A multi-task CNN approach for lung nodule malignancy classification and characterization

作者：

Highlights：

• Malignancy prediction is better with branches for nodule characterization.

• Most relevant features for malignancy were lobulation, spiculation and texture.

• Image features should be considered for nodule malignancy classification.

摘要

•Malignancy prediction is better with branches for nodule characterization.•Most relevant features for malignancy were lobulation, spiculation and texture.•Image features should be considered for nodule malignancy classification.

论文关键词：Lung cancer,Malignancy,Multitasking classification,Convolutional neural networks,Deep learning

论文评审过程：Received 28 September 2020, Revised 23 April 2021, Accepted 21 June 2021, Available online 25 June 2021, Version of Record 1 July 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.115469