Distinguishing tracheal and esophageal tissues with hyperspectral imaging and fiber-optic sensing

Spectral Reflectance as a Unique Tissue Identifier in Healthy Humans and Inhalation Injury Subjects

Tracheal intubation is the preferred method of airway management, a common emergency trauma medicine problem. Currently, methods for confirming tracheal tube placement are lacking, and we propose a novel technology, spectral reflectance, which may be incorporated into the tracheal tube for verification of placement. Previous work demonstrated a unique spectral profile in the trachea, which allowed differentiation from esophageal tissue in ex vivo swine, in vivo swine, and human cadavers. The goal of this study is to determine if spectral reflectance can differentiate between trachea and other airway tissues in living humans and whether the unique tracheal spectral profile persists in the presence of an inhalation injury. Reflectance spectra were captured using a custom fiber-optic probe from the buccal mucosa, posterior oropharynx, and trachea of healthy humans intubated for third molar extraction and from the trachea of patients admitted to a burn intensive care unit with and without inhalation injury. Using ratio comparisons, we found that the tracheal spectral profile was significantly different from buccal mucosa or posterior oropharynx, but the area under the curve values are not high enough to be used clinically. In addition, inhalation injury did not significantly alter the spectral reflectance of the trachea. Further studies are needed to determine the utility of this technology in a clinical setting and to develop an algorithm for tissue differentiation.

Deep Learning Analysis of In Vivo Hyperspectral Images for Automated Intraoperative Nerve Detection

Nerves are critical structures that may be difficult to recognize during surgery. Inadvertent nerve injuries can have catastrophic consequences for the patient and lead to life-long pain and a reduced quality of life. Hyperspectral imaging (HSI) is a non-invasive technique combining photography with spectroscopy, allowing non-invasive intraoperative biological tissue property quantification. We show, for the first time, that HSI combined with deep learning allows nerves and other tissue types to be automatically recognized in in vivo hyperspectral images. An animal model was used, and eight anesthetized pigs underwent neck midline incisions, exposing several structures (nerve, artery, vein, muscle, fat, skin). State-of-the-art machine learning models were trained to recognize these tissue types in HSI data. The best model was a convolutional neural network (CNN), achieving an overall average sensitivity of 0.91 and a specificity of 1.0, validated with leave-one-patient-out cross-validation. For the nerve, the CNN achieved an average sensitivity of 0.76 and a specificity of 0.99. In conclusion, HSI combined with a CNN model is suitable for in vivo nerve recognition.

Spectral Reflectance Can Differentiate Tracheal and Esophageal Tissue in the Presence of Bodily Fluids and Soot

Endotracheal intubation is a common life-saving procedure implemented in emergency care to ensure patient oxygenation, but it is difficult and often performed in suboptimal conditions leading to high rates of patient complications. Undetected misplacement in the esophagus is a preventable complication that can lead to fatalities in 5–10% of patients who undergo emergency intubation. End-tidal carbon dioxide monitoring and other proper placement detection methods are useful, yet the problem of misplacement persists. Our previous work demonstrated the utility of spectral reflectance sensors for differentiating esophageal and tracheal tissues, which can be used to confirm proper endotracheal tube placement. In this study, we examine the effectiveness of spectral characterization in the presence of saline, blood, “vomit”, and soot in the trachea. Our results show that spectral properties of the trachea that differentiate it from the esophagus persist in the presence of these substances. This work further confirms the potential usefulness of this novel detection technology in field applications.

Testing of novel spectral device sensor in swine model of airway obstruction

Loss of a patent airway is a significant cause of prehospital death. Endotracheal intubation is the gold standard of care but has a high rate of failure and complications, making development of new devices vital. We previously showed that tracheal tissue has a unique spectral profile which could be utilized to confirm correct airway device placement. Therefore, the goals of this study were twofold: 1- to develop an airway obstruction model and 2- use that model to assess how airway compromise affects tissue reflectance. Female swine were anesthetized, intubated, and instrumented. Pigs were allowed to breathe spontaneously and underwent either slow- or rapid-onset obstruction until a real-time pulse oximeter reading of ≤50%. At baseline, 25%, 50%, 75%, and 100% obstruction, a fiber-optic reflection probe was inserted into the trachea and esophagus to capture reflectance spectra. Both slow- and rapid-onset obstruction significantly decreased arterial oxygen concentration (sO₂ ) and increased partial pressure of CO₂ (pCO₂ ). The presence of the tracheal-defining spectral profile was confirmed and remained consistent despite changes in sO₂ and pCO₂ . This study validated a model of slow- and rapid-airway obstruction that results in significant hypoxia and hypercapnia. This is valuable for future testing of airway device components that may improve airway management. Additionally, our data support the ability of spectral reflectance to differentiate between tracheal and esophageal tissues in the presence of a clinical condition that decreases oxygen saturation.

Tissue classification of oncologic esophageal resectates based on hyperspectral data

PURPOSE

Esophageal carcinoma is the eighth most common cancer worldwide. Esophageal resection with gastric pull-up is a potentially curative therapeutic option. After this procedure, the specimen is examined by the pathologist to confirm complete removal of the cancer. An intraoperative analysis of the resectate would be less time-consuming and therefore improve patient safety.

METHODS

Hyperspectral imaging (HSI) is a relatively new modality, which has shown promising results for the detection of tumors. Automatic approaches could support the surgeon in the visualization of tumor margins. Therefore, we evaluated four supervised classification algorithms: random forest, support vector machines (SVM), multilayer perceptron, and k-nearest neighbors to differentiate malignant from healthy tissue based on HSI recordings of esophago-gastric resectates in 11 patients.

RESULTS

The best performances were obtained with a cancerous tissue detection of 63% sensitivity and 69% specificity with the SVM. In a leave-one patient-out cross-validation, the classification showed larger performance differences according to the patient data used. In less than 1 s, data classification and visualization was shown.

CONCLUSION

In this work, we successfully tested several classification algorithms for the automatic detection of esophageal carcinoma in resected tissue. A larger data set and a combination of several methods would probably increase the performance. Moreover, the implementation of software tools for intraoperative tumor boundary visualization will further support the surgeon during oncologic operations.

Using spectral reflectance to distinguish between tracheal and oesophageal tissue: applications for airway management

Proper placement of the tracheal tube requires confirmation, and the predominant method in addition to clinical signs is the presence of end-tidal carbon dioxide. Such is the importance of confirmation that novel methods may also have a place. We previously demonstrated using ex-vivo swine tissue a unique spectral reflectance characteristic of tracheal tissue that differs from oesophageal tissue. We hypothesised that this characteristic would be present in living swine tissue and human cadavers. Reflectance spectra in the range 500-650 nm were captured using a customised fibreoptic probe, compact spectrometer and white light source from both the trachea and the oesophagus in anesthetised living swine and in human cadavers. A tracheal detection algorithm using ratio comparisons of reflectance was developed. The existence of the unique tracheal characteristic in both in-vivo swine and cadaver models was confirmed (p < 0.0001 for all comparisons between tracheal and oesophageal tissue at all target wavelengths in both species). Furthermore, our proposed tracheal detection algorithm exhibited a 100% positive predictive value in both models. This has potential utility for incorporation into airway management devices.