Nowadays, wireless communication technology in microwave frequency range has been applied in both commercial and industrial applications. In addition, it has received a great attention in applying to medical applications such as telemedicine and medical imaging [1–3]. The breast cancer has the highest rate of incidence in women and it leads to the cause of death among women around the world [4]. Therefore, early diagnosis of breast cancer can increase the chances of successful treatment. Currently, the clinical methods are often used for diagnosing breast cancer such as X-ray mammography, Ultrasound imaging and magnetic resonance imaging (MRI) [5]. However, X-ray mammography has false negative diagnoses between 4% to 30% and false positive more than 5% [6]. In addition, mammography is often painful from breast compression during examination [7] and it can also generate ionizing radiation causing the cancerous of healthy tissue [8]. Ultrasound imaging is a painless technique compared to X-ray mammography. However, this technique has low accurate to distinguish between normal and malignant tissues [9]. The MRI is a medical imaging technique that has high sensitivity and accuracy. However, the MRI examination is very expensive [10]. To overcome these limitations, breast cancer detection through electromagnetic radiation in microwave frequency range has received attention in recent years. The advantages of microwave-based detection techniques are accurate, low cost, low complexity, nonionizing radiation and comfortable breast detections. These techniques are based on the difference of electrical properties between normal and malignant tissues to detect breast tumors. The electrical properties of malignant tissue are 5 to 10 times larger than those of normal breast tissue [11]. The antenna is one of important elements to use in the microwave breast imaging system. To increase resolution of microwave imaging system, the several antennas are arranged around the breast [12]. However, the detection accuracy may be reduced due to the mutual coupling between antenna elements, which are placed close to each other. To overcome these problems, a single scanning antenna is used for breast cancer detection and microwave imaging system [13]. It is found that the ultra-wideband (UWB) antennas are used because wide range of frequency points are used in microwave imaging algorithm and breast cancer detection. The antipodal Vivaldi antennas (AVAs) are suitable for breast cancer detection system because it can provide ultra-wide bandwidth and good radiation characteristics [14]. The design of AVA to obtain the desired performance, while maintaining the compact size and low manufacturing complexity, is challenging. Various techniques have been reported to improve the performance of AVA such as the use of metamaterial, the dielectric director or dielectric lens, corrugation or slot edge and parasitic patch [15]. The different types of metamaterial unit cells such as negative index metamaterial (NIM) [16], zero index metamaterial (ZIM) [17] and epsilon near zero (ENZ) [18] metamaterial are loaded into the AVA structure to enhance the gain. However, metamaterial unit cells increase complexity in antenna design. Moreover, the dielectric director or dielectric lens as an extension of antenna substrate is introduced in front of the AVA to improve the antenna gain, especially at higher frequencies [19]. The dielectric lens can be produced by using special shape of dielectric substrate [20,21]. Though this technique provides good performance but it leads to increase the antenna size and does not improve the gain at lower frequencies. Therefore, the corrugation on the both edges of the AVA is used to enhance the gain and radiation pattern especially at low frequency range without increasing the antenna size. It is found that cutting the exponential slots on both sides of conducting arms can increase the electrical length of the AVA, leading to extend the lowest edge frequency and improve radiation pattern [22,23]. Moreover, the several uniform corrugations with simple slot shapes are considered [24]. However, it is well known that the slot length and shape of corrugation have a direct impact on the reflection coefficient and radiation characteristics. Therefore, the non-uniform corrugation is focused instead of the uniform corrugation [25,26]. Furthermore, the various shapes of parasitic are added in the flare aperture for enhancing the directivity and gain of AVA [27–30]. It is known that the design of parasitic patch increases complexity because it depends on the shape, size and position of the parasitic patch. However, the electric field distribution of AVA with parasitic patch is focused toward the end-fire direction which is desired for near-field microwave imaging medical applications.

This paper presents the design and analysis of ultra-wideband antipodal Vivaldi antennas for breast cancer detection. Two candidate AVA structures i.e., AVA with non-uniform corrugation and AVA with parasitic patch and corrugation are presented. Then, the design and fabrication procedure of breast phantom by using a gelatine-oil mixture technology are detailed. The prototype of proposed AVAs are simulated, fabricated and measured with breast phantom in order to detect the cancerous cell inside the breast. In addition, the different AVA structures i.e., original AVA, AVA with dielectric lens in [19] and proposed AVAs are compared performance for breast cancer detection. The design of proposed antenna is proven ability of enhancing the directivity and improving the electric field distribution at higher frequencies without design complexity. These properties of antenna are desired for breast cancer detection and microwave imaging applications.

In this section, the design of two AVA structures i.e., AVA with non-uniform corrugation and AVA with parasitic patch and corrugation are proposed. The characteristics of the proposed AVAs were optimized by using CST microwave studio to operate in frequency range from 3–8 GHz for breast cancer detection. The reflection coefficient measurements are performed using a vector network analyzer (Agilent Technologies E5071C ENA Series 9 kHz-8.5 GHz) between 1 and 8.5 GHz. Moreover, a far-field anechoic chamber is used to measured the radiation pattern and gain.

The simulated and measured results for the four AVA structures i.e., original AVA, AVA with non-uniform corrugation, AVA with parasitic patch and corrugation and AVA with dielectric lens presented in [19] in term of S₁₁ and gain are summarized in Tab. 3. It is known that increasing the antenna size, which will increase effective surface of antenna, provides a higher gain. Therefore, AVA with dielectric lens presented in [19] with dimension of 75 mm × 140 mm has the highest gain, follows by, AVA with parasitic patch and corrugation, AVA with non-uniform corrugation and original AVA. As the same antenna dimension, AVA with non-uniform corrugation can enhance the gain of original AVA around 1 dBi at 6 GHz. It is known that breast cancer detection is based on near-field scenario, which the incident wave shape varies with radiation distance and the plane wave can only assumed in the far-field region. Therefore, electric field distributions of antenna in near-field region should be considered. The comparison of simulated electric field distributions in xy-plane of the four AVA structures listed in Tab. 3 at 5, 7 and 9 GHz is shown in Fig. 10. It is clearly obvious that the electric field distribution of AVA with parasitic patch is more focused toward end-fire direction than those of the other AVAs at higher frequencies. At low frequency, for instance, at 4 GHz, AVA with dielectric lens [19] and AVA with parasitic patch produce plane-like wave, while original AVA and AVA with non-uniform corrugation produce spherical-like wave. At high frequencies, for instance, 7 and 9 GHz, all the four AVA structures produce plane-like wave relatively well at 7 GHz but electric field distribution of AVA with dielectric lens [19] becomes weak in the center of antenna due to the off-axis radiation at 9 GHz.

The comparison between the proposed AVAs and other AVAs found from literature is shown in Tab. 4. The proposed AVAs can provide rather high gain while achieving smaller size, simple structure and low complexity in design. Therefore, the proposed AVAs is suitable for breast cancer detection, which requires good radiation properties.

For experimental proposes, the breast phantom is designed and fabricated. The breast model consists of skin and fat layers. The cancerous cell is inserted into the fat layer. It is found from the Cole-Cole model that the electrical properties of human tissue are different and vary with frequency [33]. The dielectric constant of biological tissue based on Cole-Cole model is given by (6)εω=ε∞+εS−ε∞1+(jωτ)1−αwhere εω is complex dielectric constant, εS is static frequency dielectric constant, ε∞ is infinite frequency dielectric constant, ω is angular frequency, τ is time constant and α(0<α<1) is exponent parameter. It can be seen from Eq. (6) that the dielectric constant will reduce as operating frequency increases. In addition, the contrast for relative permittivity and conductivity between malignant and normal tissues has been found around 5:1 and 6:1, respectively, across frequency range from 1–10 GHz.

For breast phantom design, there is several tissue mimicking methods found in literature [34–36]. The mimicking method based on gelatine-oil is used in this paper because it has stable mechanical properties, low cost and easy fabrication. The dielectric properties of breast tissues can be obtained by varying the concentration of gelatin-oil mixture. The composition of chemicals for breast phantom based on [36] is shown in Tab. 5.

The hemispherical breast phantom is designed with a radius of 70 mm. The phantom consists of a skin layer with the thickness of 2 mm and fat layer with the thickness of 68 mm as shown in Fig. 11. The spherical cancerous cell with the radius of 5 mm is embedded in the fat layer. This cancerous cell is located on the left side of the breast phantom and it is embedded 40 mm away from bottom surface as shown in Fig. 12. The procedure for prototyping a breast phantom is as follows. The first step, the chemical mixture for skin layer is mixed and poured into a 2 mm gap between two hemispherical containers. The first hemisphere is a base with a radius of 70 mm and the second hemisphere is a cover with a radius of 68 mm, leaving a gap of 2 mm between them. The skin layer mixture should be allowed to rest for at least two days to become solid. The cancerous cell is made using similar procedure as the skin layer and it is cut to sphere with a radius of 5 mm. The next step, the fat layer mixture can be poured into the skin layer. The cancerous cell is embedded in desired position within the fat layer. For the completed breast phantom, the solid skin layer is milky white color and flexible, while the solid fat layer is a pale-yellow color and a slightly soft. The fabricated hemisphere breast phantom is depicted in Fig. 12b.

Moreover, the relative permittivity and conductivity of the breast phantom and tumor are measured using dielectric probe and Vector Network Analyzer (VNA). The relative permittivity and conductivity curves of fat and cancer against frequency are illustrated in Fig. 13. It can be observed that the permittivity of cancerous cell is about 5 times larger than breast tissue across frequency range from 1-8 GHz, while conductivity of cancerous cell is 5 times larger than breast tissue at frequency above 6 GHz.

This section presents the measurement of proposed AVAs with breast phantom for breast cancer detection. The methodology for breast cancer detection based on using a single antenna is proposed. In order to confirm the performance of proposed AVAs for breast cancer detection, the four AVA structures i.e., the original AVA, AVA with non-uniform corrugation, AVA with parasitic patch and corrugation and AVA with dielectric lens in [19] are compared. The experimental setup of breast cancer detection system is shown in Fig. 14. The measurement system mainly consists of a VNA (Agilent Technologies E5071C ENA Series), antenna and cancerous breast phantom. Absorbers are placed underneath and enclosed the breast cancer detection system. The antenna is connected to port 1 of VNA to measure the S₁₁. The cancerous breast phantom is placed on the rotation platform and antenna is placed above the cancerous beast phantom with fixed distance. The breast phantom is rotated for 360 degree relative to the stationary antenna to simulate antenna scanning around the breast phantom. In order to clearly distinguish of S₁₁ between cancerous and non-cancerous positions, four locations of antenna are defined with the notation of A, B, C and D, where point A is position of cancerous cell as shown in Fig. 15. The S₁₁ at all antenna positions are recorded. Next, distance between antenna and skin layer (H) is adjusted until the magnitude of S₁₁ between cancerous and non-cancerous position is obviously differed. Therefore, the distance between antenna and skin layer of each antenna configurations is not fixed as the same distance. The measured S₁₁ corresponding the antenna locations i.e., A, B, C and D, when the four AVA structures are used, are shown in Fig. 16. Considering the same scanning radius, it is found that the magnitude of S₁₁ is relatively similar in the absence of cancerous cell. In addition, the magnitude of S₁₁ at cancerous position is rather different from those of non-cancerous positions. According to the condition that if the antenna meets the cancerous cell, the magnitude of S₁₁ is higher than the healthy tissue due to the high difference of electrical properties between cancerous and healthy tissue. It is found from Fig. 16 that all AVA structures can detect the cancerous cell because the difference of S₁₁ between cancerous and non-cancerous positions can be observed. However, the difference of S₁₁ between cancerous and non-cancerous positions (ΔS11=S11,cancer−S11,non - cancer) is very small and difficult to detect the cancerous cell when using the AVA with dielectric lens. This is because the electric field strength in near-field region of AVA with dielectric lens is relatively weak compared to other AVAs. Although the antenna can provide high gain but it cannot confirm that the cancerous cell will be accuracy detected. Therefore, the electric field distribution of antenna should be considered in antenna design. The maximum difference of S₁₁ between cancerous and non-cancerous position at considered frequency for the four AVA structures is summarized in Tab. 6. It can be seen from Tab. 6 that the AVA with parasitic patch has the highest value of ΔS11, follows by, AVA with non-uniform corrugation, original AVA and AVA with dielectric lens in [19]. It means that AVA with parasitic patch can distinguish between cancerous cell and healthy tissue better than other AVAs. Therefore, the parasitic patch proved capable of enhancing radiation in near-field region, which is required for breast cancer detection.

Then, the all S₁₁ at all antenna positions are collected and imported in MATLAB to image processing. The S₁₁ data sets of AVA with parasitic patch is used as an example for microwave imaging. It is known that microwave imaging based on frequency domain signals (i.e., S₁₁) describe the cancerous response at different operating frequencies. Therefore, the S₁₁ data sets are firstly analyzed. According to the phenomenon that the magnitude of the cancerous response becomes stronger when the antenna moves close to the cancerous cell and becomes weak when antenna moves away from cancerous cell. Therefore, the 2D image of the breast phantom can be created using the intensity of preprocessed S₁₁ parameters [13]. The resulting values of S₁₁ data sets at 4.75 GHz is used to create the 2D image of the cancerous breast phantom as shown in Fig. 17. The red region with the high density depicts the presence of the embedded cancerous cell inside the breast. It is clearly seen from image result that the cancerous cell can be detected at the left side of the breast phantom. Therefore, it can be concluded that the proposed AVA can detect the breast cancer while maintaining the compact size, simple structure and low complexity in design.

Two antipodal Vivaldi antennas (AVAs) i.e., AVA with non-uniform cosine-shaped corrugation and AVA with parasitic patch and corrugation for breast cancer detection was designed and analyzed. By using the non-uniform corrugation technique, the gain of original AVA is improved without increasing the antenna size. The advantage of AVA with non-uniform corrugation is its compact size and simple structure compared to other AVAs in literature. In addition, the parasitic patch placed on AVA aperture is very useful for enhancing the radiation in end-fire direction, which is necessary for breast cancer detection. Finally, the proposed AVAs are tested with breast phantom to detect the cancerous cell inside the breast. The measured results can confirm that the proposed antennas are suitable for breast cancer detection.