大規模な言語モデルで,放射線学のヨーロッパ・ディプロマを解決する:比較評価
Hamza Eren Güzel1, Laura Oleaga2, Ali Murat Koç3
1Department of Radiology, Ministry of Health Izmir City Hospital, İzmir, Turkey (H.E.G.).
Academic radiology
|February 14, 2026
まとめ
2つの高度なAIモデルであるGPT-5とGemini 2.5 Proは,欧州放射線学学士号 (EDiR) 試験に合格し,放射線学教育の可能性を示しています. しかし,完全な臨床展開のために視覚的および時間的な理解に課題が残っています.
科学分野:
- 医療イメージングにおける人工知能
- 放射線学 教育 テクノロジー テクノロジー
- 大型言語モデル (LLM) とは
背景:
- マルチモダルの大型言語モデル (LLM) は急速に進歩しており,放射線学の教育と評価において大きな可能性を秘めている.
- 標準化された医療検査でこれらのLLMのパフォーマンスを評価することは,その能力と限界を理解するために不可欠です.
研究 の 目的:
- 3つの最先端のLLM:GPT-5,Gemini 2.5 Pro,およびClaude 4.5 Sonnetのパフォーマンスを比較する.
- 引退した欧州放射線学学士号 (EDiR) 試験を使用して,彼らの能力を評価する.
主な方法:
- EDiRの試験は,複数の応答質問,ショートケース,COREケースを含むもので,LLMにゼロショット条件で実施されました.
- 入力はテキスト,静的画像,ビデオで構成され,回答は公式の欧州放射線学評議会の基準に基づいて得点されました.
- 通過する値は,過去の人間のパフォーマンスデータに基づいています.
主要な成果:
- Gemini 2.5 Pro (72.6%) とGPT-5 (67.3%) は書面コンポーネントパスマーク (50.9%) を上回ったが,Claude 4.5 Sonnetは47.5%のスコアを獲得した.
- GPT-5 (62.5%) とGemini 2.5 Pro (56.3%) は,クロード 4.5 ソネット (50.7%) と異なり,COREコンポーネント (55%のパスマーク) を通過しました.
- モデルはテキストと静的な画像のタスクに優れていたが,特定の座標のローカライゼーションとダイナミックなビデオ解釈に苦労した.
結論:
- GPT-5とGemini 2.5 Proは,EDiRの合格基準を満たしており,放射線学教育の拡張に適していることを示しています.
- セマンティック処理の進歩にもかかわらず,空間的局所化と時間的推論の限界は依然として存在し,自律的な臨床展開を妨げています.
- 臨床的視覚的アグラウンドは,放射線学の現在のLLMにとって重要な課題です.
関連する概念動画
Language
924
Language is a unique communication system that uses words and systematic rules to organize and transmit information. Unlike other forms of communication, which may involve postures, movements, odors, or vocalizations, language relies on symbols and grammar. This makes human communication distinct from that of other species, who also communicate but do not use language in the same way humans do.
Corballis and Suddendorf (2007) and Tomasello and Rakoczy (2003) highlight the role of language in...
Corballis and Suddendorf (2007) and Tomasello and Rakoczy (2003) highlight the role of language in...
924
Mathematical Modeling: Problem Solving
404
Mathematical modeling transforms real-world scenarios into mathematical expressions, allowing for structured problem-solving and analysis. This process involves defining the situation, assigning variables to measurable quantities, selecting an appropriate model, and solving the resulting equation. Such models are invaluable in finance, providing precise methods to evaluate investments, loans, and repayment structures.A widely used example is the calculation of fixed monthly payments on a loan,...
404
Components of Language
836
Language, whether spoken, signed, or written, consists of specific components: lexicon and grammar. The lexicon is the vocabulary of a language, comprising its words. Grammar is the set of rules used to convey meaning through the lexicon. For example, English grammar adds “-ed” to most verbs to indicate past tense. Words are formed by combining phonemes, which are the basic sound units of a language. Different languages have different sets of phonemes (e.g., “ah” vs.
836
Language Development
956
Children master language quickly and with relative ease, supported by both biological predisposition and reinforcement. B. F. Skinner (1957) proposed that language is learned through reinforcement, while Noam Chomsky (1965) argued that language acquisition mechanisms are biologically determined.
The critical period for language acquisition suggests that the ability to acquire language is at its peak early in life. As people age, this proficiency decreases. Language development begins very...
The critical period for language acquisition suggests that the ability to acquire language is at its peak early in life. As people age, this proficiency decreases. Language development begins very...
956
Language and Cognition
820
Language serves as a bridge between ideas and communication, influencing how individuals perceive and interact with the world. Psychologists have long debated whether language shapes thought or vice versa. This discussion gained grip with Edward Sapir and Benjamin Lee Whorf in the 1940s, who proposed that language determines thought, a concept known as linguistic determinism. They suggested that the vocabulary and structure of a language influence how its speakers think and perceive reality.
820
Growth Models with Integration: Problem Solving
65
In population modeling, integration provides a systematic way to determine accumulated quantities from known rates of change. One such application arises in ecology, where the total weight of a fish population in a body of water is referred to as its biomass. When the rate of growth of this biomass is known as a function of time, calculus can be used to determine the total biomass at a future date.Growth Rate and Biomass FunctionLet the growth rate of the fish population be represented by a...
65


