Automated Detection and Type Classification of Central Venous Catheters\n in Chest X-Rays

2019/07/02 by Vaishnavi Subramanian, Hongzhi Wang, Subramanian, Vaishnavi +9
Health Professions · Medicine · #Central Venous Catheters and Hemodialysis #Hemodynamic Monitoring and Therapy

paper · pdf · doi:10.48550/arxiv.1907.01656

Abstract

Central venous catheters (CVCs) are commonly used in critical care settings\nfor monitoring body functions and administering medications. They are often\ndescribed in radiology reports by referring to their presence, identity and\nplacement. In this paper, we address the problem of automatic detection of\ntheir presence and identity through automated segmentation using deep learning\nnetworks and classification based on their intersection with previously learned\nshape priors from clinician annotations of CVCs. The results not only\noutperform existing methods of catheter detection achieving 85.2% accuracy at\n91.6% precision, but also enable high precision (95.2%) classification of\ncatheter types on a large dataset of over 10,000 chest X-rays, presenting a\nrobust and practical solution to this problem.\n

Related

• Bottom-Up Instance Segmentation of Catheters for Chest X-Rays
• Automatic classification of multiple catheters in neonatal radiographs with deep learning
• Anatomically Guided Deep Learning System for Right Internal Jugular Line (RIJL) Segmentation and Tip Localization in Chest X-Ray
• Building a Benchmark Dataset and Classifiers for Sentence-Level Findings\n in AP Chest X-rays
• Abnormality Detection and Localization in Chest X-Rays using Deep\n Convolutional Neural Networks