Determining scanned body part from DICOM study description for relevant prior study matching.

Area of Science:

• Radiology Informatics
• Medical Imaging Analysis
• Clinical Workflow Optimization

Background:

• Radiology reporting workflow relies on reviewing prior studies.
• Picture Archiving and Communication Systems (PACS) use DICOM headers for prior study retrieval.
• Generic 'Body Part Examined' fields in DICOM headers limit accurate prior study matching.

Purpose of the Study:

• To develop a method for extracting specific body parts from DICOM Study Descriptions.
• To enhance the accuracy of prior study retrieval in radiology workflows.
• To improve filtering capabilities for relevant prior imaging studies.

Main Methods:

• A rule-based approach was developed to analyze free-text DICOM Study Descriptions.
• Algorithms were trained on a production dataset of 1200 unique study descriptions.
• The method was validated on a separate test dataset of 404 study descriptions.

Main Results:

• The proposed rule-based technique achieved 99.94% accuracy in extracting specific body parts.
• Demonstrated high precision in differentiating between generic and specific anatomical information.
• Successfully identified detailed body parts from the DICOM Study Description field.

Conclusions:

• A rule-based approach is effective for domain-specific body part extraction from DICOM headers.
• This technique can significantly improve the efficiency and accuracy of radiology reporting workflows.
• Enhanced prior study matching supports better clinical decision-making in medical imaging.