Automated Evaluation of Urodynamic Examinations Through Local Linear Models: Validation on Spinal Cord Injury...

Area of Science:

• Clinical Engineering
• Medical Devices and Systems
• Urology

Background:

• Spinal cord injury (SCI) often leads to Detrusor Overactivity (DO), complicating bladder management.
• Accurate diagnosis and measurement of DO from cystometry data are crucial for effective patient care.
• Current methods for analyzing urodynamic data can be inconsistent and lack automation.

Purpose of the Study:

• To develop and validate a robust, automated method for classifying Detrusor Overactivity (DO) events.
• To enable automatic calculation of clinical measurements, such as bladder compliance, from cystometry data in SCI patients.
• To improve the consistency and objectivity of urodynamic study assessments.

Main Methods:

• A two-stage automated approach was employed to detect DO events using local linear models and thresholding.
• A segmentation method was utilized to identify the start and end times of DO events.
• The method was developed and tested on 77 urodynamic samples from SCI individuals with 158 annotated DO events.

Main Results:

• The automated method achieved 100% accuracy in patient-level DO diagnosis.
• DO event detection showed a precision of 0.94 and recall of 0.72.
• Automated bladder compliance estimation demonstrated a median absolute error of 5.20 ml/cmH2O, with 88% accuracy in classifying bladder compliance levels.

Conclusions:

• The proposed automated method provides accurate and objective assessment of cystometry data for DO diagnosis.
• This tool empowers medical teams to consistently evaluate urodynamic studies, aiding disease characterization and clinical decision-making in SCI patients.
• The methodology offers a standardized approach for clinical assessments and can be extended to other patient populations.