Cheng Phillip M, Malhi Harshawn S
Department of Radiology, Keck School of Medicine of USC, Los Angeles, CA, USA.
USC Norris Cancer Center and Hospital, 1441 Eastlake Avenue, Suite 2315B, Los Angeles, CA, 90033-0377, USA.
J Digit Imaging. 2017 Apr;30(2):234-243. doi: 10.1007/s10278-016-9929-2.
The purpose of this study is to evaluate transfer learning with deep convolutional neural networks for the classification of abdominal ultrasound images. Grayscale images from 185 consecutive clinical abdominal ultrasound studies were categorized into 11 categories based on the text annotation specified by the technologist for the image. Cropped images were rescaled to 256 × 256 resolution and randomized, with 4094 images from 136 studies constituting the training set, and 1423 images from 49 studies constituting the test set. The fully connected layers of two convolutional neural networks based on CaffeNet and VGGNet, previously trained on the 2012 Large Scale Visual Recognition Challenge data set, were retrained on the training set. Weights in the convolutional layers of each network were frozen to serve as fixed feature extractors. Accuracy on the test set was evaluated for each network. A radiologist experienced in abdominal ultrasound also independently classified the images in the test set into the same 11 categories. The CaffeNet network classified 77.3% of the test set images accurately (1100/1423 images), with a top-2 accuracy of 90.4% (1287/1423 images). The larger VGGNet network classified 77.9% of the test set accurately (1109/1423 images), with a top-2 accuracy of VGGNet was 89.7% (1276/1423 images). The radiologist classified 71.7% of the test set images correctly (1020/1423 images). The differences in classification accuracies between both neural networks and the radiologist were statistically significant (p < 0.001). The results demonstrate that transfer learning with convolutional neural networks may be used to construct effective classifiers for abdominal ultrasound images.
本研究的目的是评估使用深度卷积神经网络进行迁移学习以对腹部超声图像进行分类的效果。来自185例连续临床腹部超声检查的灰度图像，根据技术人员为图像指定的文本注释被分为11类。裁剪后的图像被重新缩放到256×256分辨率并随机化，来自136项研究的4094张图像构成训练集，来自49项研究的1423张图像构成测试集。基于CaffeNet和VGGNet的两个卷积神经网络的全连接层，先前在2012年大规模视觉识别挑战赛数据集上进行过训练，在训练集上进行重新训练。每个网络卷积层中的权重被冻结以用作固定特征提取器。对每个网络在测试集上的准确率进行评估。一位在腹部超声方面有经验的放射科医生也将测试集中的图像独立分类为相同的11类。CaffeNet网络准确分类了测试集中77.3%的图像（1100/1423张图像），前两名准确率为90.4%（1287/1423张图像）。更大的VGGNet网络准确分类了测试集中77.9%的图像（1109/1423张图像），VGGNet的前两名准确率为89.7%（1276/1423张图像）。放射科医生正确分类了测试集中71.7%的图像（1020/1423张图像）。两个神经网络与放射科医生之间的分类准确率差异具有统计学意义（p<0.001）。结果表明，使用卷积神经网络进行迁移学习可用于构建有效的腹部超声图像分类器。
