胸部X線写真における肺結節検出のためのマルチモダルLLM,VLMおよびエージェントベースのフレームワークの比較精度研究
Daria Khovanova1, Yuriy Vasilev1,2, Anton Vladzymyrskyy1,3
1Research and Practical Clinical Center for Diagnostics and Telemedicine Technologies of the Moscow Health Care Department, Moscow, Russia.
Frontiers in digital health
|February 12, 2026
まとめ
この研究では,胸部X線写真 (CXR) で肺結節を検出するための大型言語モデル (LLM) を評価しました. MedRAXとBiomedCLIPは有望であったが,現在のLLMの精度は臨床実用には不十分である.
科学分野:
- 医療における人工知能
- 医学画像分析 医学画像分析
- 自然言語処理 (Natural Language Processing) とは,自然言語処理で処理される言語のことです.
背景:
- 人工知能 (AI) は,臨床環境にますます統合されています.
- 大型言語モデル (LLM) は,AIアシスタントのための有望なソリューションを提供します.
- 独立したパフォーマンスの評価は,臨床AIの実現可能性にとって極めて重要です.
研究 の 目的:
- 胸部X線写真 (CXR) で肺結節を検出するマルチモダルLLMの精度を評価する.
- ノードル検出のための様々な特許およびオープンソースのLLMのパフォーマンスを比較する.
主な方法:
- Llama 3.2 Vision 90B,Claude 3.5 Sonnet,Claude 3.7 Sonnet,Gemini 2.0 Pro Experimental,Perplexity,CXR-LLaVA,XrayGPT,BiomedCLIP,MedRAXなど9つのマルチモダルのLLMが評価されました.これらのLLMは,Llama 3.2 Vision 90B,Claude 3.5 Sonnet,Claude 3.7 Sonnet,Gemini 2.0 Pro Experimental,Perplexity,CXR-LLaVA,XrayGPT,BiomedCLIP,MedRAXなどで評価されました.
- モデルは100のCXR (50の肺結節を持つ) のデータセットを分析した.
- 診断精度メトリックとROC曲線が計算され,McNemarのテストが比較に使用されました.
主要な成果:
- MedRAXフレームワークとBiomedCLIPは,最も高い精度 (0.711) を達成しました.
- Claude 3.7 Sonnetは,独占的な単一モデルLLMの中で最高のパフォーマンスを発揮しました (精度0.651).
- Llama 3.2 Vision 90B,Claude 3.5 Sonnet,およびGemini 2.0 Pro Experimentalは同様の精度 (0.602) を示した.
結論:
- MedRAXとBiomedCLIPは,肺結節検出において優れたパフォーマンスを示しました.
- プロプライエタリモデルとオープンソースモデルの間に大きな違いは見つかりませんでしたので,オープンソースのLLMの精錬の可能性を示唆しています.
- 現在のLLMの精度は,臨床実施には不十分であり,データセットの制限により,結果は探索的である.
関連する概念動画
Improving Translational Accuracy
15.0K
Base complementarity between the three base pairs of mRNA codon and the tRNA anticodon is not a failsafe mechanism. Inaccuracies can range from a single mismatch to no correct base pairing at all. The free energy difference between the correct and nearly correct base pairs can be as small as 3 kcal/ mol. With complementarity being the only proofreading step, the estimated error frequency would be one wrong amino acid in every 100 amino acids incorporated. However, error frequencies observed in...
15.0K
Improving Translational Accuracy
3.7K
3.7K
Uncertainty in Measurement: Accuracy and Precision
104.7K
Scientists typically make repeated measurements of a quantity to ensure the quality of their findings and to evaluate both the precision and the accuracy of their results. Measurements are said to be precise if they yield very similar results when repeated in the same manner. A measurement is considered accurate if it yields a result that is very close to the true or the accepted value. Precise values agree with each other; accurate values agree with a true value.
104.7K
Flail Chest-I
767
Overview of Flail Chest
Flail chest is a severe and potentially life-threatening condition characterized by the fracture of three or more adjacent ribs in multiple places. It is most commonly caused by direct impacts and trauma, such as motor vehicle accidents or injuries from a steering wheel impact. It can also occur due to falls in elderly individuals with osteoporosis, or assaults involving sharp objects.
Pathophysiology
The pathophysiology of flail chest is complex, involving fractures of...
Flail chest is a severe and potentially life-threatening condition characterized by the fracture of three or more adjacent ribs in multiple places. It is most commonly caused by direct impacts and trauma, such as motor vehicle accidents or injuries from a steering wheel impact. It can also occur due to falls in elderly individuals with osteoporosis, or assaults involving sharp objects.
Pathophysiology
The pathophysiology of flail chest is complex, involving fractures of...
767
Flail Chest-II
657
Managing flail chest, a condition characterized by a segment of the chest wall moving independently from the rest of the thoracic cage, requires a comprehensive approach. It includes a thorough assessment of the patient's condition, a diagnostic evaluation to determine the extent of the injury, and the implementation of appropriate medical interventions tailored to the individual's needs.
Assessment:
1. Clinical Evaluation:
History:
Assessment:
1. Clinical Evaluation:
History:
657
Chest Physiotherapy
1.9K
Chest Physiotherapy (CPT) is a therapeutic technique used in respiratory care to improve ventilation, clear bronchial secretions, and enhance the efficiency of respiratory muscles. This therapy includes three primary procedures: postural drainage, percussion, and vibration. It can be performed on spontaneously breathing patients and those who are intubated and mechanically ventilated.
Purpose
CPT is primarily used for patients with excessive bronchial secretions who have difficulty clearing...
Purpose
CPT is primarily used for patients with excessive bronchial secretions who have difficulty clearing...
1.9K


