Semantic de-identification of burned-in PHI in DICOM medical images: a deep learning–NLP pipeline validated on clinical and phantom TMM datasets
A semantic de-identification pipeline integrating YOLOv11n-based text detection, domain-optimized EasyOCR, and a hybrid natural language processing (NLP) classification module combining regular expressions, keyword matching, and named entity recognition is proposed, confirming that the pipeline preserves quantitative p...