Extraction of Cole parameters from the electrical bioimpedance spectrum using stochastic optimization algorithms.

Area of Science:

• Biomedical Engineering
• Bioelectrical Impedance Analysis
• Computational Biology

Background:

• Accurate extraction of Cole parameters from bioimpedance spectroscopy (BIS) data is crucial for evaluating tissue electrical properties.
• Current methods face challenges with noise sensitivity and parameter accuracy, limiting their physiological and pathological assessment capabilities.

Purpose of the Study:

• To propose and evaluate an improved Cole parameter extraction method using the bacterial foraging optimization (BFO) algorithm.
• To assess the accuracy, noise robustness, and classification performance of the BFO method compared to least squares (LS) and other evolutionary algorithms (GA, PSO).

Main Methods:

• Simulated datasets were used to compare the BFO fitting method against the least squares (LS) fitting method for parameter extraction accuracy and noise sensitivity.
• Experimental BIS data from forearm measurements at different positions were analyzed.
• Extracted Cole parameters from BFO, LS, GA, and PSO methods were used as features for various classifiers.

Main Results:

• The BFO method exhibited higher accuracy and better robustness to noise in parameter extraction compared to the LS method.
• Classifiers utilizing feature sets extracted by the BFO method achieved higher accuracy on both training and test data.
• The BFO method demonstrated improved performance in classifying forearm positions based on bioimpedance measurements.

Conclusions:

• The proposed BFO-based Cole parameter extraction method offers significant advantages in accuracy and noise resilience over conventional techniques.
• This enhanced feature extraction improves the performance of classification tasks for bioimpedance data.
• The BFO method holds promise for more reliable analysis of tissue electrical properties in various physiological and pathological conditions.