Classifier of intestinal contractile activity degree based on internal electroenterogram recording.

Area of Science:

• Gastroenterology
• Computational Biology
• Biomedical Engineering

Background:

• The intestinal interdigestive motor migratory complex (IMMC) plays a crucial role in gastrointestinal motility.
• Anomalies in IMMC patterns are indicative of various gastrointestinal pathologies.
• Accurate classification of IMMC states is essential for diagnosing motor function disorders.

Purpose of the Study:

• To develop an automatic classifier for discriminating between different degrees of intestinal contractile activity within the IMMC.
• To utilize electroenterogram (EGG) recordings for classifying IMMC states.
• To evaluate the efficacy of machine learning algorithms, specifically Linear Discriminant Analysis (LDA) and Support Vector Machines (SVM), for this classification task.

Main Methods:

• Internal electroenterogram (EGG) recordings were used as the data source.
• Spectral and statistical parameters were extracted from EGG signals to serve as features.
• Classifiers based on Linear Discriminant Analysis (LDA) and linear Support Vector Machines (SVM) were implemented and compared.
• Performance was evaluated based on classification accuracy for binary (quiescence vs. maximum activity) and multiclass (quiescence, irregular, maximum activity) scenarios.

Main Results:

• Linear Support Vector Machines (SVM) demonstrated slightly higher accuracy than Linear Discriminant Analysis (LDA).
• A binary SVM classifier achieved approximately 91% accuracy in distinguishing between quiescence and maximum contractile activity.
• A multiclass SVM classifier achieved approximately 74% accuracy in differentiating between quiescence, irregular, and maximum contractile activity.

Conclusions:

• Preliminary results indicate that SVM classifiers are effective tools for the automatic classification of intestinal contractile activity degrees.
• The developed classification system shows potential for identifying the IMMC and diagnosing anomalies in intestinal motor function.
• Further improvements in accuracy may be achievable by incorporating additional features and exploring non-linear SVM classifiers.