Development of a multi-frequency system for medical application of Focused Impedance Method (FIM)

Abstract

Monitoring of anatomical structures and physiological processes by electrical impedance techniques has attracted scientists as it is noninvasive, nonionizing and the instrumentation is relatively simple. Focused Impedance Method (FIM) is attractive in this context as it has enhanced sensitivity at the central region beneath the electrode configuration and can localize a region of interest minimizing contribution from neighboring regions. The present work is divided into 3 components. Firstly, a multifrequency FIM (MFFIM) system capable of measuring all three versions of FIM (FIM4, FIM-6 and FIM-8 with 4, 6 and 8 electrodes respectively) in the frequency range 10kHz – 1MHz was developed using IC chips readily available in Bangladesh. Secondly, FEM simulations were used to investigate the 3D sensitivity profiles of all versions of FIM in order to be able to use this system appropriately in a given situation. Thirdly, the feasibility of using MFFIM in the study of localized lung ventilation disorders and breast tumour classification was explored – using both simulation and human subject measurements. For the MFFIM instrumentation, a microcontroller based multi-frequency signal generator and a balanced Howland current source with high output impedance (476𝑘𝛺 at 10𝑘𝐻𝑧 and 58.3𝑘𝛺 at1𝑀𝐻𝑧) were implemented for driving currents of constant amplitude into biological tissues with error <1%. The peak values of voltage signals were measured using two different approaches: analogue synchronous peak detection and digital demodulation. The overall accuracy of the measurement (error< 2%, except 1MHz), resolution better than 0.2Ω and frame rate (> 1.35𝑠𝑒𝑐 −1 ) of the designed MFFIM system was deemed adequate for noninvasive impedance measurements on the human body. The effects of the drive/receive electrode separation, electrode diameter, volume conductor inhomogeneity and proximity of electrode to the boundary on the sensitivity profiles were investigated using FEM simulations and compared for the 3 FIM and standard tetra-polar impedance measurement (TPIM) electrode configurations. Two new parameters, Percentage Depth Sensitivity (PDS) and Degree of localization (dol) were introduced in the present work which suggested that FIM-4 offers superior depth penetration and localization ability when compared to other FIM and TPIM configurations. An experiment was also designed and performed to investigate 3D sensitivity distributions, which gave good agreement with the simulated results. An FEM based model of the human thorax was created in the Comsol Multiphysics software and focused impedance measurements were simulated with an FIM-4 electrode probe placed on the surface of the thorax. The changes in impedance between maximum inspiration and expiration on different vertical levels of thorax were in the range 5-24% for the FEM model whereas the changes were in the range 5-17% for a human subject. The impedance changes at different vertical levels on the thorax model matched well with experimental findings on the human subject. A simulation study showed that the impedance change between maximum inspiration and expiration, measured using an FIM-4 probe with appropriate electrode separation, was substantially lower in the region of thorax where the underlying lung is filled with fluid (5.1%) compared to that without fluid (16.8%) for a fluid volume of 663cm3 . Based on this simulation study, a new technique is proposed to detect fluid accumulation in lungs using FIM measurements. An experimental study was performed on 16 female subjects for classification of palpable breast tumours (whether the tumour is benign or malignant) using the developed MFFIM device. Applying k-nearest neighbors (k-NN) algorithms the sensitivity, specificity and prediction accuracy of breast tumour classification was found to be 75%, 87.5% and 81.25% respectively for k=3. Tissue anisotropy was found to be an important feature for classification of breast tumours. The area under ROC curve for the feature representing tissue anisotropy was 0.82. Overall, the present work has produced a reliable and accurate multi-frequency universal FIM system for noninvasive impedance measurements on the human body, suitable for a resource constrained setting with limited availability of electronic components. The work has expanded knowledge of the spatial sensitivity characteristics of FIM and demonstrated how this new knowledge can help the application of this technique in physiological measurements. In this thesis this knowledge is applied in the detection of lung disease and to the characterization of breast tumours, through modeling and preliminary measurements on human subjects.