1
|
Artificial intelligence-guided image acquisition on patients with implanted electrophysiological devices: results from a pivotal prospective multi-center clinical trial. Eur Heart J 2020. [DOI: 10.1093/ehjci/ehaa946.0006] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.3] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Indexed: 11/12/2022] Open
Abstract
Abstract
Background
A novel, recently FDA-authorized software uses deep learning (DL) to provide prescriptive transthoracic echocardiography (TTE) guidance, allowing novices to acquire standard TTE views. The DL model was trained by >5,000,000 observations of the impact of probe motion on image orientation/quality. This study evaluated whether novice-acquired TTE images guided by this software were of diagnostic quality in patients with and without implanted electrophysiological (EP) devices, focusing on RV size and function, which were thought to be sensitive to EP devices. Some aspects of the study have previously been presented.
Methods
240 patients (61±16 years old, 58% male, 33% BMI >30 kg/m2, 91% with cardiac pathology) were recruited. 8 nurses without echo experience each acquired 10 view TTEs in 30 patients guided by the software. 235 of the patients were also scanned by a trained sonographer without assistance from the software. 5 Level 3 echocardiographers independently assessed the diagnostic quality of the TTEs acquired by the nurses and sonographers to evaluate the effect of EP devices on DL software performance.
Results
Nurses using the AI-guided acquisition software acquired TTEs of sufficient quality to make qualitative assessments of right ventricular (RV) size and function in greater than 80% of cases for patients with and without implanted EP devices (Table). There was no significant difference between nurse- and sonographer-acquired scans.
Conclusion
These results indicate that new DL software can guide novices to obtain TTEs that enable qualitative assessment of RV size even in the presence of implanted EP devices. The results of the comparison to sonographer-acquired exams indicate the software performance is robust to presence of pacemaker/ICD leads visible in the images (Figure).
Nurse-acquired TTE with visible ICD lead
Funding Acknowledgement
Type of funding source: Private company. Main funding source(s): Caption Health, Inc.
Collapse
|
2
|
Abstract
Abstract
Funding Acknowledgements
Bay Labs, Inc; San Francisco, CA
Background/Introduction: When used by experienced examiners, the utility of point-of-care (POC) ultrasound for assessing cardiac anatomy and function has been well established. However, in some clinical circumstances (Primary Care offices, Intensive Care Unit, some Emergency Rooms, or in remote settings) in which a rapid assessment of cardiac anatomy and dynamics can facilitate patient care, an examiner experienced at POC scanning may not be immediately available.
Purpose
To help novice users acquire clinically useful standard cardiac views using novel machine learning (ML) software.
Methods
We used an investigational device that employs ML software to provide real-time adaptive guidance of transducer position and orientation to help novice users acquire tomographic views of the heart. We tested the utility of this approach when 4 nurses with no prior training in sonography performed POC studies on 16 subjects (10 healthy, 6 with cardiac abnormalities; 9 men; body mass index normal in 6, overweight in 6, and obese in 4 subjects). Each nurse underwent didactic training and 4 hours of supervised practice using the ML program. Each nurse scanned each study subject using a scanner equipped with ML software to acquire 10 digital two-dimensional image clips, including: parasternal long axis, short axis at the aortic valve, mitral valve, and mid-left ventricle (LV), apical 2-, 4-, and 5-chamber, subcostal 4-chamber, and longitudinal views of the inferior vena cava (IVC). All video clips (n = 640) were later reviewed independently by 5 level 3-trained cardiologists who were blinded to subject, scanner, and each other"s assessments. The expert readers reviewed each set of 10 clips to determine if the following variables could be assessed qualitatively: LV size and function; right ventricular (RV) size and function; aortic, mitral and tricuspid valves; pericardial effusion; left atrial size; IVC size.
Results
The majority of expert readers concurred, independently, that the sets of images acquired by nurses using ML guidance allowed qualitative assessment of LV size and function in 98%, pericardial effusion in 98%, RV size and function in 92%, and aortic and mitral valve anatomy and dynamics in 94-97% of cases. Qualitative assessment of LA size was feasible in 95%. Images of the IVC were judged as adequate for assessment in 58%.
Conclusion
This preliminary study suggests the potential value of novel ML software by demonstrating that nurses with limited training can acquire tomographic images useful for qualitative assessment of the cardiac chambers and valves in more than 90% of the subjects examined. This approach might be useful when timely POC cardiac assessment is indicated in settings where an experienced examiner is not available. Further refinements in the guiding software are needed to improve the success rate of IVC imaging, since IVC size can be a useful indicator of volume status.
Collapse
|
3
|
Learning invariant and variant components of time-varying natural images. J Vis 2010. [DOI: 10.1167/7.9.964] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/24/2022] Open
|
4
|
Shape representation in V4: Investigating position-specific tuning for boundary conformation with the standard model of object recognition. J Vis 2010. [DOI: 10.1167/5.8.671] [Citation(s) in RCA: 2] [Impact Index Per Article: 0.1] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/24/2022] Open
|