Scope and Objectives

We propose to build a low cost, super-resolution, portable ultrasound imaging device designed specifically to address healthcare issues in rural India and other developing countries. The system shall reconstruct images using lesser number of measurements than the standard scanners. The reconstruction shall be done using our specialized sparse recovery based techniques that offer super-resolution capabilities for layered phantoms. For imaging organs, the quality of the reconstruction would be comparable to that obtained using the full set of measurements.

Deliverables

A prototype of a low-cost portable ultrasound scanner consisting of a transducer and a handheld device or a laptop containing software for image reconstruction. A detailed technical document explaining the various steps involved, from data acquisition to image reconstruction. Publications in premier journals and top-tier conferences. Students graduating with a sound knowledge of sub-Nyquist methodologies and hands-on experience in working with ultrasound systems.

Videos

Scientific Output

Sub-Nyquist super-resolution ultrasound imaging system with a single-element transducer. We proposed a statistical beamformer (called ST - soft-thresholding) which considers the signal of interest and noise interference as a random variable. Moreover, we also optimize over the apodization weights (called IR - Iteratively-reweighted), which result in a high-resolution image (as seen in the attached figure). It results in improved lateral resolution compared to the conventional Delay-and-Sum beamformer and the current state-of-the-art beamformer called iMAP. We also attach an image measuring axial and lateral resolution for various beamforming algorithms.

Results and outcome till date

We demonstrated a proof of concept low-rate ultrasound system, equipped with a wobbler transducer, by scanning a goat's heart and kidney. For this experiment, we used L and T Sonata LX ultrasound scanner. We used Sum-of-Sincs (SoS) kernel-based sub Nyquist sampling technique to reconstruct each scan line. Since the system used a wobbler, the signal had low SNR. To counter that, we propose to use an array transducer instead of a single element probe like a wobbler. Analog vs Digital domain subsampling: After several rounds of technical discussions with hardware experts from the Department of Electronic Systems Engineering (DESE) at IISc, we chose to go with digital domain sub Nyquist sampling due to hurdles encountered in implementing an analog-domain sub-Nyquist system in hardware. Design considerations for the multichannel (64/128 array) acquisition system and per-channel simulation of the FFT module in Xilinx Vivado. Since the system acquires only real signals, we could embed two real signals (two channels) in the real and imaginary parts of the FFT computations. Demonstration of super-resolution using sparse signal recovery techniques (based on l-1 minimization) on data generated using Field-II simulation software. The signal was considered to be sparse in the time-domain and the measurements are in the Fourier domain. Development of a maximum a posteriori estimation based novel beamforming algorithm taking noise statistics into account for contrast enhancement. Optimal apodization window design which leads to high resolution ultrasound images. PCB layout has been completed. Rx and Tx beamforming algorithms have been simulated for implementing in FPGA. Super-resolution limits being tested on teflon samples of varying thicknesses.

Societal benefit and impact anticipated

With the low-cost portable ultrasound scanner, we could potentially have mobile diagnostic centers around India. This results in immediate assistance in remote parts of India (where a standard ultrasound scan would usually take weeks). Also, with the high resolution images, one could train Machine Learning algorithms on it to identify certain anomalies in the image, thereby assisting doctors in the diagnosis.

Next steps

We have started with the hardware building process, and soon we should have an early-level prototype for acquiring raw data on phantoms. With that, we can fine-tune our algorithms and improve them further to adapt to the hardware design. To acquire fewer measurements, we are considering sampling strategies wherein few ADC channels would be required. We are also considering sequential panning of an aperture to acquire the sectoral image than using the whole aperture of the transducer. The next step would be to send the PCB for fabrication, and to implement and test the beamforming algorithms in FPGA.

Publications and reports

A. G. Mahurkar, P. K. Pokala, C. S. Seelamantula, ITERATIVELY-REWEIGHTED BEAMFORMING FOR HIGH-RESOLUTION ULTRASOUND IMAGING, IEEE Intl. Symposium on Biomedical Imaging 2019.

Patents

C. S. Seelamantula, S. Nagesh, S. Mulleti, S. Rudresh, D. Mazumder, R. M. Vasu, "Lowrate super-resolution ultrasound imaging system," Indian patent 6178/CHE/2015, filed on November 17, 2015. We have received the first examination report on the patent on August 8, 2018 and we are preparing a point-by-point response to the questions raised.

Scholars and Project Staff

M. Manjula Administrative Assistant March 2017 - July 2019 Neeraj Pal Project Assistant September 2017 - December 2017 Pavan Kulkarni Project Associate October 2017 - July 2019 Suprosanna Shit Project Associate July 2017 - November 2017 Swati Hegde Project Associate November 2017 - January 2018 Amol G Mahurkar Project Assistant February 2018 - December 2018 Rashmi S. Project Assistant March 2017 - April 2017 Raghavendra D.R Project Associate April 2018 - January 2019 Aditya K. Project Assistant May 2018 - November 2018 Saurav K. Shastri Project Assistant June 2018 - June 2019 Dwiref Snehal OZA Project Assistant June 2018 - December 2018 Satya Prakash Project Assistant June 2018 - November 2018 Venod M. System Administrator July 2018 - October 2018 Jishnu Sadasivan Research Associate September 2018 - February 2019 Sabbella Sai Hemanth Reddy Project Assistant October 2018 - March 2019

Challenges faced

The bandwidth of the raw data is posing a great hurdle.

Other information

We're also teaming up with professional ultrasound manufacturers (Surabhi systems, Coimbatore).