A Low Cost Electrical Impedance Tomography (EIT) for Pulmonary Disease Modelling and Diagnosis. Venkatratnam Chitturi, Nagi Farrukh Faculty of Mechanical Engineering Universiti Tenaga Nasional, Selangor, Malaysia Vinesh Thiruchelvam, Thang Ka Fei School of Engineering Asia Pacific University Bukit Jalil, Malaysia chitturi@apu.edu.my, Farrukh@uniten.edu.my, Dr.Vinesh@apu.edu.my, ka.fei@apiit.edu.my ABSTRACT This paper reports the design, development and implementation of a low cost electrical impedance tomography system. A 16 electrode EIT system was designed around a phantom tank. A 60 khz constant sinusoidal current of up to 1 ma is induced in the first and second electrode and the root-mean-square potential on all the adjacent pairs of electrodes placed clockwise is measured directly by the analogue pins of an Arduino Mega microcontroller. This data is fed and then executed in MATLAB which is interfaced with EIDORS software to reconstruct the image based on the acquired data. This is the most economical EIT system designed and tested so far as compared to other EIT systems. KEYWORDS Electrical Impedance Tomography, electrodes, reconstruction, EIDORS. 1. INTRODUCTION Numerous efforts have been made towards the development of more accurate, reliable and faster methods to detect the pulmonary disorders through X-rays, Computed Tomography (CT) scanning, Magnetic Resonance Imaging (MRI) and Ultrasound techniques. Electrical Impedance Tomography (EIT) on the other hand works in a radiation free environment by inducing an electrical current into the medium and extracting the voltage measurements. It is relatively low cost and is easy to operate and maintain. Compared to its counter parts, EIT is free from any form of exposure of X-rays or ionisation radiation. The speed by which the data can be extracted by EIT is very fast due to which the physiological features of the organs can be detected easily. EIT can be used as a bedside monitoring system for patients in clinics or hospitals for those who find it difficult to afford CT scanning or MRI scanning. Electrical Impedance Tomography (EIT) is a tool that allows reconstruction of images of any object. It reconstructs an image based on electrical voltage measurements acquired from electrodes attached evenly around a conductive ring. Current is induced by neighbouring method and the voltage measurements are carried out adjacently as shown in Figure 1[1]. Figure 1: Sequence of voltage Measurements Performed on the Conductive Ring (C. Montellano et al., 2011) ISBN: 978-0-9891305-4-7 2014 SDIWC 83
2. LITERATURE REVIEW Electrical Impedance Tomography (EIT) was introduced in the early 1980s by Barber and Brown and was proposed by B. H. Brown that EIT can be utilized for imaging the human thorax to construct images of lungs and monitoring the ventilation in human lungs. I. Frerichs (2002) after reviewing the experimental and clinical activities of EIT regarding imaging of lungs and monitoring of pulmonary ventilation suggested that there were numerous scientific supports for EIT that helps in extracting information on lung function providing bedside monitoring [2]. Thus, EIT possesses high potential to become a useful technique for reconstructing lung images possessing features such as monitoring over long period of time with no risk of harmful side effects on patients [3]. Although there is capability of EIT to be utilized in the medical field, it has some limitations due to which there is hesitancy in adopting it for routine medical diagnosis. Some of the major limitations include low spatial resolution, vulnerability to noise and electrode errors which can disrupt medical imaging [4]. Keeping these problems in mind, Andrew Adler (1995) developed an idea to reconstruct lung images using artificial neural network (ANN). He proposed that artificial neural network has the ability to develop a model without the intervention of mathematical methods [5]. Andrew Adler (1996) also found out that one of the most significant problems in lung image reconstruction in EIT to detect pulmonary disorders is faced by the electrode movement placed on the thorax of human due to breathing. Since the electrodes are frequently moving due to exhaling and inhaling, their sensitivity is affected. This indicates that the conductivity in the organs changes and is subject to certain errors. To overcome this, simulation was first performed on a two dimensional anatomical finite element model of the thorax to observe the movement of electrodes due to breathing. The data obtained was then utilized by combining with the electrical properties of the lungs to simulate the EIT measurements from which images were reconstructed. Other reasons for poor image quality in EIT technique are low spatial resolution and noise. To curb this problem, X.Y. Chen et al. (2009) proposed an idea of introducing prior information of the physical geometry of the lungs before processing the image. This was an enhancement to the common approach in EIT image reconstruction by providing a solution for the inverse problem of EIT. A modeling package called as COMSOL have been utilized in this method to produce simulation of physical processes through partial differential equations [6]. Another approach for improving spatial resolution in EIT was introduced by Mouloud A. Dena et al. (2010) by developing a prototype for better imaging tool for human lungs using EIT. They have utilized respiratory mechanics in the tool by introducing a physiological model having the exact lung volume and behaving as the human lung behaves [7]. This proposed idea works in a way that when the electrodes are connected to human thorax and the current is induced and voltages are measured, the image reconstruction is performed, the algorithm checks the image obtained from human lung with the physiological model and if the image is same or close to the model, it displays on the monitor. Hence the image will not be too sensitive to noise and the low frequency properties of the model created may act as low pass filters against earlier disturbances. ISBN: 978-0-9891305-4-7 2014 SDIWC 84
For better image resolution of lungs and high density images, an idea for computation of 3-D model for lung imaging was proposed by Shuai Zhang et al. (2012). In this proposed idea, finite element method (FEM) has been used to construct a 3-D model for human thorax that has a cylindrical geometry. With the help of nodal total variation regularized algorithm, the resistivity distribution of the lungs gets wellconstructed giving out high resolution images. 3. PROPOSED METHOD An EIT phantom tank is developed with a shallow plastic bucket having 25 cm as its diameter and of 29 cm height. Sixteen identical stainless steel electrodes are equally spaced on the tank inner wall and fixed over the inner wall using stainless screws, nuts and washers. The electrodes are of rectangular shape (35mm x 10 mm) having 0.05 mm thickness. Stainless steel electrodes have specifically been used to make sure that the corrosion of localized pitting which leads to the creation of small holes does not occur if the phantom tank is filled with 0.9% NaCl solution. The EIT system is required to accept a constant current to the phantom boundary and measure the voltage potentials developed across the surface electrodes. The working of the system is such that a 60 khz constant sinusoidal current of up to 1 ma will be induced in the first and second electrode and the root-mean-square potential on all the adjacent pairs of electrodes placed clockwise will be measured. The advantage of applying current to the electrodes and measuring voltages is that the noise due to spatial variation is reduced when compared to applying voltage to the electrodes and measuring currents. The sinusoidal signal is generated by building a voltage controlled oscillator (VCO) using an ICL8038 IC and feeding it into VCCS which is an enhanced Howland generator developed with LMC6482 ICs. The AC current is injected into an automatic electrode switching module (ESM) which is developed with high speed CMOS analogue multiplexers called CD4067B ICs. Once the sinusoidal AC current is injected into a pair of electrodes, the voltage potentials are measured from across the adjacent pair of electrodes. The switching of the current injection after one cycle of voltage measurements to the adjacent pair of electrodes is done automatically by the multiplexers that are interfaced with the Arduino microcontroller by a program developed to control the multiplexers through Arduino based on the conditions specified. As the current injection is changing at the pairs, voltages are measured across the adjacent electrodes through the microcontroller and saved as.mat file. This process repeats itself until it achieves 13 x 16 = 208 sets of voltage measurements. Finally, this data will be fed into processing unit and executed in MATLAB which is interfaced with EIDORS (Electrical Impedance and Diffuse Optical Tomography Reconstruction Software which is a program introduced by EIT research) software to reconstruct the image based on the data acquired. The complete block diagram of the working system is shown in Figure 2. Figure 2: Block Diagram of the working system ISBN: 978-0-9891305-4-7 2014 SDIWC 85
3.1 Electrode Current. The EIT electronic instrumentation comprises of a constant current injector (CCI) which is made up of a voltage controlled oscillator (VCO) feeding the voltage signal to a voltage controlled current source (VCCS). The VCO is developed by a variable frequency function generator IC ICL8038 as shown in figure 3. ICL8038 is a high precision and a high frequency function generator that is able to produce high accuracy sine, square, triangular, saw tooth and pulse waveforms with a minimum of external components. The frequency of the function generator can be selected ranging from 0.001 Hz to more than 300 khz using external components such as resistors or capacitors and sweeping can be achieved through an external voltage. The circuit diagram of the function generator developed for the EIT system to produce a sinusoidal signal is shown in figure 3. the output signal of the function generator gets distorted easily if its connected directly to the VCCS. Due to the high output impedance of the function generator circuit, a voltage buffer has been used to transfer the voltage of the function generator into the voltage controlled current source (VCCS) which have comparatively low input impedance. A high pass filter with R4 = 100 k ohm and C2 = 47 µ Farad has been used after the sine wave generation to filter the DC components out of the output signal of the function generator and remove or reduce the amplitudes of frequencies lower than the cut off frequency. The voltage controlled current source is found to be the most commonly used current source for EIT systems because of its ability to provide controllable means of current injection into the load impedance. The VCCS built for the system is basically an enhanced Howland current source which is a dual Op-Amp based voltage controlled current source. To get the steady output across the load, this VCCS uses a resistor matching to complete the feedback loops. The circuit diagram of VCCS built for the EIT system is shown in figure 4. Figure 3: Voltage Controlled Oscillator (VCO) The buffer circuit has been developed with the function generator using TL071 OP- Amp because the output current of the function generator is low and it has been observed by connecting it to the VCCS that Figure 4: Voltage Controlled Current Source ISBN: 978-0-9891305-4-7 2014 SDIWC 86
3.2 Current Switching The electrode current switching is developed using Multiplexers. The multiplexers used in this system are CD4067B, they are high speed CMOS analogue multiplexers. For the EIT system, two 1:16 multiplexers (MUX1 and MUX2) are used to develop an automatic switching system for current to inject into the adjacent pair of electrodes after completing one set of voltage measurements. The circuit diagram for the multiplexers is shown in figure 5. The digital bits required to operate and control the multiplexers (MUX1 and MUX2) to switch the current in the electrodes are generated by Arduino microcontroller and are fed into the multiplexers connected to the surface electrodes. Hence, 8-bit parallel digital data is required to operate the two 1:16 analogue multiplexers (MUX1 and MUX2) simultaneously as one 1:16 multiplexer has four control pins. Here, 8-bit parallel digital data are generated using Arduino microcontroller and a program is written to convert it to 16-bit parallel digital data required for automatic switching of the current for the surface electrodes of 16 electrode EIT system. The output sinusoidal current having 60 khz and 4 volt peak-peak is injected into the input of MUX1. When the control pins of the MUX1 are enabled by the Arduino pins in a 4-bit format, the corresponding output channel of the MUX1 will open and the current from the VCCS will flow into the respective electrode attached at the phantom tank and the input of the MUX2 is connected to the ground of the VCCS. The AC current through the multiplexers will first be injected into the first and second electrode pair and the voltage measurements from the adjacent pairs placed clock wise will be measured. When one cycle of voltage measurement is completed, the Arduino will enable the multiplexers control pins so that the corresponding output channel is shifted to the next channel. The switching sequence of the output digital pins from the Arduino and the corresponding channel of the multiplexers are shown in table 1. Figure 5: Current Switching Module ISBN: 978-0-9891305-4-7 2014 SDIWC 87
Table 1: Current Switching Channels Table 2: Voltage measurements from 1 to 16 electrodes The electrodes from within the phantom tank are connected to the MUX1 and MUX2 of the electronic part of the system for the automatic current switching module and to the analogue pins of the Arduino microcontroller for extracting the voltages and recording them in.mat file for the image reconstruction process. 4. TEST RESULTS An empty plastic bottle (insulator) was inserted in the phantom tank and the voltage measurements were recorded and saved it in a format as shown in table 2. Final reconstructed image of the empty plastic bottle after simulation by EIDORS is shown in figure 6. ISBN: 978-0-9891305-4-7 2014 SDIWC 88
Figure 6: Resistivity Image Created by EIDORS Testing of image reconstruction was first done for empty phantom tank and as shown in figure 7. Figure 7: Phantom tank Image Reconstruction Since water is conductive by nature, some of the resistivity elements can be spotted in the image reconstruction of an empty phantom. Finally, Human lung design prototype is made and placed in the phantom tank. The system when run acquired the voltage data corresponding to resistivity of lungs and reconstructed the image is as shown in figure 8. Figure 8: Human Lungs Image Reconstruction 5. CONCLUSION It is observed that EIT scans were limited to imaging when diameter of the object is between 5 cm to 15 cm. To achieve better resolution of the images, Silver/Silver chloride electrodes will be used. All forward and inverse problem solvers existing in EIDORS will be tested to determine which one provides the most satisfactory results. Further, data acquisition can be faster and accurate by replacing the microcontroller with DAQ cards with 32 or more electrodes. 6. REFERENCES [1] Montellano, C. and Garay, E.C.L. and Rodriguez, S. and Rogeli, P. Development of an electrical impedance tomograph 8th International Conference on Electrical Engineering Computing Science and Automatic Control (CCE), pp. 1-4, 2011. [2] Inéz Frerichs*, Sven Pulletz, Gunnar Elke, Günther Zick and Norbert Weiler. Electrical Impedance Tomography in Acute Respiratory Distress Syndrome The Open Nuclear Medicine Journal. Vol 2. (1). p. 110-118, 2010. [3] Shuai Zhang and Guizhi Xu and Xueying Zhang and Bo Zhang and Hongbin Wang and Yaoyuan Xu and Ning Yin and Ying Li and Weili Yan. Computation of a 3-D Model for Lung Imaging With Electrical Impedance Tomography Magnetics. 48. (2). p. 651-654, 2012. [4] Bradley Graham Michael. Enhancements in Electrical Impedance Tomography (EIT) Image Reconstruction for 3D Lung Imaging. Faculty of Graduate and Postdoctoral Studies. Ottawa, Canada: University of Ottawa, 2007. (Thesis) [5] Andrew Adler. Measurement of Pulmonary Function with Electrical Impedance Tomography. Faculty of Engineering. Québec: Montreal University, 1996. (Thesis) [6] Chen, X.Y. and Wang, H.X. and Newell, J.C, Measuring Technology and Mechatronics Automation (ICMTMA) Third International Conference on Lung Ventilation Reconstruction by Electrical Impedance Tomography Vol 2. pp. 489-492, 2011. [7] Denai, M.A. and Mahfouf, M. and Mohamad- Samuri, S. and Panoutsos, G. and Brown, Brian H. and Mills, G.H. Absolute Electrical Impedance Tomography (aeit) Guided Ventilation Therapy in Critical Care Patients: Simulations and Future Trends. Information Technology in Biomedicine.14. (3). p. 641-649, 2010. ISBN: 978-0-9891305-4-7 2014 SDIWC 89