Comprehensive language-image pre-training for 3D medical image understanding

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wald, Tassilo, Hamamci, Ibrahim Ethem, Gao, Yuan, Bond-Taylor, Sam, Sharma, Harshita, Ilse, Maximilian, Lo, Cynthia, Melnichenko, Olesya, Schwaighofer, Anton, Codella, Noel C. F., Wetscherek, Maria Teodora, Maier-Hein, Klaus H., Korfiatis, Panagiotis, Salvatelli, Valentina, Alvarez-Valle, Javier, Pérez-García, Fernando
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915729146642432
author Wald, Tassilo
Hamamci, Ibrahim Ethem
Gao, Yuan
Bond-Taylor, Sam
Sharma, Harshita
Ilse, Maximilian
Lo, Cynthia
Melnichenko, Olesya
Schwaighofer, Anton
Codella, Noel C. F.
Wetscherek, Maria Teodora
Maier-Hein, Klaus H.
Korfiatis, Panagiotis
Salvatelli, Valentina
Alvarez-Valle, Javier
Pérez-García, Fernando
author_facet Wald, Tassilo
Hamamci, Ibrahim Ethem
Gao, Yuan
Bond-Taylor, Sam
Sharma, Harshita
Ilse, Maximilian
Lo, Cynthia
Melnichenko, Olesya
Schwaighofer, Anton
Codella, Noel C. F.
Wetscherek, Maria Teodora
Maier-Hein, Klaus H.
Korfiatis, Panagiotis
Salvatelli, Valentina
Alvarez-Valle, Javier
Pérez-García, Fernando
contents Vision-language pre-training, i.e., aligning images with paired text, is a powerful paradigm to create encoders that can be directly used for tasks such as classification, retrieval, and segmentation. In the 3D medical image domain, these capabilities allow vision-language encoders (VLEs) to support radiologists by retrieving patients with similar abnormalities, predicting likelihoods of abnormality, or, with downstream adaptation, generating radiological reports. While the methodology holds promise, data availability and domain-specific hurdles limit the capabilities of current 3D VLEs. In this paper, we overcome these challenges by injecting additional supervision via a report generation objective and combining vision-language with vision-only pre-training. This allows us to leverage both image-only and paired image-text 3D datasets, increasing the total amount of data to which our model is exposed. Through these additional objectives, paired with best practices of the 3D medical imaging domain, we develop the Comprehensive Language-Image Pre-training (COLIPRI) encoder family. Our COLIPRI encoders achieve state-of-the-art performance in report generation, semantic segmentation, classification probing, and zero-shot classification. The model is available at https://huggingface.co/microsoft/colipri.
format Preprint
id arxiv_https___arxiv_org_abs_2510_15042
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Comprehensive language-image pre-training for 3D medical image understanding
Wald, Tassilo
Hamamci, Ibrahim Ethem
Gao, Yuan
Bond-Taylor, Sam
Sharma, Harshita
Ilse, Maximilian
Lo, Cynthia
Melnichenko, Olesya
Schwaighofer, Anton
Codella, Noel C. F.
Wetscherek, Maria Teodora
Maier-Hein, Klaus H.
Korfiatis, Panagiotis
Salvatelli, Valentina
Alvarez-Valle, Javier
Pérez-García, Fernando
Computer Vision and Pattern Recognition
Machine Learning
Vision-language pre-training, i.e., aligning images with paired text, is a powerful paradigm to create encoders that can be directly used for tasks such as classification, retrieval, and segmentation. In the 3D medical image domain, these capabilities allow vision-language encoders (VLEs) to support radiologists by retrieving patients with similar abnormalities, predicting likelihoods of abnormality, or, with downstream adaptation, generating radiological reports. While the methodology holds promise, data availability and domain-specific hurdles limit the capabilities of current 3D VLEs. In this paper, we overcome these challenges by injecting additional supervision via a report generation objective and combining vision-language with vision-only pre-training. This allows us to leverage both image-only and paired image-text 3D datasets, increasing the total amount of data to which our model is exposed. Through these additional objectives, paired with best practices of the 3D medical imaging domain, we develop the Comprehensive Language-Image Pre-training (COLIPRI) encoder family. Our COLIPRI encoders achieve state-of-the-art performance in report generation, semantic segmentation, classification probing, and zero-shot classification. The model is available at https://huggingface.co/microsoft/colipri.
title Comprehensive language-image pre-training for 3D medical image understanding
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2510.15042