AI-Powered Navigation System for Steering POCUS in the COVID-ICU

Leveraging the Capabilities of AI: Novice Neurology-Trained Operators Performing Cardiac POCUS in Patients with Acute Brain Injury

BACKGROUND

Cardiac point-of-care ultrasound (cPOCUS) can aid in the diagnosis and treatment of cardiac disorders. Such disorders can arise as complications of acute brain injury, but most neurologic intensive care unit (NICU) providers do not receive formal training in cPOCUS. Caption artificial intelligence (AI) uses a novel deep learning (DL) algorithm to guide novice cPOCUS users in obtaining diagnostic-quality cardiac images. The primary objective of this study was to determine how often NICU providers with minimal cPOCUS experience capture quality images using DL-guided cPOCUS as well as the association between DL-guided cPOCUS and change in management and time to formal echocardiograms in the NICU.

METHODS

From September 2020 to November 2021, neurology-trained physician assistants, residents, and fellows used DL software to perform clinically indicated cPOCUS scans in an academic tertiary NICU. Certified echocardiographers evaluated each scan independently to assess the quality of images and global interpretability of left ventricular function, right ventricular function, inferior vena cava size, and presence of pericardial effusion. Descriptive statistics with exact confidence intervals were used to calculate proportions of obtained images that were of adequate quality and that changed management. Time to first adequate cardiac images (either cPOCUS or formal echocardiography) was compared using a similar population from 2018.

RESULTS

In 153 patients, 184 scans were performed for a total of 943 image views. Three certified echocardiographers deemed 63.4% of scans as interpretable for a qualitative assessment of left ventricular size and function, 52.6% of scans as interpretable for right ventricular size and function, 34.8% of scans as interpretable for inferior vena cava size and variability, and 47.2% of scans as interpretable for the presence of pericardial effusion. Thirty-seven percent of screening scans changed management, most commonly adjusting fluid goals (81.2%). Time to first adequate cardiac images decreased significantly from 3.1 to 1.7 days (p < 0.001).

CONCLUSIONS

With DL guidance, neurology providers with minimal to no cPOCUS training were often able to obtain diagnostic-quality cardiac images, which informed management changes and significantly decreased time to cardiac imaging.

Artificial intelligence (AI) versus expert: A comparison of left ventricular outflow tract velocity time integral (LVOT-VTI) assessment between ICU doctors and an AI tool

Purpose

The application of point of care ultrasound (PoCUS) in medical education is a relatively new course. There are still great differences in the existence, quantity, provision, and depth of bedside ultrasound education. The left ventricular outflow tract velocity time integral (LVOT‐VTI) has been successfully used in several studies as a parameter for hemodynamic management of critically ill patients, especially in the evaluation of fluid responsiveness. While LVOT‐VTI has been broadly used, valuable applications using artificial intelligence (AI) in PoCUS is still limited. We aimed to identify the degree of correlation between auto LVOT‐VTI and the manual LVOT‐VTI acquired by PoCUS trained ICU doctors.

Methods

Among the 58 ICU doctors who attended PoCUS training from 1 September 2019 to 30 November 2020, 46 ICU doctors who trained for more than 3 months were enrolled. At the end of PoCUS training, each of the enrolled ICU doctors acquired echocardiography parameters of a new ICU patient in 2 h after new patient was admitted. One of the two bedside expert sonographers would take standard echocardiogram of new ICU patients within 24 h. For ICU doctors, manual LVOT‐VTI was obtained for reference and auto LVOT‐VTI was calculated instantly by using an AI software tool. Based on the image quality of the auto LVOT‐VTI, ICU patients was separated into ideal group (n = 31) and average group (n = 15).

Results

Left ventricular end‐diastolic dimension (LVEDd, p = 0.1028), left ventricular ejection fraction (LVEF, p = 0.3251), left atrial dimension (LA‐d, p = 0.0962), left ventricular E/A ratio (p = 0.160), left ventricular wall motion (p = 0.317) and pericardial effusion (p = 1) had no significant difference between trained ICU doctors and expert sonographer. ICU patients in average group had greater sequential organ failure assessment (SOFA) score (7.33 ± 1.58 vs. 4.09 ± 0.57, p = 0.022) and lactic acid (3.67 ± 0.86 mmol/L vs. 1.46 ± 0.12 mmol/L, p = 0.0009) with greater value of LVEDd (51.93 ± 1.07 vs. 47.57 ± 0.89, p = 0.0053), LA‐d (39.06 ± 1.47 vs. 35.22 ± 0.98, p = 0.0334) and percentage of decreased wall motion (p = 0.0166) than ideal group. There were no significant differences of δLVOT‐VTI (|manual LVOT‐VTI – auto LVOT‐VTI|/manual VTI*100%) between the two groups (8.8% ± 1.3% vs. 10% ± 2%, p = 0.6517). Statistically, significant correlations between manual LVOT‐VTI and auto LVOT‐VTI were present in the ideal group (R² = 0.815, p = 0.00) and average group (R² = 0.741, p = 0.00).

Conclusions

ICU doctors could achieve the satisfied level of expertise as expert sonographers after 3 months of PoCUS training. Nearly two thirds of the enrolled ICU doctors could obtain the ideal view and one third of them could acquire the average view. ICU patients with higher SOFA scores and lactic acid were less likely to acquire the ideal view. Manual and auto LVOT‐VTI had statistically significant agreement in both ideal and average groups. Auto LVOT‐VTI in ideal view was more relevant with the manual LVOT‐VTI than the average view. AI might provide real‐time guidance among novice operators who lack expertise to acquire the ideal standard view.

Training strategies for point of care ultrasound in the ICU

PURPOSE OF REVIEW

Ultrasound in critical care medicine (CCUS) is a relatively young tool that has been evolving rapidly as skillsets, applications and technology continue to progress. Although ultrasound is identified as a core competency in intensive care unit (ICU) training, there remains significant variability and inconsistencies in the delivery of ultrasound training. The goal of this narrative review is to explore areas of consensus and highlight areas where consensus is lacking to bring attention to future directions of ultrasound training in critical care medicine.

RECENT FINDINGS

There exists considerable variation in competencies identified as basic for CCUS. Recent efforts by the European Society of Intensive Care Medicine serve as the most up to date iteration however implementation is still limited by regional expertise and practice patterns. Major barriers to ultrasound training in the ICU include a lack of available experts for bedside teaching and a lack of familiarity with new technology.

SUMMARY

Though international uptake of CCUS has made many gains in the past 20 years, further adoption of technology will be required to overcome the traditional barriers of CCUS training. Although the availability and time constraints of experts will remain a limitation even with wireless capabilities, the ability to expand beyond the physical constraints of an ultrasound machine will vastly benefit efforts to standardize training and improve access to knowledge.