, we observe an increase in student classification performance as the size of the unlabeled dataset increases, indicating that there is potential for this method to scale further with additional unlabeled data. Our model outperforms the high-resolution teacher model for celiac disease in accuracy, F1-score, precision, and recall while requiring 4 times fewer computations. For lung adenocarcinoma, our results at 1.25x magnification are within 1.5% of the results for the teacher model at 10x magnification, with a reduction in computational cost by a factor of 64. Our model on renal cell carcinoma at 1.25x magnification performs within 1% of the teacher model at 5x magnification while requiring 16 times fewer computations. Furthermore, our celiac disease outcomes benefit from additional performance scaling with the use of more unlabeled data. In the case of 0.625x magnification, using unlabeled data improves accuracy by 4% over the tissue-level baseline. Therefore, our approach can improve the feasibility of deep learning solutions for digital pathology on standard computational hardware and infrastructures.
登录
查看更多内容
DOI：
10.1097/pas.0b013e3181b8cf03
发表时间：
2009-12
期刊：
The American journal of surgical pathology
影响因子：
--
作者：
Girard N;Deshpande C;Lau C;Finley D;Rusch V;Pao W;Travis WD
通讯作者：
Travis WD
影响因子：
3.5
作者：
Griebel L;Prokosch HU;Köpcke F;Toddenroth D;Christoph J;Leb I;Engel I;Sedlmayr M
通讯作者：
Sedlmayr M
DOI：
10.1016/j.patol.2019.02.004
发表时间：
2019-10-01
期刊：
Revista espanola de patologia : publicacion oficial de la Sociedad Espanola de Anatomia Patologica y de la Sociedad Espanola de Citologia
影响因子：
--
作者：
Gutierrez Olivares, Victor Manuel;Gonzalez Torres, Luz Mery;Niebles De la Hoz, Maria Camila
通讯作者：
Niebles De la Hoz, Maria Camila
影响因子：
--
作者：
Chen, Pingjun;Yang, Lin
通讯作者：
Yang, Lin
通过选择性知识蒸馏进行野外低分辨率人脸识别
影响因子：
10.6
作者：
Ge, Shiming;Zhao, Shengwei;Li, Jia
通讯作者：
Li, Jia