Integrated Relaxation Pressure Classification and Probe Positioning Failure Detection in High-Resolution Esophageal Manometry Using Machine Learning

Abstract

High-resolution esophageal manometry is used for the study of esophageal motility disorders, with the help of catheters with up to 36 sensors. Color pressure topography plots are generated and analyzed and using the Chicago algorithm a final diagnosis is established. One of the main parameters in this algorithm is integrated relaxation pressure (IRP). The procedure is time consuming. Our aim was to firstly develop a machine learning based solution to detect probe positioning failure and to create a classifier to automatically determine whether the IRP is in the normal range or higher than the cut-off, based solely on the raw images. The first step was the preprocessing of the images, by finding the region of interest—the exact moment of swallowing. Afterwards, the images were resized and rescaled, so they could be used as input for deep learning models. We used the InceptionV3 deep learning model to classify the images as correct or failure in catheter positioning and to determine the exact class of the IRP. The accuracy of the trained convolutional neural networks was above 90% for both problems. This work is just the first step in fully automating the Chicago Classification, reducing human intervention.

Automated Chicago Classification for Esophageal Motility Disorder Diagnosis Using Machine Learning

“…The IRP was measured during the first ten seconds after the commencement of the swallow, which was regarded as the white vertical line. More information about the dataset and the IRP classification algorithm can be found in our previous article [ 8 ].…”

“…We mention that in our previously published study we developed an automatic classifier to assess whether the IRP is in the normal range or greater than the cut-off, and to detect the probe placement failure, based simply on the raw pictures [ 8 ]. The previous study was the first step in automating the Chicago classification process based on Machine Learning that follows the same steps as a human expert.…”

“…The previous study was the first step in automating the Chicago classification process based on Machine Learning that follows the same steps as a human expert. The photos were initially pre-processed by locating the region of interest—the precise moment of swallowing [ 8 ]. Further, the photos were scaled and resized such that deep learning models could utilize them as input.…”

“…Further, the photos were scaled and resized such that deep learning models could utilize them as input. To categorize the photos as successful or unsuccessful catheter positioning and to establish the precise class of the IRP, we employed the InceptionV3 deep learning model [ 8 ]. For both challenges, the trained CNN’accuracy exceeded 90%.…”

“…In this research, we provide a technique for automating the Chicago classification process based on Machine Learning. Initially, we developed a classifier [ 8 ] based purely on the raw pressure topography photos to automatically assess whether the IRP is within the usual range or is over the cut-off. In the Chicago algorithm, determining the IRP type is one of the most crucial tasks.…”