Automated detection and type classification of central venous catheters
A system for automated detection and type classification of central venous catheters. The system includes an electronic processor that is configured to, based on an image, generate a segmentation of a potential central venous catheter using a segmentation method and extract, from the segmentation, one or more image features associated with the potential central venous catheter. The electronic processor is also configured to, based on the one or more image features, determine, using a first classifier, whether the image includes a central venous catheters and determine, using a second classifier, a type of central venous catheter included in the image.
1. A system for automated detection and type classification of central venous catheters, the system comprising:
an electronic processor, the electronic processor configured to
based on an image, generate a segmentation of a potential central venous catheter using a segmentation method, wherein the image is a chest X-ray;
extract, from the segmentation, one or more image features associated with the potential central venous catheter, wherein the one or more image features are extracted using spatial prior, wherein a spatial prior is generated from a plurality of chest X-rays including central venous catheters annotated by radiologists, wherein two spatial priors, one for central venous catheters inserted through a left-hand side of a patient's body and one for central venous catheters inserted through a right-hand side of a patient's body, are generated for each type of central venous catheter;
based on the one or more image features, determine, using a first classifier, whether the image includes a central venous catheters; and
based on the one or more image features, determine, using a second classifier, a type of central venous catheter included in the image.
2. The system according to claim 1 , wherein the electronic processor is configured to extract image features by performing a pixel-wise multiplication of the segmentation with the each determined spatial prior.
3. The system according to claim 1 , wherein the electronic processor is further configured to determine, using a first U-Net, one or more chest anatomical structures in the image.
4. The system according to claim 3 , wherein the one or more image features are extracted using Euclidean distance distributions of the segmentation relative to a center of one or more chest anatomical structures.
5. The system according to claim 1 , wherein types of central venous catheters include peripherally inserted central catheters, internal jugular catheters, subclavian catheters, and Swan-Ganz catheters.
6. The system according to claim 1 , wherein the segmentation method is a second U-Net and the first and second classifiers are random forests.
7. The system according to claim 6 , wherein the second U-Net is trained using a plurality of chest X-rays, one or more of which include a central venous catheter annotated by a radiologist.
8. A method for automated detection and type classification of central venous catheters, the method comprising:
based on an image, generating a segmentation of a potential central venous catheter using a segmentation method, wherein the image is a chest X-ray;
extracting, from the segmentation, one or more image features associated with the potential central venous catheter wherein the one or more image features are extracted using spatial priors, and wherein a spatial prior is generated from a plurality of chest X-rays including central venous catheters annotated by radiologists, wherein two spatial priors, one for central venous catheters inserted through a left-hand side of a patient's body and one for central venous catheters inserted through a right-hand side of a patient's body, are generated for each type of central venous catheter;
based on the one or more image features, determining, using a first classifier, whether the image includes a central venous catheters; and
based on the one or more image features, determining, using a second classifier, a type of central venous catheter included in the image.
9. The method according to claim 8 , wherein extracting, from the segmentation, one or more image features associated with the potential central venous catheter includes performing a pixel-wise multiplication of the segmentation with the each determined spatial prior.
10. The method according to claim 8 , the method further comprising determining, using a first U-Net, one or more chest anatomical structures in the image.
11. The method according to claim 10 , wherein the one or more image features are extracted using Euclidean distance distributions of the segmentation relative to a center of one or more chest anatomical structures.
12. The method according to claim 8 , wherein types of central venous catheters include peripherally inserted central catheters, internal jugular catheters, subclavian catheters, and Swan-Ganz catheters.
13. The method according to claim 8 , wherein the segmentation method is a second U-Net and the first and second classifiers are random forests.
14. The method according to claim 13 , wherein the second U-Net is trained using a plurality of chest X-rays, one or more of which include a central venous catheter annotated by a radiologist.