Single‐slice microwave imaging of breast cancer by reverse time migration

Abstract

Purpose Microwave imaging of breast cancer is considered and a new microwave imaging prototype including the imaging algorithm, the antenna array, and the measurement configuration is presented. The prototype aims to project the geometrical features of the anomalies inside the breast to a single‐slice image at the coronal plane depending on the complex dielectric permittivity variation among the tissues to aid the diagnosis. Methods The imaging prototype uses a solid cylindrical dielectric platform, where a total of 24 optimized Vivaldi antennas are embedded inside to form a uniform circular antenna array. The center of the platform is carved to create a hollow part for placement of the breast and the multistatic, microwave scattering parameters are collected with the antenna array around the hollow center. The dielectric platform further enhances the microwave impedance matching against the breast fat tissue and preserves the vertical polarization during the measurements. In the imaging phase, a computationally efficient inverse electromagnetic scattering method—reverse time migration (RTM)—is considered and adapted in terms of scattering parameters to comply with the actual measurements. Results The prototype system is experimentally tested against tissue‐mimicking breast phantoms with realistic dielectric permittivity profiles. The reconstructed single‐slice images accurately determined the locations and the geometrical extents of the tumor phantoms. These experiments not only verified the microwave imaging prototype but also provided the first experimental results of the imaging algorithm. Conclusions The presented prototype system implementing the RTM method is capable of reconstructing single‐slice, nonanatomical images, where the hotspots correspond to the geometrical projections of the anomalies inside the breast.

Ultra‐wideband imaging with an improved backward projection algorithm for far‐field applications

“…The simulations are also conducted for performance comparison. The evaluation metric used for the comparison is the structural similarity (SSIM) described in Bilgin et al 34…”

“…The simulations are also conducted for performance comparison. The evaluation metric used for the comparison is the structural similarity (SSIM) described in Bilgin et al 34 $$$\mathrm{SSIM}({E}_{1},{E}_{2})=\frac{(2{\mu }_{1}{\mu }_{2}+{c}_{{\rm{a}}})(2{\sigma }_{12}+{c}_{{\rm{b}}})}{\left({\mu }_{1}^{2}+{\mu }_{2}^{2}+{c}_{{\rm{a}}}\right)\left({\sigma }_{1}^{2}+{\sigma }_{2}^{2}+{c}_{{\rm{b}}}\right)},$$$where $$\mu $$ and $$\sigma $$ are the mean and standard deviation of the images respectively, $${\sigma }_{12}$$ is the covariance of $${E}_{1}$$ and $${E}_{2}$$, the constants $${c}_{a}={(0.01L)}^{2}$$ and $${c}_{b}={(0.03L)}^{2}$$, and $$L$$ is the dynamic range of the pixel values. Structural similarity ranges between 0 and 1, and the value of SSIM is equal to 1 when two images are identical.…”

Review and Analysis of Tumour Detection and Image Quality Analysis in Experimental Breast Microwave Sensing

“…a T max is the maximum pixel intensity of the tumour response, T mean is the mean tumour response, C max is the maximum clutter response, C mean is the mean clutter response, F max is the maximum fibroglandular response, σ is the standard deviation of image intensities, r max is the position of the maximum image response, r tum is the known position of the tumour, I(r) is the image intensity at position r, and I true (r) is the true object property at position r. In the definition of the SSIM, µ i and σ i are the average pixel intensity and the standard deviation of the pixel intensities of the image I i and c a , c b is defined as c a = (0.01L) 2 and c b = (0.03L) 2 , with L set to the dynamic range of the intensity values in [176]. b The FWHM may refer to the volume or area of an image corresponding to the voxels/pixels that have intensities greater than 50% of the maximum image intensity, or it may refer to the FWHM along a particular dimension within the image.…”

“…The SSIM is an additional summary measure of image quality; it is a measure of the similarity of two images, first proposed by [203], and is commonly used in medical imaging. The application of the SSIM in BMS [176] has the same limitations as the MSE. It is a summary image accuracy metric, and requires complete a priori knowledge of the ground truth property distribution.…”