Lokalne modele sztucznej inteligencji do automatycznej analizy badań MRI mózgu u pacjentów z epilepsją: badanie retrospektywne

PubMed➕ 19.11.2026J Imaging Inform Med

On-Premises Open-Weight Large Language Models for Classifying Epilepsy Brain MRI Reports: A Retrospective Study

W skrócie

Badacze sprawdzili, czy zainstalowane na komputerach szpitala programy AI potrafiają automatycznie analizować pisemne opisy badań MRI mózgu u pacjentów z epilepsją, bez wysyłania danych do internetu. Okazało się, że program AI działający lokalnie radzit sobie równie dobrze jak doświadczeni lekarze, zwłaszcza gdy podano mu jasne zasady definiujące rodzaje zmian w mózgu. Wynik ten pokazuje, że AI może w przyszłości pomóc lekarzom w szybkiej selekcji pacjentów do badań naukowych, zachowując bezpieczeństwo ich danych medycznych.

Oryginalny abstract (angielski)

Free-text reports limit the use of imaging data for research cohort screening. This study evaluated whether locally deployed open-weight large language models (LLMs) can classify brain magnetic resonance imaging (MRI) reports in epilepsy without external data transfer, and which forms of domain knowledge improve classification. In this single-center retrospective study, 1210 epilepsy-protocol brain MRI reports were labeled by a board-certified neuroradiologist for abnormality status and primary imaging finding (eight classes). A held-out test set of 206 reports was classified by a biomedical (OpenBioLLM-8B) and a general-purpose (Qwen3.5-9B) LLM on a single 24-GB graphics processing unit, with and without a rule set and example retrieval, and compared with a fine-tuned radiology-specific BERT model, an independent neuroradiology fellow, and a same-base control model. The primary outcome was the macro-averaged F1 score for the primary finding. Without task-specific training, the general-purpose model achieved a finding macro-F1 of 0.63 (95% confidence interval (CI), 0.53-0.70), comparable to the fine-tuned BERT model (0.59; difference, - 0.04; 95% CI, - 0.13 to 0.07) and to inter-reader agreement (0.61; κ = 0.59). A rule set stating institutional category definitions improved the biomedical model (difference, 0.23; 95% CI, 0.11-0.33) without significantly reducing performance in any model, whereas biomedical fine-tuning conferred no advantage over the control model and example retrieval did not help. Residual errors concentrated where the readers disagreed. Locally deployed LLMs performed comparably to inter-reader agreement, which may support human-verified cohort screening; in this institution-specific task, explicit category definitions were more useful than biomedical specialization.

Metadane publikacji

Journal
J Imaging Inform Med
Data publikacji
18.11.2026
PMID
43052637
DOI
10.1007/s10278-026-02533-7
Autorzy
Kang SM, Kim S, Choi Y, Koo YS
Słowa kluczowe
Artificial intelligence, Epilepsy, Magnetic resonance imaging, Natural language processing, Radiology reports
Źródło
PubMed