Abstract
BACKGROUND: Colonoscopy is an essential procedure that allows screening for colorectal cancer and other lower bowel diseases. However, the procedure itself poses the risk of inducing bleeding and perforation that could go undetected if it occurs behind the distal tip camera. A soft optical sensor has been developed as an add-on device onto the colonoscope to detect bleeding. Previously, the device was successfully tested on an ex-vivo bovine colon model. Advancing towards clinical trials, here we present results of the in-vivo evaluation of our device. METHODS: The device is evaluated using two Yorkshire pigs across two testing sessions by expert endoscopists. The sensor was assessed in terms of bleeding detection accuracy, navigation time, NASA TLX, safety, and device durability. RESULTS: The soft optical blood sensor is able to achieve an accuracy of 92%, precision of 83%, sensitivity of 100%, and specificity of 87%. The average navigation time indicates that there is no significant clinical workflow interruption caused by the device. The average NASA TLX demonstrates that equipping the device does not contribute significantly to the increased workload. Lastly, the sensor remains intact and the colon has no damage observed after testing trials. CONCLUSIONS: The sensor demonstrates its capability to reliably detect bleeding in pre-clinical trials without interrupting the clinical workflow. For the future work, the sensor could be evaluated with a larger cohort of users to progress towards clinical translation.