Large language models enable multi-modality integration for brain tumor diagnosis and prognosis
KeAi Communications Co., Ltd.Peer-Reviewed Publication
For brain tumors, radiology reports provide essential imaging perspectives while pathology reports deliver microscopic confirmation, but each type of report typically requires domain experts to interpret separately. This separation can make it difficult to form a consistent basis for diagnosis and to reliably link findings to patient survival. Leveraging the integrative capabilities of large language models (LLMs), both sources can now be analyzed within a unified framework, reducing fragmentation and improving the accuracy of diagnostic classification and survival prediction.
To address this, a team led by Dr. Zhuoqi Ma (1st author) and Dr. Zhicheng Jiao (corresponding) from the Department of Radiology at Brown University and Brown University Health developed a large language model (LLM)-based pipeline that integrates radiology and pathology reports within a unified framework. By leveraging the integrative capabilities of LLMs, both sources can be analyzed together and improving the accuracy of diagnostic classification and survival prediction. Their findings demonstrate the potential of this approach to enhance diagnostic reliability and support precision neuro-oncology.
- Journal
- Meta-Radiology
- Funder
- Robert E. Leet and Clara Guthrie Patterson Trust Mentored Research Award