针对医疗文本的特殊性，建议采用以下组合方案：
- Model Selection：强制指定
model=mineru参数，该模型内置医学术语库（含ICD-10编码） - preprocessing enhancement：对模糊的手写处方，先用PP-StructureV3进行图像增强，再提交识别
- 术语修正：在API请求中添加
specialty=medical头部字段，激活领域自适应算法 - Dual Checksum Mechanism：同时运行MonkeyOCR快速版获取初结果，用Dolphin模型二次校验关键字段
- 结果格式化：选择Markdown输出格式，系统会自动标注药品剂量、诊断结果等关键字段
This answer comes from the articleAI Fast Station: document parsing tool for comparing OCR models in one clickThe
May not be reproduced without permission:AI productivity tools " Solving the problem of missed terminology detection in OCR recognition of medical records
English 
简体中文 
日本語 
Deutsch 
Português do Brasil