Estimation of histopathological types from breast MRI findings using a large language model
-
By
-
Rie Kanasaki
-
Kazufumi Suzuki
-
Takami Ota
-
Sadako Akashi-Tanaka
-
Yoji Nagashima
-
Shuji Sakai
-
April 14, 2026
-
Clinical Scorecard: Utilizing a Large Language Model to Infer Histopathological Types from Breast MRI Observations
At a Glance
| Category | Detail |
|---|
| Condition | |
| Key Mechanisms | |
| Target Population | Patients aged 30-88 years with suspected malignancy undergoing contrast-enhanced breast MRI. |
| Care Setting | |
Key Highlights
Guideline-Based Recommendations
Diagnosis
Management
- Integrate LLM predictions with clinical assessments for treatment planning, ensuring a multidisciplinary approach.
Monitoring & Follow-up
Risks
Patient & Prescribing Data
Treatment decisions should rely on pathological assessments despite LLM predictions, particularly for patients aged 30-88.
Clinical Best Practices
- Ensure comprehensive reporting of MRI findings according to BI-RADS MRI lexicon.
- Maintain a multidisciplinary approach involving radiologists and pathologists.
- Regularly assess the accuracy of LLM predictions against clinical outcomes.
- Incorporate ongoing training of LLMs based on new data.
References
