Diagnosis analysis of rectal function through using ensemble empirical mode decomposition-deep belief networks...

Area of Science:

• Biomedical Engineering
• Signal Processing
• Artificial Intelligence

Background:

• Rectal motility function is a key indicator of rectal health.
• Diagnosing rectal function post-artificial anal sphincter implantation presents challenges due to signal instability and noise.
• Existing methods struggle with effective feature extraction from complex rectal pressure signals.

Purpose of the Study:

• To develop and validate a robust rectal function diagnosis model.
• To improve the accuracy of distinguishing normal from abnormal rectal motility functions.
• To provide technical support for patients with artificial anal sphincters.

Main Methods:

• Ensemble Empirical Mode Decomposition (EEMD) was employed to decompose noisy rectal pressure signals, reducing modal mixing.
• Deep Belief Networks (DBNs) were utilized for deep learning to extract multi-dimensional features from decomposed signals.
• A classification model was optimized by adjusting the number of Restricted Boltzmann Machines and DBN layers.

Main Results:

• EEMD effectively reduced mode aliasing compared to standard Empirical Mode Decomposition (EMD).
• The proposed EEMD-DBNs model successfully distinguished between normal and abnormal rectal pressure signals.
• An optimized DBN model (3 RBMs, 5 layers) achieved an 85% diagnosis rate, balancing accuracy and speed.

Conclusions:

• The EEMD-DBNs model offers a reliable method for diagnosing rectal motility function.
• This approach provides valuable technical support for the recovery of rectal function in patients with artificial anal sphincters.
• The study demonstrates the potential of advanced signal processing and deep learning in clinical diagnostics.