CLiP, catheter and line position dataset

Area of Science:

• Medical Imaging
• Radiology
• Artificial Intelligence in Healthcare

Background:

• Accurate placement of medical devices like endotracheal tubes (ETT), nasoenteric tubes (NET), and central venous catheters (CVCs) is critical for patient safety and treatment efficacy.
• Complications arising from malpositioned devices can lead to significant patient harm and increased healthcare costs.
• Existing datasets may lack the detailed annotations required for robust training of AI models for device position verification.

Purpose of the Study:

• To present a comprehensive, manually annotated dataset of ETT, NET, and CVCs from chest radiographs.
• To facilitate the development and validation of AI algorithms for automated detection and assessment of catheter placement.
• To improve the accuracy and reliability of catheter position assessment in clinical practice.

Main Methods:

• Utilized the publicly available NIH ChestXRay14 dataset, comprising 30,083 chest radiographs.
• Performed manual annotation and segmentation of 17,999 instances of endotracheal tubes (ETT), nasoenteric tubes (NET), and central venous catheters (CVCs).
• The dataset includes 50,612 image-level annotations, providing extensive data for machine learning model training.

Main Results:

• A large-scale, meticulously annotated dataset for three common types of catheters in chest radiographs has been created.
• The dataset enables detailed analysis and model development for catheter localization and positional assessment.
• The availability of this dataset is expected to advance research in AI-driven medical image analysis for critical care.

Conclusions:

• The described dataset represents a valuable resource for advancing research in medical device placement verification using artificial intelligence.
• This resource will aid in the development of automated systems to ensure correct catheter positioning, thereby reducing potential complications.
• The dataset has the potential to significantly impact patient safety and diagnostic accuracy in radiology and critical care settings.