Deep multi-scale resemblance network for the sub-class differentiation of adrenal masses on computed tomography images

作者：

Highlights：

• Proposed a similarity loss to derive features that was able to tolerate large intra-class variations.

• Introduced a multi-scale feature embedding to integrate complementary feature representations produced at different scales.

• Proposed to augment the training data with randomly sampled pairs to reduce the influence of imbalanced training data.

摘要

•Proposed a similarity loss to derive features that was able to tolerate large intra-class variations.•Introduced a multi-scale feature embedding to integrate complementary feature representations produced at different scales.•Proposed to augment the training data with randomly sampled pairs to reduce the influence of imbalanced training data.

论文关键词：Classification,Adrenal masses,Convolutional neural networks (CNN)

论文评审过程：Received 4 June 2020, Revised 23 March 2022, Accepted 22 April 2022, Available online 9 August 2022, Version of Record 11 August 2022.

论文官网地址：https://doi.org/10.1016/j.artmed.2022.102374