Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Automatic Report Generation for Histopathology images using pre-trained Vision Transformers and BERT

About

Deep learning for histopathology has been successfully used for disease classification, image segmentation and more. However, combining image and text modalities using current state-of-the-art (SOTA) methods has been a challenge due to the high resolution of histopathology images. Automatic report generation for histopathology images is one such challenge. In this work, we show that using an existing pre-trained Vision Transformer (ViT) to encode 4096x4096 sized patches of the Whole Slide Image (WSI) and a pre-trained Bidirectional Encoder Representations from Transformers (BERT) model for language modeling-based decoder for report generation, we can build a performant and portable report generation mechanism that takes into account the whole high resolution image. Our method allows us to not only generate and evaluate captions that describe the image, but also helps us classify the image into tissue types and the gender of the patient as well. Our best performing model achieves a 89.52% accuracy in Tissue Type classification with a BLEU-4 score of 0.12 in our caption generation task.

Saurav Sengupta, Donald E. Brown• 2023

Related benchmarks

TaskDatasetResultRank
Pathology report generationML-Path MESO (test)
BLEU20
9
Pathology report generationML-Path DLBC (test)
BLEU Score20.3
9
Pathology report generationML-Path ACC (test)
BLEU21.7
9
Pathology report generationML-Path LUNG (test)
BLEU Score24.2
9
Pathology report generationML-Path STAD (test)
BLEU Score (Mean)0.241
9
Pathology report generationML-Path KIDNEY (test)
BLEU Score (Mean)27.8
9
Pathology report generationML-Path Mean (test)
BLEU0.231
9
Pathology report generationML-Path BRCA (test)
BLEU Score18.7
9
Pathology report generationML-Path COAD (test)
BLEU Score (Mean)0.218
9
Pathology report generationML-Path LIHC (test)
BLEU Score22.1
9
Showing 10 of 12 rows

Other info

Follow for update