Novel automatic endotracheal position confirmation system: mannequin model algorithm evaluation

A novel endotracheal intubation accurate positioning confirmation system based on image classification algorithm is introduced and evaluated using a mannequin model.

The system comprises a miniature complementary metal oxide silicon sensor (CMOS) attached to the tip of a semi rigid stylet and connected to a digital signal processor (DSP) with an integrated video acquisition component. Video signals acquired and processed by an algorithm implemented on the processor. During mannequin intubations, video signals were continuously recorded. A total of 10 videos were recorded. From each video, 7 images of esophageal intubation and 8 images of endotracheal intubation (in which the carina could be clearly seen) were extracted, yielding a total of 150 images taken from arbitrary positions and angles which were processed by the confirmation algorithm.

The performance of the confirmation algorithm was evaluated using a leave-one-out method: in each iteration, 149 images were used to train the system and estimate the models, and the remaining image was used to test the system. This process was repeated 150 times such that each image participated once in testing. The system correctly identified 80 out of 80 endotracheal intubations and 70 out of 70 esophageal intubations.

This fully automatic image recognition system was used successfully to discriminate airway carina and non-carina endotracheal tube positioning. The system had a 100% success rate using a mannequin model and therefore further investigation including live tissue model and human research should follow.