Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Shi, Pengcheng, Zhang, Minghui, Song, Kehan, Liu, Jiaqi, Gu, Yun, Zhang, Xinglin
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2603.00479
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911474654380032
author Shi, Pengcheng
Zhang, Minghui
Song, Kehan
Liu, Jiaqi
Gu, Yun
Zhang, Xinglin
author_facet Shi, Pengcheng
Zhang, Minghui
Song, Kehan
Liu, Jiaqi
Gu, Yun
Zhang, Xinglin
contents Automated radiology report generation is key for reducing radiologist workload and improving diagnostic consistency, yet generating accurate reports for 3D medical imaging remains challenging. Existing vision-language models face two limitations: they do not leverage segmentation-pretrained encoders, and they inject visual features only at the input layer of language models, losing multi-scale information. We propose U-VLM, which enables hierarchical vision-language modeling in both training and architecture: (1) progressive training from segmentation to classification to report generation, and (2) multi-layer visual injection that routes U-Net encoder features to corresponding language model layers. Each training stage can leverage different datasets without unified annotations. U-VLM achieves state-of-the-art performance on CT-RATE (F1: 0.414 vs 0.258, BLEU-mean: 0.349 vs 0.305) and AbdomenAtlas 3.0 (F1: 0.624 vs 0.518 for segmentation-based detection) using only a 0.1B decoder trained from scratch, demonstrating that well-designed vision encoder pretraining outweighs the benefits of 7B+ pre-trained language models. Ablation studies show that progressive pretraining significantly improves F1, while multi-layer injection improves BLEU-mean. Code is available at https://github.com/yinghemedical/U-VLM.
format Preprint
id arxiv_https___arxiv_org_abs_2603_00479
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle U-VLM: Hierarchical Vision Language Modeling for Report Generation
Shi, Pengcheng
Zhang, Minghui
Song, Kehan
Liu, Jiaqi
Gu, Yun
Zhang, Xinglin
Computer Vision and Pattern Recognition
Automated radiology report generation is key for reducing radiologist workload and improving diagnostic consistency, yet generating accurate reports for 3D medical imaging remains challenging. Existing vision-language models face two limitations: they do not leverage segmentation-pretrained encoders, and they inject visual features only at the input layer of language models, losing multi-scale information. We propose U-VLM, which enables hierarchical vision-language modeling in both training and architecture: (1) progressive training from segmentation to classification to report generation, and (2) multi-layer visual injection that routes U-Net encoder features to corresponding language model layers. Each training stage can leverage different datasets without unified annotations. U-VLM achieves state-of-the-art performance on CT-RATE (F1: 0.414 vs 0.258, BLEU-mean: 0.349 vs 0.305) and AbdomenAtlas 3.0 (F1: 0.624 vs 0.518 for segmentation-based detection) using only a 0.1B decoder trained from scratch, demonstrating that well-designed vision encoder pretraining outweighs the benefits of 7B+ pre-trained language models. Ablation studies show that progressive pretraining significantly improves F1, while multi-layer injection improves BLEU-mean. Code is available at https://github.com/yinghemedical/U-VLM.
title U-VLM: Hierarchical Vision Language Modeling for Report Generation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.00479