Functional gut disorders or disordered gut function? Small bowel dysmotility evidenced by an original technique

Area of Science:

• Gastroenterology
• Medical Imaging
• Computational Biology

Background:

• Functional bowel disorders (FBDs) affect a significant portion of the population.
• The underlying pathophysiology of FBDs remains incompletely understood.
• Objective assessment of intestinal motility in FBDs is challenging.

Purpose of the Study:

• To determine the prevalence of abnormal intestinal motility in patients with FBDs.
• To validate a novel computer vision and machine learning method for analyzing capsule endoscopy images.
• To differentiate motility patterns in FBD patients from healthy controls.

Main Methods:

• Capsule endoscopy (Pillcam SB1) was performed on 80 FBD patients and 70 healthy controls.
• A computer vision program analyzed endoluminal images for motility events and patterns.
• A machine learning classifier (one-class support vector machine) established normality and identified abnormalities.

Main Results:

• The classifier identified 29% of FBD patients (23/80) as having abnormal intestinal motility.
• Healthy subjects were correctly classified as normal with 97% accuracy (68/70).
• Abnormal motility patterns in FBD patients included hyper- and hypodynamic subtypes, unrelated to clinical features.

Conclusions:

• Abnormal intestinal motility is demonstrable in a significant subset of FBD patients.
• The findings suggest a pathological disturbance in gut physiology contributes to FBDs.
• Computer-aided analysis of capsule endoscopy offers a promising tool for FBD research.