Artificial intelligence versus expert: a comparison of rapid visual inferior vena cava collapsibility assessment between POCUS experts and a deep learning algorithm
Abstract Objectives We sought to create a deep learning algorithm to determine the degree of inferior vena cava (IVC) collapsibility in critically ill patients to enable novice point‐of‐care ultrasound (POCUS) providers. Methods We used publicly available long short term memory (LSTM) deep learning...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2020-10-01
|
| Series: | Journal of the American College of Emergency Physicians Open |
| Subjects: | |
| Online Access: | https://doi.org/10.1002/emp2.12206 |
| _version_ | 1818401343561269248 |
|---|---|
| author | Michael Blaivas Srikar Adhikari Eric A. Savitsky Laura N. Blaivas Yiju T. Liu |
| author_facet | Michael Blaivas Srikar Adhikari Eric A. Savitsky Laura N. Blaivas Yiju T. Liu |
| author_sort | Michael Blaivas |
| collection | DOAJ |
| description | Abstract Objectives We sought to create a deep learning algorithm to determine the degree of inferior vena cava (IVC) collapsibility in critically ill patients to enable novice point‐of‐care ultrasound (POCUS) providers. Methods We used publicly available long short term memory (LSTM) deep learning basic architecture that can track temporal changes and relationships in real‐time video, to create an algorithm for ultrasound video analysis. The algorithm was trained on public domain IVC ultrasound videos to improve its ability to recognize changes in varied ultrasound video. A total of 220 IVC videos were used, 10% of the data was randomly used for cross correlation during training. Data were augmented through video rotation and manipulation to multiply effective training data quantity. After training, the algorithm was tested on the 50 new IVC ultrasound video obtained from public domain sources and not part of the data set used in training or cross validation. Fleiss’ κ was calculated to compare level of agreement between the 3 POCUS experts and between deep learning algorithm and POCUS experts. Results There was very substantial agreement between the 3 POCUS experts with κ = 0.65 (95% CI = 0.49–0.81). Agreement between experts and algorithm was moderate with κ = 0.45 (95% CI = 0.33–0.56). Conclusions Our algorithm showed good agreement with POCUS experts in visually estimating degree of IVC collapsibility that has been shown in previously published studies to differentiate fluid responsive from fluid unresponsive septic shock patients. Such an algorithm could be adopted to run in real‐time on any ultrasound machine with a video output, easing the burden on novice POCUS users by limiting their task to obtaining and maintaining a sagittal proximal IVC view and allowing the artificial intelligence make real‐time determinations. |
| first_indexed | 2024-12-14T07:50:58Z |
| format | Article |
| id | doaj.art-6339f567fe2044d1957bdcee4206a2d7 |
| institution | Directory Open Access Journal |
| issn | 2688-1152 |
| language | English |
| last_indexed | 2024-12-14T07:50:58Z |
| publishDate | 2020-10-01 |
| publisher | Wiley |
| record_format | Article |
| series | Journal of the American College of Emergency Physicians Open |
| spelling | doaj.art-6339f567fe2044d1957bdcee4206a2d72022-12-21T23:10:43ZengWileyJournal of the American College of Emergency Physicians Open2688-11522020-10-011585786410.1002/emp2.12206Artificial intelligence versus expert: a comparison of rapid visual inferior vena cava collapsibility assessment between POCUS experts and a deep learning algorithmMichael Blaivas0Srikar Adhikari1Eric A. Savitsky2Laura N. Blaivas3Yiju T. Liu4Department of Emergency Medicine, St. Francis Hospital, School of Medicine University of South Carolina Columbus South Carolina USADepartment of Emergency Medicine, School of Medicine University of Arizona Tucson Arizona USADepartment of Emergency Medicine, UCLA David Geffen School of Medicine UCLA Ronald Reagan Medical Center Los Angeles California USADepartment of Emergency Medicine, Harbor‐UCLA Medical Center, David Geffren School of Medicine UCLA Los Angeles California USAMichigan State University‐East Lansing East Lansing Michigan USAAbstract Objectives We sought to create a deep learning algorithm to determine the degree of inferior vena cava (IVC) collapsibility in critically ill patients to enable novice point‐of‐care ultrasound (POCUS) providers. Methods We used publicly available long short term memory (LSTM) deep learning basic architecture that can track temporal changes and relationships in real‐time video, to create an algorithm for ultrasound video analysis. The algorithm was trained on public domain IVC ultrasound videos to improve its ability to recognize changes in varied ultrasound video. A total of 220 IVC videos were used, 10% of the data was randomly used for cross correlation during training. Data were augmented through video rotation and manipulation to multiply effective training data quantity. After training, the algorithm was tested on the 50 new IVC ultrasound video obtained from public domain sources and not part of the data set used in training or cross validation. Fleiss’ κ was calculated to compare level of agreement between the 3 POCUS experts and between deep learning algorithm and POCUS experts. Results There was very substantial agreement between the 3 POCUS experts with κ = 0.65 (95% CI = 0.49–0.81). Agreement between experts and algorithm was moderate with κ = 0.45 (95% CI = 0.33–0.56). Conclusions Our algorithm showed good agreement with POCUS experts in visually estimating degree of IVC collapsibility that has been shown in previously published studies to differentiate fluid responsive from fluid unresponsive septic shock patients. Such an algorithm could be adopted to run in real‐time on any ultrasound machine with a video output, easing the burden on novice POCUS users by limiting their task to obtaining and maintaining a sagittal proximal IVC view and allowing the artificial intelligence make real‐time determinations.https://doi.org/10.1002/emp2.12206artificial intelligencecritical caredeep learningfluid responsivenessinferior vena cavapoint‐of‐care ultrasound |
| spellingShingle | Michael Blaivas Srikar Adhikari Eric A. Savitsky Laura N. Blaivas Yiju T. Liu Artificial intelligence versus expert: a comparison of rapid visual inferior vena cava collapsibility assessment between POCUS experts and a deep learning algorithm Journal of the American College of Emergency Physicians Open artificial intelligence critical care deep learning fluid responsiveness inferior vena cava point‐of‐care ultrasound |
| title | Artificial intelligence versus expert: a comparison of rapid visual inferior vena cava collapsibility assessment between POCUS experts and a deep learning algorithm |
| title_full | Artificial intelligence versus expert: a comparison of rapid visual inferior vena cava collapsibility assessment between POCUS experts and a deep learning algorithm |
| title_fullStr | Artificial intelligence versus expert: a comparison of rapid visual inferior vena cava collapsibility assessment between POCUS experts and a deep learning algorithm |
| title_full_unstemmed | Artificial intelligence versus expert: a comparison of rapid visual inferior vena cava collapsibility assessment between POCUS experts and a deep learning algorithm |
| title_short | Artificial intelligence versus expert: a comparison of rapid visual inferior vena cava collapsibility assessment between POCUS experts and a deep learning algorithm |
| title_sort | artificial intelligence versus expert a comparison of rapid visual inferior vena cava collapsibility assessment between pocus experts and a deep learning algorithm |
| topic | artificial intelligence critical care deep learning fluid responsiveness inferior vena cava point‐of‐care ultrasound |
| url | https://doi.org/10.1002/emp2.12206 |
| work_keys_str_mv | AT michaelblaivas artificialintelligenceversusexpertacomparisonofrapidvisualinferiorvenacavacollapsibilityassessmentbetweenpocusexpertsandadeeplearningalgorithm AT srikaradhikari artificialintelligenceversusexpertacomparisonofrapidvisualinferiorvenacavacollapsibilityassessmentbetweenpocusexpertsandadeeplearningalgorithm AT ericasavitsky artificialintelligenceversusexpertacomparisonofrapidvisualinferiorvenacavacollapsibilityassessmentbetweenpocusexpertsandadeeplearningalgorithm AT lauranblaivas artificialintelligenceversusexpertacomparisonofrapidvisualinferiorvenacavacollapsibilityassessmentbetweenpocusexpertsandadeeplearningalgorithm AT yijutliu artificialintelligenceversusexpertacomparisonofrapidvisualinferiorvenacavacollapsibilityassessmentbetweenpocusexpertsandadeeplearningalgorithm |