# Attention-based multiple instance learning for predominant growth pattern prediction in lung adenocarcinoma wsi using foundation models

arXiv: https://arxiv.org/abs/2604.21530 · HTML: https://mlsanity.com/paper/2604.21530/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 5 of its 11 references that are in the map.
- First posted on arXiv 23 Apr 2026.

## Builds on (most cited first, 5 of 5)

- [Data Efficient and Weakly Supervised Computational Pathology on Whole Slide Images](https://mlsanity.com/paper/2004.09666/) (2020)
- [Albumentations: fast and flexible image augmentations](https://mlsanity.com/paper/1809.06839/) (2018)
- [Visual Transformers: Token-based Image Representation and Processing for Computer Vision](https://mlsanity.com/paper/2006.03677/) (2020)
- [Pathologist-level classification of histologic patterns on resected lung adenocarcinoma slides with deep neural networks](https://mlsanity.com/paper/1901.11489/) (2019)
- [CellOMaps: A Compact Representation for Robust Classification of Lung Adenocarcinoma Growth Patterns](https://mlsanity.com/paper/2501.08094/) (2025)

## Abstract (arXiv)

Lung adenocarcinoma (LUAD) grading depends on accurately identifying growth patterns, which are indicators of prognosis and can influence treatment decisions. Common deep learning approaches to determine the predominant pattern rely on patch-level classification or segmentation, requiring extensive annotations. This study proposes an attention-based multiple instance learning (ABMIL) framework to predict the predominant LUAD growth pattern at the whole slide level to reduce annotation burden. Our approach integrates pretrained pathology foundation models as patch encoders, used either frozen or fine-tuned on annotated patches, to extract discriminative features that are aggregated through attention mechanisms. Experiments show that fine-tuned encoders improve performance, with Prov-GigaPath achieving the highest agreement (kappa = 0.699) under ABMIL. Compared to simple patch-aggregation baselines, ABMIL yields more robust predictions by leveraging slide-level supervision and spatial attention. Future work will extend this framework to estimate the full distribution of growth patterns and validate performance on external cohorts.