提高增强型人类智能以区分 Burkitt 淋巴瘤与弥漫性大 B 细胞淋巴瘤病例。

Improving Augmented Human Intelligence to Distinguish Burkitt Lymphoma From Diffuse Large B-Cell Lymphoma Cases.

机构信息

Department of Pathology, Scientific Computing and Imaging Institute, University of Utah, Salt Lake City.

ARUP Laboratories, Scientific Computing and Imaging Institute, University of Utah, Salt Lake City.

出版信息

Am J Clin Pathol. 2020 May 5;153(6):743-759. doi: 10.1093/ajcp/aqaa001.

DOI:10.1093/ajcp/aqaa001

PMID:32067039

Abstract

OBJECTIVES

To assess and improve the assistive role of a deep, densely connected convolutional neural network (CNN) to hematopathologists in differentiating histologic images of Burkitt lymphoma (BL) from diffuse large B-cell lymphoma (DLBCL).

METHODS

A total of 10,818 images from BL (n = 34) and DLBCL (n = 36) cases were used to either train or apply different CNNs. Networks differed by number of training images and pixels of images, absence of color, pixel and staining augmentation, and depth of the network, among other parameters.

RESULTS

Cases classified correctly were 17 of 18 (94%), nine with 100% of images correct by the best performing network showing a receiver operating characteristic curve analysis area under the curve 0.92 for both DLBCL and BL. The best performing CNN used all available training images, two random subcrops per image of 448 × 448 pixels, random H&E staining image augmentation, random horizontal flipping of images, random alteration of contrast, reduction on validation error plateau of 15 epochs, block size of six, batch size of 32, and depth of 22. Other networks and decreasing training images had poorer performance.

CONCLUSIONS

CNNs are promising augmented human intelligence tools for differentiating a subset of BL and DLBCL cases.

摘要

目的

评估和提高深度密集卷积神经网络（CNN）对血液病理学家在区分伯基特淋巴瘤（BL）和弥漫性大 B 细胞淋巴瘤（DLBCL）组织学图像方面的辅助作用。

方法

使用来自 BL（n=34）和 DLBCL（n=36）病例的总共 10818 张图像来训练或应用不同的 CNN。网络的区别在于训练图像的数量和图像的像素、是否有颜色、像素和染色增强、网络的深度等参数。

结果

正确分类的病例为 18 例中的 17 例（94%），其中表现最佳的网络对 100%的图像分类正确，该网络的接收者操作特征曲线分析曲线下面积为 0.92，用于 DLBCL 和 BL。表现最佳的 CNN 使用了所有可用的训练图像、每张图像的两个随机子裁剪（448×448 像素）、随机 H&E 染色图像增强、图像的随机水平翻转、图像对比度的随机调整、验证错误平台减少 15 个周期、块大小为 6、批量大小为 32、深度为 22。其他网络和减少的训练图像表现较差。