TransMIL: Transformer based Correlated Multiple Instance Learning for Whole Slide Image Classication

By Zhucheng Shao, Hao Bian, Yang Chen, Yifeng Wang, Jian Zhang, Xiangyang Ji, Yongbing Zhang. Neural Information Processing Systems, 2021

1,314 citations recorded by the source corpus.

Multiple instance learning (MIL) is a powerful tool to solve the weakly supervised classification in whole slide image (WSI) based pathology diagnosis. However, the current MIL methods are usually based on independent and identical distribution hypothesis, thus neglect the correlation among different instances. To address this problem, we proposed a new framework, called correlated MIL, and provided a proof for convergence. Based on this framework, we devised a Transformer based MIL (TransMIL), which explored both morphological and spatial information. The proposed TransMIL can effectively deal with unbalanced/balanced and binary/multiple classification with great visualization and interpretability. We conducted various experiments for three different computational pathology problems and achieved better performance and faster convergence compared with state-of-the-art methods. The test AUC for the binary tumor classification can be up to 93.09% over CAMELYON16 dataset. And the AUC over the cancer subtypes classification can be up to 96.03% and 98.82% over TCGA-NSCLC dataset and TCGA-RCC dataset, respectively. Implementation is available at: https://github.com/szc19990412/TransMIL.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/2153963d19b21e64e55295b69de81d99aadac465

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Scaling Vision Transformers to Gigapixel Images via Hierarchical Self-Supervised Learning (2022, background): “Recent variations of MIL have also evolved to extend the aggregation layers and scoring functions [17, 64, 68, 75, 77, 78, 85].”
Artificial intelligence for digital and computational pathology (2023, background): “sensitivity from WSIs, allowing the identification of patients who do not require further testing 133 .”
DTFD-MIL: Double-Tier Feature Distillation Multiple Instance Learning for Histopathology Whole Slide Image Classification (2022, methodology): “…scores are learned by a side-branch network, in DS-MIL [18] an attention score is based on the cosine distance between features of an instance and the critical instance, while in Trans-MIL [34], they are the output of a transformer architecture that encode the mutual-correlations between instances.”
Transformers in Medical Image Analysis: A Review (2022, background): “Shao et al.[150] focused on the correlation between different instances as opposed to simply assuming that instances are independent and identically distributed.”
Transformers in Medical Imaging: A Survey (2022, methodology): “Further, the attention scores from the TransMIL have been visualized to demonstrate interpretability, as shown in Fig.”