Training radiomics-based CNNs for clinical outcome prediction: Challenges, strategies and findings
作者：
Highlights：
• We propose a novel AI solution with ML strategies for predicting patient outcomes.
• A new method of pre-processing radiology images is proposed for model training.
• Our models predict different clinical outcomes on pre-treatment H&N cancer data.
• A visual explanation module is leveraged to improve the interpretability of results.
• We conduct extensive experiments and comparisons, and then summarise main findings.
摘要
•We propose a novel AI solution with ML strategies for predicting patient outcomes.•A new method of pre-processing radiology images is proposed for model training.•Our models predict different clinical outcomes on pre-treatment H&N cancer data.•A visual explanation module is leveraged to improve the interpretability of results.•We conduct extensive experiments and comparisons, and then summarise main findings.
论文关键词：Cancer outcome prediction,Head & neck cancers,Deep neural networks,Radiomic features,Visualization interpretation,Radiotherapy
论文评审过程：Received 2 July 2021, Revised 20 November 2021, Accepted 30 November 2021, Available online 6 December 2021, Version of Record 8 December 2021.
论文官网地址：https://doi.org/10.1016/j.artmed.2021.102230
