Normal Guidance is what Attention Needs

2026-05-26

AI Summary

This work addresses the challenging task of slice-level classification in 3D medical imaging under a weakly supervised setting where only whole-scan binary labels are available. To this end, the authors propose Normal Guidance, a novel regularization method that, for the first time, incorporates a bell-shaped prior to constrain the attention distribution within a Transformer-based multiple instance learning framework. This approach explicitly encourages the model to focus on more discriminative central slices. Evaluated on three large-scale medical imaging datasets comprising over four million slices, the method significantly improves slice-level localization performance while maintaining state-of-the-art scan-level classification accuracy.

Abstract

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.

Problem

• weakly supervised learning
• slice-level classification
• 3D medical images
• multiple instance learning
• attention mechanism

Innovation

• Normal Guidance
• attention-based MIL
• weakly supervised learning
• slice-level classification
• medical image analysis