Multi-modal Ultrasound Imaging for Breast Cancer Detection

Abstract

This work describes preliminary results of a two-modality imaging system aimed at the early detection of breast cancer. The first technique is based on compounding conventional echographic images taken at regular angular intervals around the imaged breast. The other modality obtains tomographic images of propagation velocity using the same circular geometry. For this study, a lowcost prototype has been built. It is based on a pair of opposed 128-element, 3.2 MHz array transducers that are mechanically moved around tissue mimicking phantoms. Compounded images around 360º provide improved resolution, clutter reduction, artifact suppression and reinforce the visualization of internal structures. However, refraction at the skin interface must be corrected for an accurate image compounding process. This is achieved by estimation of the interface geometry followed by computing the internal ray paths. On the other hand, sound velocity tomographic images from time of flight projections have been also obtained. Two reconstruction methods, Filtered Back Projection (FBP) and 2D Ordered Subset Expectation Maximization (2D OSEM), were used as a first attempt towards tomographic reconstruction. These methods yield useable images in short computational times that can be considered as initial estimates in subsequent more complex methods of ultrasound image reconstruction. These images may be effective to differentiate malignant and benign masses and are very promising for breast cancer screening.

Time domain reconstruction of sound speed and attenuation in ultrasound computed tomography using full wave inversion

“…After the forward simulation, random Gaussian noise was added to the recorded data to give a signal-to-noise ratio (SNR) of 50 dB. This is similar to the SNR achievable with our experimental USCT prototype (Medina-Vald es et al, 2015;Perez-Liva et al, 2015). Due to the presence of noise in the data, the TV-based edge-preserving regularization explained in Sec.…”

“…The USCT geometry was a ring array of radius 54 mm with N ¼ 200 uniformly distributed transducers. This circular configuration is employed in several prototypes of USCT scanner for breast cancer detection (Duric et al, 2009;Li et al, 2008b;Andr e et al, 2013;Medina-Vald es et al, 2015) and ensures a constant spatial resolution and good angular coverage. The simulations were conducted on a 128 mm  128 mm grid represented by 256  256 grid points with a 20 grid point perfectly matched layer positioned outside the domain (Treeby and Cox, 2010a).…”

“…Ultrasound computed tomography (USCT) is a noninvasive radiation-free medical imaging technique with promising capabilities to resolve soft tissue structures in the body. In this technique, the tissue is imaged using a set of ultrasound transducers that are located surrounding a region of interest (Ruiter et al, 2012;Medina-Vald es et al, 2015). Each of the transducers acts as the source of an ultrasound field in turn, while the rest of the transducers record the temporal ultrasound signals that carry information about the different structures encountered during propagation.…”

Improved misfit function for attenuation and speed reconstruction in ultrasound computed tomography

“…After the forward simulation, random Gaussian noise was added to the recorded data to give a signal-to-noise ratio (SNR) of 40 dB. This is similar to the SNR achievable with our experimental USCT prototype 22 .…”

Convolutional Neural Network untuk Pendeteksian Patah Tulang Femur pada Citra Ultrasonik B–Mode

“…Citra B-mode memiliki kualitas yang rendah dan dipenuhi speckle pada seluruh bidang citranya. Pada beberapa kasus, speckle ini sangat membantu dalam memisahkan organ/jaringan lunak yag satu dengan yang lain, seperti pada pendeteksian janin [10], segmentasi pembuluh darah [11], [12], pendeteksian penyakit jantung [13], mata [14], kanker payudara [15], [16], dan penyakit prostat [17] maupun jaringan lain dalam tubuh [18]˗ [20].…”