Gadsby, Brett
(2023)
Endotracheal tube placement with fibre optic sensing.
PhD thesis, University of Nottingham.
Abstract
Unrecognised oesophageal intubation is often described as a ‘Never Event’; an entirely preventable and extremely serious incident. However, there is still prevalence, with severe consequences for the patient. The current gold standards of visually confirming passage through the vocal cords, observing chest movement, and end-tidal capnography all have limitations. There is an opportunity for an alternative method to determine the correct endotracheal tube placement in the trachea. One such solution is described here, through the development of a compact fibre optic sensor integrated into a standard endotracheal tube. The sensor contains no electrical components inside the invasive portion of the device, and it is magnetic resonance imaging safe.
The spectral characteristics of the trachea and oesophagus are observed using a fibre optic probe, with the presence of oxyhaemoglobin being used to distinguish the tissues. A novel epoxy sensor is then designed using plastic optical fibres. A rounded top with a square base sensor shape had the least impact on the overall performance of the ETT. Furthermore, by forming the sensor with two illumination fibres, pulse oximetry can be performed. A fibre separation distance between 1.27 and 2.54 mm is optimal. However, a Monte Carlo simulation demonstrates a viable separation of 0.5 to 3 mm. ThereforeHowever, by reducing this to 0.88 mm, a more compact sensor can be produced whilst still retaining a classification rate (average correct identification of trachea and oesophagus) of >89.2% in ex-vivo models.
Two methods for emitting light perpendicularly to the fibre axis are explored, demonstrating that bending the fibres caused an optical power loss of 29.0%, whereas cleaving the fibres at 45° produced an 81.9% loss. The position of the sensor on the endotracheal tube is investigated, finding that integration behind the cuff is preferable. However, placement outside the main lumen of the ETT is also a viable option.
Computational methods to process spectral measurements are developed. Data for these methods was provided by two experiments on two different sensor types, providing a high tissue classification rate for both the trachea and oesophagus. The first experiment consisted of 9 ex-vivo porcine samples, producing a correct tissue identification of up to 100.0%. The second consisted of 10 sensors on 1 ex-vivo porcine sample, yielding a maximum correct tissue identification of 89.2% when a support vector machine classifier was used. Application of the sensor in an animal study, which consisted of 3 porcine subjects, generated a maximum correct tissue identification of 98.31.6% using principal component analysis and aa support vector machine. The data were recorded over a combined time of 348 minutes, obtained during varying cuff pressures, endotracheal tube orientations, and movement.
Finally, suggested future developments to the sensor design and computational methods demonstrate a potential route to improving tissue identification rates. Changes to the experimental protocol are described to verify classification rates. Concepts for exchanging the spectrometer for photodiodes and optical filters to reduce the cost of the opto-electronic units display potential. The technology has applications in wider healthcare, with examples of integration within naso-/oro- gastric tubes and other invasive medical devices given.
Actions (Archive Staff Only)
 |
Edit View |
Tools
Tools
