Random Bundle: Brain Metastases Segmentation Ensembling through Annotation Randomization
This work addresses segmentation accuracy for brain metastases, a critical medical imaging task, though it appears incremental as it builds on existing ensembling strategies with specific modifications.
The paper tackles brain metastases segmentation by introducing Random Bundle (RB), an ensembling method that trains networks on datasets with 50% of annotated lesions censored and uses a lopsided bootstrap loss to recover performance, resulting in a 39% improvement in mAP and more than tripling sensitivity at 80% precision.
We introduce a novel ensembling method, Random Bundle (RB), that improves performance for brain metastases segmentation. We create our ensemble by training each network on our dataset with 50% of our annotated lesions censored out. We also apply a lopsided bootstrap loss to recover performance after inducing an in silico 50% false negative rate and make our networks more sensitive. We improve our network detection of lesions's mAP value by 39% and more than triple the sensitivity at 80% precision. We also show slight improvements in segmentation quality through DICE score. Further, RB ensembling improves performance over baseline by a larger margin than a variety of popular ensembling strategies. Finally, we show that RB ensembling is computationally efficient by comparing its performance to a single network when both systems are constrained to have the same compute.