Automated hematoma segmentation and outcome prediction for patients with traumatic brain injury

作者：

Highlights：

• A new multi-view network is developed to segment total acute hematoma on brain CT scans from patients with acute traumatic brain injury.

• A novel mixed loss function is proposed to improve the model's generalization.

• The proposed hematoma segmentation framework was validated on a heterogeneous dataset.

• Quantitative hematoma characteristics greatly improve the 6-month mortality prediction.

摘要

•A new multi-view network is developed to segment total acute hematoma on brain CT scans from patients with acute traumatic brain injury.•A novel mixed loss function is proposed to improve the model's generalization.•The proposed hematoma segmentation framework was validated on a heterogeneous dataset.•Quantitative hematoma characteristics greatly improve the 6-month mortality prediction.

论文关键词：Traumatic brain injury,Hematoma segmentation,Outcome prediction,Convolutional neural network,Deep learning

论文评审过程：Received 24 December 2019, Revised 4 May 2020, Accepted 9 June 2020, Available online 13 June 2020, Version of Record 20 June 2020.

论文官网地址：https://doi.org/10.1016/j.artmed.2020.101910