Research

Paper

AI LLM March 18, 2026

DiffVP: Differential Visual Semantic Prompting for LLM-Based CT Report Generation

Authors

Yuhe Tian, Kun Zhang, Haoran Ma, Rui Yan, Yingtai Li, Rongsheng Wang, Shaohua Kevin Zhou

Abstract

While large language models (LLMs) have advanced CT report generation, existing methods typically encode 3D volumes holistically, failing to distinguish informative cues from redundant anatomical background. Inspired by radiological cognitive subtraction, we propose Differential Visual Prompting (DiffVP), which conditions report generation on explicit, high-level semantic scan-to-reference differences rather than solely on absolute visual features. DiffVP employs a hierarchical difference extractor to capture complementary global and local semantic discrepancies into a shared latent space, along with a difference-to-prompt generator that transforms these signals into learnable visual prefix tokens for LLM conditioning. These difference prompts serve as structured conditioning signals that implicitly suppress invariant anatomy while amplifying diagnostically relevant visual evidence, thereby facilitating accurate report generation without explicit lesion localization. On two large-scale benchmarks, DiffVP consistently outperforms prior methods, improving the average BLEU-1-4 by +10.98 and +4.36, respectively, and further boosts clinical efficacy on RadGenome-ChestCT (F1 score 0.421). All codes will be released at https://github.com/ArielTYH/DiffVP/.

Metadata

arXiv ID: 2603.17718
Provider: ARXIV
Primary Category: cs.CV
Published: 2026-03-18
Fetched: 2026-03-19 06:01

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.17718v1</id>\n    <title>DiffVP: Differential Visual Semantic Prompting for LLM-Based CT Report Generation</title>\n    <updated>2026-03-18T13:38:26Z</updated>\n    <link href='https://arxiv.org/abs/2603.17718v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.17718v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>While large language models (LLMs) have advanced CT report generation, existing methods typically encode 3D volumes holistically, failing to distinguish informative cues from redundant anatomical background. Inspired by radiological cognitive subtraction, we propose Differential Visual Prompting (DiffVP), which conditions report generation on explicit, high-level semantic scan-to-reference differences rather than solely on absolute visual features. DiffVP employs a hierarchical difference extractor to capture complementary global and local semantic discrepancies into a shared latent space, along with a difference-to-prompt generator that transforms these signals into learnable visual prefix tokens for LLM conditioning. These difference prompts serve as structured conditioning signals that implicitly suppress invariant anatomy while amplifying diagnostically relevant visual evidence, thereby facilitating accurate report generation without explicit lesion localization. On two large-scale benchmarks, DiffVP consistently outperforms prior methods, improving the average BLEU-1-4 by +10.98 and +4.36, respectively, and further boosts clinical efficacy on RadGenome-ChestCT (F1 score 0.421). All codes will be released at https://github.com/ArielTYH/DiffVP/.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CV'/>\n    <published>2026-03-18T13:38:26Z</published>\n    <arxiv:primary_category term='cs.CV'/>\n    <author>\n      <name>Yuhe Tian</name>\n    </author>\n    <author>\n      <name>Kun Zhang</name>\n    </author>\n    <author>\n      <name>Haoran Ma</name>\n    </author>\n    <author>\n      <name>Rui Yan</name>\n    </author>\n    <author>\n      <name>Yingtai Li</name>\n    </author>\n    <author>\n      <name>Rongsheng Wang</name>\n    </author>\n    <author>\n      <name>Shaohua Kevin Zhou</name>\n    </author>\n  </entry>"
}