Standard endoscopy is a screening method that allows the detection of GI tract pathologies and lesions. However, it was demonstrated that endoscopy suffers from a significant miss rate for cancer detection. Therefore, there is a need to establish a standard protocol that can quantify the quality of endoscopic procedures, assist physicians in analyzing their performance and train them on how to select navigation techniques that result in different levels of quality inspection. In this paper, we propose a method to evaluate the performance of endoscopists by analyzing the trajectory produced by the endoscopic camera. The proposed method consists of three main modules: i) the image analysis module for estimating camera positions; ii) Kalman filter for estimating velocity and acceleration profiles; and iii) analysis module to analyze the trajectory based on the density of observed images and the acceleration profile. The approach was preliminary tested with a dataset of images collected from an ex-vivo experiment, with porcine tissue.
Trajectory analysis of endoscopic capsule images: A feasibility study
Mura M.;Ciuti G.;Dario P.;
2016-01-01
Abstract
Standard endoscopy is a screening method that allows the detection of GI tract pathologies and lesions. However, it was demonstrated that endoscopy suffers from a significant miss rate for cancer detection. Therefore, there is a need to establish a standard protocol that can quantify the quality of endoscopic procedures, assist physicians in analyzing their performance and train them on how to select navigation techniques that result in different levels of quality inspection. In this paper, we propose a method to evaluate the performance of endoscopists by analyzing the trajectory produced by the endoscopic camera. The proposed method consists of three main modules: i) the image analysis module for estimating camera positions; ii) Kalman filter for estimating velocity and acceleration profiles; and iii) analysis module to analyze the trajectory based on the density of observed images and the acceleration profile. The approach was preliminary tested with a dataset of images collected from an ex-vivo experiment, with porcine tissue.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.