Hyperparameter selection for dataset‐constrained semantic segmentation: Practical machine learning optimization

作者：Chris Boyd, Gregory Brown, Timothy John Kleinig, Wolfgang Mayer, Joseph A. Dawson, Mark Jenkinson, Eva Bezak · 发表于：Journal of Applied Clinical Medical Physics · 年份：2024 · DOI：10.1002/acm2.14542 · 被引用次数：9 · 研究领域：Advanced X-ray and CT Imaging、Radiomics and Machine Learning in Medical Imaging、Advanced Radiotherapy Techniques

PURPOSE/AIM: This paper provides a pedagogical example for systematic machine learning optimization in small dataset image segmentation, emphasizing hyperparameter selections. A simple process is presented for medical physicists to examine hyperparameter optimization. This is also applied to a case-study, demonstrating the benefit of the method. MATERIALS AND METHODS: An unrestricted public Computed Tomography (CT) dataset, with binary organ segmentation, was used to develop a multiclass segmentation model. To start the optimization process, a preliminary manual search of hyperparameters was conducted and from there a grid search identified the most influential result metrics. A total of 658 different models were trained in 2100 h, using 13 160 effective patients. The quantity of results was analyzed using random forest regression, identifying relative hyperparameter impact. RESULTS: Metric implied segmentation quality (accuracy 96.8%, precision 95.1%) and visual inspection were found to be mismatched. In this work batch normalization was most important, but performance varied with hyperparameters and metrics selected. Targeted grid-search optimization and random forest analysis of relative hyperparameter importance, was an easily implementable sensitivity analysis approach. CONCLUSION: The proposed optimization method gives a systematic and quantitative approach to something intuitively understood, that hyperparameters change model performance. Even just grid search optimiza...