Multi-Scale Attention-based Multiple Instance Learning for Classification of Multi-Gigapixel Histology Images
This work addresses cancer diagnosis and prognosis for medical professionals by providing a method to classify histology images with slide-level labels, but it is incremental as it applies existing techniques like multiple instance learning and attention to a new specific domain.
The paper tackled the problem of predicting latent membrane protein 1 (LMP1) status in nasopharyngeal carcinoma from multi-gigapixel histology images using a deep learning pipeline with multiple instance learning and attention mechanisms, achieving an average accuracy of 0.936, AUC of 0.995, and F1-score of 0.862 in cross-validation.
Histology images with multi-gigapixel of resolution yield rich information for cancer diagnosis and prognosis. Most of the time, only slide-level label is available because pixel-wise annotation is labour intensive task. In this paper, we propose a deep learning pipeline for classification in histology images. Using multiple instance learning, we attempt to predict the latent membrane protein 1 (LMP1) status of nasopharyngeal carcinoma (NPC) based on haematoxylin and eosin-stain (H&E) histology images. We utilised attention mechanism with residual connection for our aggregation layers. In our 3-fold cross-validation experiment, we achieved average accuracy, AUC and F1-score 0.936, 0.995 and 0.862, respectively. This method also allows us to examine the model interpretability by visualising attention scores. To the best of our knowledge, this is the first attempt to predict LMP1 status on NPC using deep learning.