Portable Ultrasound Bladder Volume Measurement Over Entire Volume Range Using a Deep Learning Artificial Intelligence...

Area of Science:

• Urology
• Medical Imaging
• Artificial Intelligence

Background:

• Accurate bladder volume measurement is crucial for diagnosing lower urinary tract symptoms.
• Conventional methods using portable ultrasound bladder scanners (PUBS) may have limitations in accuracy.

Purpose of the Study:

• To compare the accuracy of AI-assisted bladder volume (AI-BV) measurement against conventional bladder volume (C-BV) measurement using PUBS.
• To determine if AI algorithms can improve bladder volume estimation.

Main Methods:

• Prospective investigation of 250 patients undergoing filling cystometry.
• Bladder volume was measured using PUBS (C-BV) and a deep learning AI model (AI-BV).
• Comparison of AI-BV and C-BV against true bladder volume (T-BV).

Main Results:

• AI-BV showed no significant difference compared to T-BV (p=0.081), unlike C-BV (p=0.001).
• The mean difference between AI-BV and T-BV was significantly smaller than between C-BV and T-BV (6.5 vs 15.0 mL).
• Both AI-BV (R²=0.90) and C-BV (R²=0.91) were highly correlated with T-BV.

Conclusions:

• Deep learning AI-BV offers more accurate bladder volume estimation than conventional methods in the studied cohort.
• Further studies are needed to determine clinical relevance and performance in external populations.