Non-destructive characterization of bone mineral content by machine learning-assisted electrochemical impedance spectroscopy

Abstract

Continuous quantitative monitoring of the change in mineral content during the bone healing process is crucial for efficient clinical treatment. Current radiography-based modalities, however, pose various technological, medical, and economical challenges such as low sensitivity, radiation exposure risk, and high cost/instrument accessibility. In this regard, an analytical approach utilizing electrochemical impedance spectroscopy (EIS) assisted by machine learning algorithms is developed to quantitatively characterize the physico-electrochemical properties of the bone, in response to the changes in the bone mineral contents. The system is designed and validated following the process of impedance data measurement, equivalent circuit model designing, machine learning algorithm optimization, and data training and testing. Overall, the systematic machine learning-based classification utilizing the combination of EIS measurements and electrical circuit modeling offers a means to accurately monitor the status of the bone healing process.

Electrical impedance detects early stages of bone healing: An in vivo explanatory study of tibial fractures in rabbits

“…Further steps also involve the model being tested in in vivo large animal models prior to human clinical trials. Additionally, a recent study by Banerjee et al developed an analytical method combining electrical impedance spectroscopy and machine learning for the quantitative assessment of bone mineral content [ 1 ], which can be another innovational part of electrical impedance spectroscopy use.…”

Emerging nanomaterials to enhance electrochemical impedance spectroscopy for biomedical applications

“…A set of frequencies is chosen to maximize the impedimetric difference between the target tissue and its environment. A computational algorithm can evaluate the EIS data at these frequencies to predict the type and severity of disease, as evidenced by a recent work performed by Banerjee et al (2022), which successfully applied a machine learning model to characterize bone mineral content.…”