Modified 16-Quasi Log Periodic Antenna Array for Microwave Imaging of Breast Cancer Detection

Abstract

In this paper, an effective system for microwave imaging of breast tumor detection using modified 16-planar log periodic antenna (PLPA) array is presented. The modified PLPA operates in the band from 2 to 5 GHz with stable directional patterns in the end-fire direction. Once the results of a single antenna element have been validated, the design is extended to include 16 antenna elements. All 16 transceiver antennas are vertically placed around the phantom in a circular manner where one antenna acts as a transmitter and the rest work as receivers. Delay and Sum (DAS) algorithm is used for post processing the acquired scattered signals from the sensors to reconstruct the image of the breast and to identify the existence of breast tumors. The electromagnetic simulators CST and HFSS are used to design the system, while MATLAB is used to process the data. The developed PLPA array-based microwave imaging system performs admirably, making it one of the most effective systems for detecting tumor cells.

1. Introduction

Women die from breast cancer more often than any other cancer worldwide. Breast cancer develops in a very complex way, involving multiple cell types, and prevention of the disease remains a challenge. Early detection of breast cancer can greatly reduce the possibility of its progression. Breast cancer patients in some developed nations have a 5-year survivorship of more than 80% due to early detection and treatment [1]. For detection of breast cancers, various imaging technologies, such as ultrasound, X-ray, Computed Tomography (CT), Magnetic Resonance Imaging (MRI), and nuclear medicine imaging, have been developed over the last decades [2]. As of now, microwave imaging is a promising technique to identify abnormalities in the human body based on the electrical characteristics of both healthy and malignant tissues. As opposed to other medical imaging methods, microwave imaging is characterized by the fact that it is low in complexity, low in cost, and does not generate ionizing radiation.

Several ultra-wideband (UWB) antennas for Microwave Imaging System (MIS) have been proposed in recent decades, including the slotted antennas [3,4,5], dipole antenna [6], coplanar waveguide (CPW) antenna [7], vivaldi antennas [8,9,10], Vee dipole [11], EBG antenna [12], tunable antenna [13], and resistively loaded bowtie antenna [14]. The antenna is a significant piece of equipment that collects signals from adjacent scattered objects or sends signals to nearby objects. The performance of a MIS is mostly determined by the antennas that are employed as transmitters and receivers in the platform. To build a therapeutically effective MIS, it is critical to clearly understand the characteristics of the microwave antenna. The antenna used for transmitting and receiving should be planar, wideband, compact, directional, and have a high radiation efficiency in order to produce an effective MIS with high resolution and dynamic range. The wideband features allow for great resolution and precise target localization while imaging. Furthermore, information spanning a wide frequency spectrum can be acquired in a single test, allowing for fast data collecting and microwave detection. Very short microwave energy pulses are required in a near field wideband detection, which is an important requirement for medicinal applications involving electromagnetic fields [15]. In this case, the planar log periodic antenna is a suitable candidate for MIS as it has high gain, low profile, wide bandwidth, and high directional patterns [16].

There are a number of simulations and experimental results that have been shown to detect breast tumors using MIS [17,18,19,20,21,22,23,24,25]. In [17], an experimental near-field MIS prototype at Dartmouth College was described. It used sixteen transceiving monopole elements between 300 and 1000 MHz. A tank between the antennas and the breast served as a coupling medium. The array of antennas was moved vertically and tuned using the mechanical switches. In [18], a monostatic radar-based MIS was developed to scan patients in the clinical trial. The system consisted of antipodal Vivaldi antenna as a sensor, laser, and cylindrical container with canola oil as a coupling medium to reduce skin reflections. The Delay and Sum algorithm (DAS) produced consistent results in the wideband range but was limited to only eight patients. In [19], a multistatic 16-antenna hemi-spherical array operating in the 2–4 GHz frequency range was used in the time-domain radar system. The results of an investigation into measurement repeatability and its impact on image reconstruction were presented. The measurement repeatability was high due to careful patient positioning in the system and in relation to the immersion medium. The measurement setup consisted of reflectors, pulse generators, couplers, and amplifier. In [20], an UWB-based near-field MIS using a raster scan algorithm was presented. The researchers presented the measured and simulated results using a homogenous and a heterogeneous 3D breast phantom. The proposed electromagnetic transverse horn antenna which is directional in nature helped to provide the scanning system with a stronger dispersion signal from dielectric phantom. The sensors of the system were in direct contact to the imaging domain and the image quality was improved by applying deconvolution technique.

In [21], an UWB biomedical MIS featured a planar antenna array that included 12 tapered slot antenna elements. To extend the dynamic range of the imaging system, the array was immersed in a specially designed matching liquid, whose dielectric constant was in accordance with the dielectric constant of the imaged object. For testing, a suitable platform was designed and built to accommodate the array, phantom, Vector Network Analyzer, switching system, and a coupling liquid medium. Although the platform was built and the experiment was conducted, there was no information regarding the imaging technique and signal processing mechanism. Furthermore, the time domain system as designed in [22] consisted of 20 antenna elements working in range of 100 MHz–3 GHz for medical applications. The system consisted of a waveform generator, power amplifier, clock generator, analog-to-digital converter (ADC), and track-and-hold circuit. Complementary metal-oxide semiconductor (CMOS) circuit-based MIS was developed in [23]. A total of 16 UWB sensors were used around the breast phantom along with central master controller. The scattered data was collected by conducting the experiment multiple times and the Delay and Sum algorithm was used to perform the image restoration. As shown in [24], handheld MIS was developed with 16 antenna sensors arranged in a 4 × 4 cross shaped fashion. The sensors are controlled by a single-pole-eight-throw (SP8T) switching module, and the breast imaging is created by the DAS algorithm. This system is very useful because it is portable; however, the design structure is complex, making it difficult to detect tumor. In [25], 16 modified UWB Vivaldi antenna elements were designed to develop an application for the detection of breast cancer. Eight of the sensors were arranged vertically, with the remaining eight arranged horizontally, around the homogeneous breast phantom. Data from 240 scanned positions was collected, and imaging performance was calculated using MATLAB software.

In this paper, a compact coplanar waveguide (CPW) fed planar log periodic antenna (PLPA) covering operating band of UWB from 2–5 GHz is designed for microwave imaging system. The novelty of the work is the use of quasi log periodic antenna as the sensor element in MIS. A total of 16 antennas are placed around the phantom in a circular manner. The antenna in the imaging system is used as a sensor to send and receive microwave signals to the human body. To make the use of this principle and apply it to a practical microwave imaging system for detecting tumors, an UWB pulse is sent to the suspected area of the target breast phantom using the prototyped transceiver antenna. The signals reflected by backscattering are picked up by the prototyped antenna and then analyzed with a suitable computing system to determine whether or not there is a tumor. The results are presented to verify the operation of the system. Due to features, such as non-ionizing, less expensive, and no side effects, the proposed imaging technique can frequently be used to detect tumors in breast.

2. Antenna Design

The modified PLPA antenna prototype was designed and fabricated on FR4 substrate with permittivity 4.4, thickness 1.6 mm, and loss tangent 0.02. The prototype was planar and compact with the dimension of 50 × 40 mm². The antenna structure is shown in Figure 1. The patch and FR4 substrate are represented by the orange and peach color, respectively. The substrate material was preferred to be FR4 as it is inexpensive, easily available, simplicity fabrication, and design resilient. The design patch was fed via 50 Ω CPW feedline by using SMA connector. The CPW feedline is best suited for obtaining wider bandwidth, better matching, ease of integration, and lower dispersion. The proposed antenna covers the frequency band from 2–5 GHz. The antenna was made up of a simplified single printed layer fed by CPW, which means that all the dipoles and feedline are printed on a single side of the substrate. The 50 Ω feedline was connected to a single strip transmission line, along which the five dipole elements are arrayed on the top layer of the FR 4 substrate.

The PLPA structural dimensions were optimized as: L = 50 mm, W = 40 mm, w_f = 3 mm, w_g = 18 mm, l_g = 10 mm, l₁ = 3 mm, l₂ = 5 mm, l₃ = 2.5 mm, l₄ = 2.5 mm, l₅ = 2.5 mm, l₆ = 3.5 mm, l₇ = 1.5 mm, w₁ = 11.5 mm, w₂ = 9.5 mm, w₃ = 6.5 mm, w₄ = 4.5 mm, w₅ = 6 mm, and w₆ = 2.5 mm. The main purpose of this prototype was to act as sensor in an array form in MIS for breast cancer detection. The larger bandwidth, better gain, and stable directional radiation patterns are necessary for the MIS.

3. Parametric Study

As the performance of the antenna mostly relies on the larger dipole, the parameters w₁ and w₆ of this structure are optimized and the parametric analysis is carried out by the HFSS simulation software.

3.1. Effect of the Length w₁

The variation of w₁ alters the bandwidth performance of the PLPA, as shown in Figure 2. The value of w₁ is changed by 1mm increments from 10.5 mm to 13.5 mm, while keeping all other parameters constant. At 3 GHz, the matching is not great for the lengths 10.5 mm and 13.5 mm. In comparison to 11.5 mm, the return loss for 12.5 mm is not improved. As a result, the most optimized w₁ value is 11.5 mm, which provides superior matching and return loss characteristics. At lower resonant frequency, the reflection coefficient is around −20 dB while at higher resonant it is −35 dB.

3.2. Effect of the Length w₆

The effect of variation of w₆ on reflection coefficient is shown in Figure 3. The value of w₁ is fixed to 11.5 mm. In 2 mm increments, the value of w₆ is changed from 2.5 to 8.5 mm. The higher resonant frequency is clearly reliant on w₆, whereas the lower resonant frequency is not. Hence, the factor of tuning comes into picture where one factor (one dimension) at a time is taken into consideration during optimization. The higher resonant continues to degrade as the value of w₆ increases from 2.5 mm to 8.5 mm, until it is completely lost. As a result, the final optimized w₆ value obtained is 2.5 mm.

4. Analysis and Discussion

LPKF protomat (S103) circuit board plotters were used to fabricate the proposed antenna prototype and a Tektronix USB Vector Network analyzer (TTR506A) was used to evaluate its performance. An experimental prototype of the antenna can be seen in Figure 4. It appears that the simulated and measured return loss for the antenna design, shown in Figure 5, are almost in good agreement. The observed measurement results indicate that the antenna operates in the frequency range of 2–5 GHz (−10 dB criteria). An apparent slight deviation between experimental results and simulation data can be attributed mainly to the effect of the SMA connector, SMA soldering, cables, and fabrication tolerance. In simulation, the antenna is excited using a waveguide port, but in practice it is excited via the SMA connector. The characteristics thus obtained aid the PLPA in its role as a sensor in MIS.

For better understanding of the designed antenna performance, the antenna current density distribution at 3.5 GHz and 5 GHz are shown in Figure 6. The current distribution at 3.5 GHz can be seen to be highly concentrated on the large dipole and along the transmission strip line, resulting in a directional radiation pattern towards the smaller dipole elements. A similar current distribution can be seen at higher frequencies of 5 GHz. Over the entire radiating patch, high density is concentrated around the larger and adjacent dipoles, resulting in a directed pattern.

The length of the large dipole element (L_dipole), is given by the following equation:

$${\mathrm{L}}_{\mathrm{dipole}}=[{\mathrm{w}}_1+{\mathrm{w}}_{\mathrm{f}}+{\mathrm{w}}_6]\approx \frac{\mathrm{c}}{2{\mathrm{f}}_{\mathrm{H}}\sqrt{{\mathsf{\epsilon }}_{\mathrm{eff}}}}$$

(1)

$${\mathsf{\epsilon }}_{\mathrm{eff}}=\frac{{\mathsf{\epsilon }}_{\mathrm{r}}+1}{2}+\frac{{\mathsf{\epsilon }}_{\mathrm{r}}+1}{2}\frac{1}{\sqrt{1+12\mathrm{h}/{\mathrm{l}}_7}}$$

(2)

where c = 3×10⁸ m/s, ε_eff is the effective permittivity, ε_r is the substrate dielectric constant, h is the height of substrate, l₇ is the width of the dipole element (all dipole widths are equal), and f_H is the highest frequency in the operating band. According to Equation (1), the theoretical and practical lengths of the larger dipole element are 16.2 mm and 17 mm, respectively.

Figure 7 depicts the 2D and 3D radiation patterns at 3.5 and 5 GHz. The 2D patterns at both frequencies are taken from XZ plane (ɸ = 0° plane). The proposed antenna is directional, with the main radiator patch pointing in the direction of the boresight as seen in the far field measurement. Over the operational band, the primary lobes of the patterns are fixed towards the end-fire direction. Hence, the radiation patterns are stable over the operating frequency band and thus enhance the sensor element in MIS for tumor detection. The CPW fed PLPA covers bandwidth of 3 GHz (−10 dB criteria) from 2–5 GHz with maximum peak gain of 4.6 dBi. The peak gain over the entire band is shown in Figure 8. At a lower frequency, the antenna has upward gain while it achieves gain stability at higher frequency.

5. Microwave Imaging Setup

With the aid of CST and HFSS simulation software, the 16-antenna array is tested for its ability to detect breast tumor by means of a breast phantom, seen in Figure 9. The top and slant view of the antenna array system around the breast phantom is depicted in Figure 9a,b, respectively. In this study, two breast phantoms are examined. The phantom is made of different layers of skin and fat tissue with an inserted tumor. The antenna array is placed in a circular manner around the phantom of radius 60 mm with an air gap of 20 mm, thus the sensor elements are not in direct contact to the breast phantom. The dielectric constant of the skin, fat, and tumor are 38, 5.14, and 67, respectively. In the first phantom, a cylindrical tumor with a radius of 5 mm and a height of 20 mm is inserted at location x = 30 mm and y = 30 mm (position A) from the center of the phantom. In the second phantom, an identical tumor is inserted at location x = –30mm and y = 30mm (position B) from the center of the phantom.

Different configurations of 3D circular antenna arrays were initially simulated; after optimization, 16 element array is selected. The main idea is to incorporate as many elements as possible to the extent that the coupling between elements is in the acceptable range. Data collection at several angular positions will be less complex if there are more elements and hence avoids the unnecessary rotation of the system. However, with the increase of the array elements, the simulation time will increase dramatically. Due to the large number of mesh cells to resolve, it is difficult to optimize this large array by HFSS or CST domain solver. Using optimization and symmetry techniques to reduce the simulation time, the 16-array configuration was selected and it helped in successfully reconstructing images from the collected data using MATLAB. Figure 10 illustrates the array bandwidth range for a 16 element array ranging from 2 to 5 GHz. Approximately 8 cm separates the antennas from the center of the phantom. The array elements are spaced at 22.5 degrees apart, which results in transmission coefficients below −20 dB in the presence of the breast phantom. As shown in Figure 10, the antenna array exhibits reflection and transmission coefficients with less than −10 dB over a bandwidth of 2 to 5 GHz.

In the proposed MIS, there are 16 antennas, one of which acts as a transmitter, while the remaining 15 antennas act as receivers to collect scattered signals from the phantom. The entire process is repeated for all antennas, resulting in 240 (16 × 15) scanned data points from the breast phantom. The collected backscattered signals are then analyzed using DAS algorithm with a suitable computing system to determine the presence of tumor. A picture of the imaging result can be seen in Figure 11 for two different locations of tumor in breast phantom. The ranges for x, y, and z are measured in meters. The 3D phantom is in hemispherical shape. As we move along the z axis the radius of the circular slice starts decreasing till it reaches the apex point. The results are obtained at different horizontal planes as slices along the z axis. The Figure 11 shows 2D images at z = 0.005 m and z = 0.02 m positions (i.e., a slice of image at z location of 5 mm and 20 mm). Therefore, if the tumor is located along the z axis it will be detected and shown in the horizontal slices at that z-location. As can be seen, the proposed system successfully detects the presence of tumor with a dark red color. A comparison of different MIS in literature with the proposed one is presented in

Table 1. Comparison of different ultra-wideband (UWB) microwave imaging system with the proposed one.

Reference	Antenna	Dimension (mm2)	Frequency (GHz)	Number of Elements	Arrangement	Imaging Method
[2]	CPW fed EBG Antenna	76 × 44	3.1–7.6	2	2 element arrays	DMAS (Delay Multiply and Sum Algorithm)
[12]	EBG antenna	31.02 × 31.68	7.1–7.4	2	2 element arrays	Confocal imaging
[21]	Slotted antenna	40 × 22	3.1–10.6	12	Planar array	-
[24]	Planar slot UWB	11 × 13.1	3.1–10.6	16	4 × 4 crossed shaped	Confocal imaging
[25]	Vivaldi antenna	40 × 40	2.5–11	16	8 vertically 8 horizontally	DAS
[26]	UWB Flexible antenna	20 × 14	4–6	2	2 element arrays	DMAS
[27]	Balanced antipodal Vivaldi antenna	-	1–13	1	36 positions	TSAR (Tissue Sensing Adaptive Radar)
Proposed	Quasi Log Periodic	50 × 40	2–5	16	Circular array	DAS

. In this regard, the system designed proves to be a good method for detecting breast cancer using microwave imaging.

6. Conclusions

The design, simulation, and fabrication of a portable, lightweight, and compact PLPA element was presented in this paper. The 50 × 40 mm² PLPA characteristics were investigated and verified for microwave imaging. There exists a modified sensor element in the proposed MIS, and the designed system consists of 16 antenna elements operating at 2 to 5 GHz. The back scattered signals collected by the circular antenna array elements are processed using DAS algorithm in MATLAB to reconstruct the image to detect the presence of breast tumor. An artifact removal method based on rotation subtraction was applied in this technique. PLPA array-based MIS presents an excellent performance, making it one of the most efficient methods for detecting tumor cells.