Pelvic floor pressure distribution profile in urinary incontinence: a classification study with feature selection

Area of Science:

• Biomedical Engineering
• Medical Diagnostics
• Urogynecology

Background:

• Female urinary incontinence (UI) diagnosis lacks objective measures.
• Pelvic floor pressure distribution profiles offer potential for objective assessment.
• Novel instrumented probes enable detailed pressure mapping.

Purpose of the Study:

• To evaluate pelvic floor pressure profiles for objective UI diagnosis.
• To develop a feature extraction, selection, and classification system.
• To identify features discriminating between continent and incontinent women.

Main Methods:

• Acquired intravaginal spatiotemporal pressure profiles from 48 women (24 continent, 24 incontinent).
• Performed four pelvic floor maneuvers: maximum contraction, Valsalva, endurance, and wave.
• Employed feature extraction, recursive feature elimination, branch and bound, and k-NN, SVM, logistic regression classifiers.

Main Results:

• The best feature subset, identified via branch and bound, achieved 95.8% accuracy.
• Recursive feature elimination with SVM yielded 77.1% accuracy, 74.1% precision, and 83.3% recall.
• No single classifier consistently outperformed others; k-NN showed statistical inferiority in one maneuver.

Conclusions:

• Pelvic floor pressure distribution profiles show potential for objective UI assessment.
• The findings provide insights into pelvic floor function related to UI.
• Further development is needed for a fully automatic diagnostic system.