Saved in:
Bibliographic Details
Main Authors: Harvey, Ethan, Loevlie, Dennis Johan, Hughes, Michael C.
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.27306
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913164501712896
author Harvey, Ethan
Loevlie, Dennis Johan
Hughes, Michael C.
author_facet Harvey, Ethan
Loevlie, Dennis Johan
Hughes, Michael C.
contents We consider training classifiers for 3D medical images using only one binary label for the entire volume rather than a label for each 2D slice. In such weakly supervised settings, can we learn accurate classifiers for slice-level predictions? Attention-based multiple instance learning (MIL) can produce an attention score for every slice. Yet recent work demonstrates that a simple center-focused baseline that ignores image content can outperform attention-based and transformer-based MIL at slice-level classification of 3D brain scans. We show this baseline also outperforms existing MIL at slice-level classification of thoracic and abdominal CT scans. Motivated by this baseline, we propose Normal Guidance, a regularization technique that encourages the learned attention distribution to follow a bell-shaped curve. Across three medical imaging datasets totaling over 4 million 2D slices, we show our Normal Guidance enables attention-based and transformer-based MIL methods to deliver significantly better slice-level localization than the state-of-the-art while remaining competitive at whole-scan classification.
format Preprint
id arxiv_https___arxiv_org_abs_2605_27306
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Normal Guidance is what Attention Needs
Harvey, Ethan
Loevlie, Dennis Johan
Hughes, Michael C.
Machine Learning
We consider training classifiers for 3D medical images using only one binary label for the entire volume rather than a label for each 2D slice. In such weakly supervised settings, can we learn accurate classifiers for slice-level predictions? Attention-based multiple instance learning (MIL) can produce an attention score for every slice. Yet recent work demonstrates that a simple center-focused baseline that ignores image content can outperform attention-based and transformer-based MIL at slice-level classification of 3D brain scans. We show this baseline also outperforms existing MIL at slice-level classification of thoracic and abdominal CT scans. Motivated by this baseline, we propose Normal Guidance, a regularization technique that encourages the learned attention distribution to follow a bell-shaped curve. Across three medical imaging datasets totaling over 4 million 2D slices, we show our Normal Guidance enables attention-based and transformer-based MIL methods to deliver significantly better slice-level localization than the state-of-the-art while remaining competitive at whole-scan classification.
title Normal Guidance is what Attention Needs
topic Machine Learning
url https://arxiv.org/abs/2605.27306