Revisiting LLM Adaptation for 3D CT Report Generation: A Study of Scaling and Diagnostic Priors

• URL: http://arxiv.org/abs/2606.17213v1
• Date: Mon, 15 Jun 2026 18:51:31 GMT
• Title: Revisiting LLM Adaptation for 3D CT Report Generation: A Study of Scaling and Diagnostic Priors
• Abstract summary: Large language models (LLMs) and vision-language models (VLMs) have demonstrated strong adaptability to natural images.<n> RAD3D-Prefix is a framework that minimizes the need for extensive parameter training.<n>Our method requires minimal trainable parameters and mitigates the risk of overfitting on small, domain-specific datasets.
• Score: 10.853232316677362
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Recent advances in multimodal learning, including large language models (LLMs) and vision-language models (VLMs), have demonstrated strong adaptability to natural images. However, extending their use to the medical domain, particularly for volumetric (3D) images, is challenging due to high computational complexity, volumetric dependencies and the semantic gap between visual features and clinical terminology. Naively fine-tuning LLMs on limited medical data often leads to overfitting and clinical hallucination, where linguistic fluency is prioritized over clinical factuality. In this study, we investigate parameter-efficient adaptation strategies for volumetric CT report generation and introduce RAD3D-Prefix, a lightweight diagnostic-prior conditioning framework that minimizes the need for extensive parameter training. This module integrates image embeddings with multi-label diagnostic classification logits, preserving critical clinical details while bridging the semantic gap. By keeping the LLM frozen, our method requires minimal trainable parameters and mitigates the risk of overfitting on small, domain-specific datasets. Through a systematic study spanning LLMs from 96.1M to 1.6B parameters, we find that fine-tuning is most beneficial for smaller LLMs, whereas freezing larger (~1B+ LLMs and training only lightweight projection layers provides a superior trade-off between performance, generalization, and computational efficiency. Across multiple automatic metrics and a clinical reader study, RAD3D-Prefix outperforms comparable parameter-efficient baselines and demonstrates strong out-of-domain generalization while using substantially fewer trainable parameters than fully fine-tuned alternatives.

Related papers

• From Many to Meaningful: Feature-Guided Zero-Shot Chronic Kidney Disease Screening Using Large Language Models [0.5442955439283729]
  This study examines the feasibility of using large language models (LLMs) for early-stage CKD screening in a zero-shot setting.<n>We propose a feature-guided framework that evaluates LLM performance using a selected set of clinically meaningful, readily available community-based features.<n>Generalizability was assessed across three heterogeneous CKD datasets spanning three countries.
  arXiv  Detail & Related papers  (2026-07-14T02:00:58Z)
• Domain-Adapted Small Language Models for Reliable Clinical Triage [4.836986736624673]
  This study evaluates whether open-source small language models (SLMs) can serve as reliable, privacy-preserving decision-support tools for clinical triage.
  arXiv  Detail & Related papers  (2026-04-29T15:00:09Z)
• Countering Catastrophic Forgetting of Large Language Models for Better Instruction Following via Weight-Space Model Merging [10.955738714664276]
  Large language models have been adopted in the medical domain for clinical documentation to reduce clinician burden.<n>Studies have reported that LLMs often "forget" a significant amount of instruction-following ability when fine-tuned using a task-specific medical dataset.<n>This study presents a model merging framework to efficiently adapt general-purpose LLMs to the medical domain.
  arXiv  Detail & Related papers  (2026-04-02T02:18:49Z)
• PREBA: Surgical Duration Prediction via PCA-Weighted Retrieval-Augmented LLMs and Bayesian Averaging Aggregation [51.96735866702332]
  PREBA is a retrieval-augmented framework that integrates PCA-weighted retrieval and Bayesian averaging aggregation.<n> PREBA significantly improves performance-for instance, reducing MAE by up to 40% and raising R2 from -0.13 to 0.62 over zero-shot inference.
  arXiv  Detail & Related papers  (2026-02-27T07:19:23Z)
• A Federated and Parameter-Efficient Framework for Large Language Model Training in Medicine [59.78991974851707]
  Large language models (LLMs) have demonstrated strong performance on medical benchmarks, including question answering and diagnosis.<n>Most medical LLMs are trained on data from a single institution, which faces limitations in generalizability and safety in heterogeneous systems.<n>We introduce the model-agnostic and parameter-efficient federated learning framework for adapting LLMs to medical applications.
  arXiv  Detail & Related papers  (2026-01-29T18:48:21Z)
• Multimodal Visual Surrogate Compression for Alzheimer's Disease Classification [69.87877580725768]
  Multimodal Visual Surrogate Compression (MVSC) learns to compress and adapt large 3D sMRI volumes into compact 2D features.<n>MVSC has two key components: a Volume Context that captures global cross-slice context under textual guidance, and an Adaptive Slice Fusion module that aggregates slice-level information in a text-enhanced, patch-wise manner.
  arXiv  Detail & Related papers  (2026-01-29T13:05:46Z)
• Mitigating Hallucinations of Large Language Models in Medical Information Extraction via Contrastive Decoding [92.32881381717594]
  We introduce ALternate Contrastive Decoding (ALCD) to solve hallucination issues in medical information extraction tasks. ALCD demonstrates significant improvements in resolving hallucination issues compared to conventional decoding methods.
  arXiv  Detail & Related papers  (2024-10-21T07:19:19Z)
• XAI4LLM. Let Machine Learning Models and LLMs Collaborate for Enhanced In-Context Learning in Healthcare [16.79952669254101]
  We introduce a knowledge-guided in-context learning framework to enable large language models to process structured clinical data.<n>Our approach integrates domain-specific feature groupings, carefully balanced few-shot examples, and task-specific prompting strategies.
  arXiv  Detail & Related papers  (2024-05-10T06:52:44Z)
• Large Language Model Distilling Medication Recommendation Model [58.94186280631342]
  We harness the powerful semantic comprehension and input-agnostic characteristics of Large Language Models (LLMs)<n>Our research aims to transform existing medication recommendation methodologies using LLMs.<n>To mitigate this, we have developed a feature-level knowledge distillation technique, which transfers the LLM's proficiency to a more compact model.
  arXiv  Detail & Related papers  (2024-02-05T08:25:22Z)
• LLM-driven Multimodal Target Volume Contouring in Radiation Oncology [46.23891509553877]
  Large language models (LLMs) can facilitate the integration of the textural information and images. We present a novel LLM-driven multimodal AI, namely LLMSeg, that is applicable to the challenging task of target volume contouring for radiation therapy. We demonstrate that the proposed model exhibits markedly improved performance compared to conventional unimodal AI models.
  arXiv  Detail & Related papers  (2023-11-03T13:38:42Z)
• To Repeat or Not To Repeat: Insights from Scaling LLM under Token-Crisis [50.31589712761807]
  Large language models (LLMs) are notoriously token-hungry during pre-training, and high-quality text data on the web is approaching its scaling limit for LLMs. We investigate the consequences of repeating pre-training data, revealing that the model is susceptible to overfitting. Second, we examine the key factors contributing to multi-epoch degradation, finding that significant factors include dataset size, model parameters, and training objectives.
  arXiv  Detail & Related papers  (2023-05-22T17:02:15Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.