Multiple instance learning for medical image classification based on instance importance
Lists
Only you see this
Abstract
Highlights•Acquiring instance-level predictions for overall bag label determination.•Introducing novel importance quantification in ImpoMIL, enhancing MIL model interpretability.•Demonstrating ImpoMIL’s capability in assigning binary importance values to frames in neonatal ultrasound videos, enhancing applicability in video analysis.•Illustrated the stages of our methods.•Included explanations on indices and weights in response to feedback.
BibTeX
@article{struski2024multiple,
author = {Lukasz Struski and Szymon Janusz and Jacek Tabor and Michal Markiewicz and Arkadiusz Lewicki},
title = {{Multiple instance learning for medical image classification based on instance importance}},
journal = {Biomedical Signal Processing and Control},
year = {2024},
doi = {10.1016/j.bspc.2023.105874},
}Loading related content...