Large-Vocabulary Segmentation for Medical Images with Text Prompts

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhao, Ziheng, Zhang, Yao, Wu, Chaoyi, Zhang, Xiaoman, Zhou, Xiao, Zhang, Ya, Wang, Yanfeng, Xie, Weidi
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916849253351424
author Zhao, Ziheng
Zhang, Yao
Wu, Chaoyi
Zhang, Xiaoman
Zhou, Xiao
Zhang, Ya
Wang, Yanfeng
Xie, Weidi
author_facet Zhao, Ziheng
Zhang, Yao
Wu, Chaoyi
Zhang, Xiaoman
Zhou, Xiao
Zhang, Ya
Wang, Yanfeng
Xie, Weidi
contents This paper aims to build a model that can Segment Anything in 3D medical images, driven by medical terminologies as Text prompts, termed as SAT. Our main contributions are three-fold: (i) We construct the first multimodal knowledge tree on human anatomy, including 6502 anatomical terminologies; Then, we build the largest and most comprehensive segmentation dataset for training, collecting over 22K 3D scans from 72 datasets, across 497 classes, with careful standardization on both image and label space; (ii) We propose to inject medical knowledge into a text encoder via contrastive learning and formulate a large-vocabulary segmentation model that can be prompted by medical terminologies in text form; (iii) We train SAT-Nano (110M parameters) and SAT-Pro (447M parameters). SAT-Pro achieves comparable performance to 72 nnU-Nets -- the strongest specialist models trained on each dataset (over 2.2B parameters combined) -- over 497 categories. Compared with the interactive approach MedSAM, SAT-Pro consistently outperforms across all 7 human body regions with +7.1% average Dice Similarity Coefficient (DSC) improvement, while showing enhanced scalability and robustness. On 2 external (cross-center) datasets, SAT-Pro achieves higher performance than all baselines (+3.7% average DSC), demonstrating superior generalization ability.
format Preprint
id arxiv_https___arxiv_org_abs_2312_17183
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Large-Vocabulary Segmentation for Medical Images with Text Prompts
Zhao, Ziheng
Zhang, Yao
Wu, Chaoyi
Zhang, Xiaoman
Zhou, Xiao
Zhang, Ya
Wang, Yanfeng
Xie, Weidi
Image and Video Processing
Computer Vision and Pattern Recognition
This paper aims to build a model that can Segment Anything in 3D medical images, driven by medical terminologies as Text prompts, termed as SAT. Our main contributions are three-fold: (i) We construct the first multimodal knowledge tree on human anatomy, including 6502 anatomical terminologies; Then, we build the largest and most comprehensive segmentation dataset for training, collecting over 22K 3D scans from 72 datasets, across 497 classes, with careful standardization on both image and label space; (ii) We propose to inject medical knowledge into a text encoder via contrastive learning and formulate a large-vocabulary segmentation model that can be prompted by medical terminologies in text form; (iii) We train SAT-Nano (110M parameters) and SAT-Pro (447M parameters). SAT-Pro achieves comparable performance to 72 nnU-Nets -- the strongest specialist models trained on each dataset (over 2.2B parameters combined) -- over 497 categories. Compared with the interactive approach MedSAM, SAT-Pro consistently outperforms across all 7 human body regions with +7.1% average Dice Similarity Coefficient (DSC) improvement, while showing enhanced scalability and robustness. On 2 external (cross-center) datasets, SAT-Pro achieves higher performance than all baselines (+3.7% average DSC), demonstrating superior generalization ability.
title Large-Vocabulary Segmentation for Medical Images with Text Prompts
topic Image and Video Processing
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2312.17183