2024 ARRS ANNUAL MEETING - ABSTRACTS

RETURN TO ABSTRACT LISTING


E3491. Adequacy of Oxipit© Artificial Intelligence Algorithm (AI) at Identifying Endotracheal Tube Placement
Authors
  1. Mustafa Haddad; Medical College of Wisconsin
  2. Glen Hanssen; Medical College of Wisconsin
  3. Chris Bennett; Medical College of Wisconsin
  4. Zachary Pick; Medical College of Wisconsin
  5. Robin Ausman; Medical College of Wisconsin
  6. Christoper Ajala; Medical College of Wisconsin
  7. Andrew Nencka; Medical College of Wisconsin
Objective:
Oxipit is an artificial intelligence program with CE marking that has different applications for interpreting chest X-rays. One of these applications, ChestEye Quality, can identify correct and incorrect endotracheal tube placement. This study was to determine the adequacy of ChestEye Quality at evaluating endotracheal tube placement.

Materials and Methods:
A single-center retrospective review of 9095 chest x-rays obtained between June 1, 2022 and December 16, 2022 in patient's older than 18 years of age was performed using the Oxipit AI algorithm. This was performed following IRB and the safety IT protocol of our institution to ensure patient data were not breached. The chest x-rays evaluated were obtained portably and nonportably and were in the AP and PA projections. The machine identified the chest x-rays with endotracheal tubes and then divided them between those that were correctly and incorrectly placed. Of those selected by the algorithm, the images and reports were then reviewed by a board-certified radiologist for discrepancies.

Results:
A total of 9095 chest x-rays were evaluated by the machine. Of those, the machine identified 264 chest x-rays with correctly placed endotracheal tubes and 21 chest x-rays with incorrectly placed endotracheal tubes. Of the 264 chest x-rays identified as having correctly placed endotracheal tubes, 253 (96%) were in concordance with the radiologist review as having correctly placed endotracheal tubes, five (2%) had no endotracheal tubes that the machine incorrectly classified as having properly placed tubes, three (1%) had tubes at the carina with reports requesting retraction of the tube, one had a tube in the right mainstem bronchus, one had a tube 7.4 cm above the carina, and another had a tracheal stent confounded for an endotracheal tube. Of the 21 chest x-rays classified as having incorrectly placed endotracheal tubes, 20 (95%) were in concordance with the radiologist review as having incorrectly placed endotracheal tubes. One of these had an endotracheal tube that was in the right mainstem bronchus that was not mentioned in the report. The remaining chest x-ray was incorrectly classified as the endotracheal tube was in an adequate position at 3.7 cm above the carina but the machine confounded tube placement with an adjacent chest tube.

Conclusion:
The Oxipit ChestEye Quality's ability to evaluate endotracheal tube placement on chest x-rays was fairly adequate at identifying chest x-rays with correctly and incorrectly placed endotracheal tubes, however, there remains a need for oversight as around 5% were incorrectly classified by the machine.