Multimodal brain imaging-based classification of functional constipation subtypes using machine learning.

Area of Science:

• Neuroscience
• Gastroenterology
• Psychiatry

Background:

• Functional constipation (FC) is prevalent and frequently co-occurs with anxiety and depression (FCAD).
• FCAD patients exhibit unique responses to treatment compared to those without anxiety and depression (FCNAD).
• Distinct impacts of FC and mental status on brain structure and function suggest potential for neuroimaging biomarkers.

Purpose of the Study:

• To investigate if brain structural and diffusion imaging features can differentiate FCAD from FCNAD.
• To develop and validate a classification model using neuroimaging data for distinguishing FCAD and FCNAD patients.

Main Methods:

• 187 FC patients underwent structural MRI and diffusion tensor imaging (DTI).
• Patients completed self-reported anxiety and depression assessments.
• A classification model was trained using extracted brain imaging features with a stagewise approach.

Main Results:

• The classification model achieved an average accuracy of 89.23% in cross-validation.
• Predicted FCAD probability correlated significantly with mental health ratings and gastrointestinal symptoms.
• Key differentiating features were identified in brain regions associated with emotion, sensation, and motor control.

Conclusions:

• Brain imaging features can effectively distinguish FCAD from FCNAD.
• These neuroimaging biomarkers may enhance personalized diagnosis and treatment strategies for functional constipation.
• The findings underscore the intricate link between gut health and brain function in FCAD.