Pulse-Echo Ultrasound Murine In Vivo Acquisitions

This dataset contains Verasonics Vantage 256 (Verasonics Inc., Kirkland, WA, USA) research scanner acquisitions of murine (rat) livers with their ex vivo measured sound speeds and lipid percentages. We encourage use of this data to validate new techniques, including, but not limited to, sound speed estimators, aberration correction methods, and beamforming algorithms. If using this dataset, please cite the following paper:Telichko, Arsenii V., et al. “Noninvasive Estimation of Local Speed of Sound by Pulse-Echo Ultrasound in a Rat Model of Nonalcoholic Fatty Liver.” Physics in Medicine &amp; Biology, vol. 67, no. 1, IOP Publishing, Jan. 2022, p. 015007. Crossref, doi:10.1088/1361-6560/ac4562.Data Organization• The Rat<code>Experiments_curated</code> &gt; <code>VerasonicsAcq</code> folder contains subfolders of saved Verasonics data from three different transmit sequences for 20 rats. Within each Rat&lt;##&gt; folder, an additional subfolder may be present. Rats 1-8 have acquisitions with the skin of the abdominal region directly exposed (ExposedLiver), rats 9-12 have acquisitions captured using a faster version of the transmit sequences (Fast), and rats 14-16 and 18-20 have an InVivo folder with the same fast transmit sequences used with the rat while alive.• The <code>fat quantification</code> folder contains spreadsheets with ex vivo measured liver sound speeds, fat percentage estimation, and lipidosis (NAFLD) scores for all rats. A separate readme.txt within the folder further describes the file contents.• The <code>processing</code> folder contains code related to data loading or image reconstruction. To load in the data and relevant probe and sequence information, run mainLoader.m. To beamform images, sos_experiments.m has example code and instructions.All Verasonics data in .mat files contains Verasonics MATLAB structs (P, RcvData, Receive, Resource, TW, TX, Trans) that provide detailed information about the acquisition and transducer used.Murine Data DetailsAs described in further detail in Telichko et al., 2022, the scanned rats included 4 (2 male, 2 female) lean and 16 (8 male, 8 female) obese Zucker rats, where the obese rats were fed a high fat diet over the time period of the scans and developed steatosis, mimicking non-alcoholic fatty liver disease (NAFLD) in humans. Every two weeks during the study, up to eight weeks, 4 (2 male, 2 female) obese rats were scanned over the abdomen and subsequently sacrificed for ex vivo liver sound speed measurements. The first 8 rats were initially scanned alive under anesthesia, but the respiration resulted in motion artifacts, resulting in the remaining rats, starting from rat 9, having additional ultrasound scans taken within 20 minutes of euthanization and injection of phosphate-buffered solution (PBS) to prevent blood clotting. The data was acquired using an L12-3v transducer with a full synthetic aperture (FSA) sequence, Hadamard-encoded sequence, and multifocal (MF) transmit sequence with multiple focused-transmit depths; multiple independent locations about the liver were captured for each rat. The ex vivo measurements were taken by placing the whole removed liver on a plane metal reflector within a heated water tank at 37°C. A piston transducer was placed above the liver, and time delays of the echoes from the metal and the liver were captured using an oscilloscope to calculate the sound speed of the liver using the known sound speed of water at 37°C. Each measurement was repeated 6-12 times for different positions of the liver. For rats 15-20, sound speed measurements were also taken using the same procedure for the 4 largest lobes of the liver closest to the abdominal wall. Additional ex vivo analysis was performed to estimate the total lipid content in the liver. Liver samples weighing 2-10 g were homogenized in deionized water and then processed as described in the paper to separate the lipids. The lipid percentage was calculated based on the weight of the isolated lipids and the volume and concentration of the homogenate. Histopathological analysis was performed to determine steatosis grading through cryosectioning and subsequent staining to highlight adipose tissue. Liver sections were evaluated using the modified Brunt system for NAFLD.Verasonics Capture DetailsA center frequency of 7.8 MHz was used for all transmit sequences. Although one set of FSA, Hadamard, and multifocal transmit sequences was used for all scans, rats 9-20 (except for rat 17) had additional fast in vivo acquisitions taken with slightly modified transmit sequences to reduce motion artifacts. The end depth for the latter acquisitions was reduced, and the receive buffer was modified as indicated in the saved Verasonics data structs.The FSA sequence involved transmitting pulses from each individual transducer element and receiving on all available elements. For the original acquisition sequence, a total of 192 * 2 transmit events were used to account for each element on the L12-3v transducer and the two receive subapertures needed at both ends on the multiplexed array to obtain channel data on all elements. The fast transmit sequence used a total of 128 * 3 transmit events, using the transmit and receive subapertures at both ends of the transducer and one in the middle. The Hadamard-encoded sequence used a Hadamard matrix to determine the polarity of pulses sent out across the entire transducer aperture for each transmit event. The original acquisition sequence used a total of 2 * 96 * 2 * 2 transmit events for the two subapertures at the ends of the transducer to cover the entire aperture in two non-overlapping halves on transmit, the positive and negative waveforms, and the two subapertures on the ends of the aperture used during receive. The fast transmit sequence used a total of 128 * 3 transmit events, with separate 128-element Hadamard encoding for the subapertures on both ends of the aperture and in the middle. The multifocal transmit sequence uses a walking aperture focused transmit sequence with multiple focal depths. The original sequence involved three focal depths at approximately 17.7, 29.6, and 41.4 mm, using the two subapertures at the ends of the transducer for a total of 192 * 3 * 2 transmit events. The fast sequence had two focal depths at approximately 15.8 and 29.6 mm using the subapertures at both ends and the center, yielding a total of 192 * 2 * 3 transmit events.<b>Publications Featuring this Data (Non-Exhaustive)</b>Manuscripts:R. Ali, T. Brevett, D. Hyun, L. L. Brickson and J. J. Dahl, "Distributed Aberration Correction Techniques Based on Tomographic Sound Speed Estimates," in IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, vol. 69, no. 5, pp. 1714-1726, May 2022, doi: 10.1109/TUFFC.2022.3162836.R. Ali et al., "Sound Speed Estimation for Distributed Aberration Correction in Laterally Varying Media," in IEEE Transactions on Computational Imaging, vol. 9, pp. 367-382, 2023, doi: 10.1109/TCI.2023.3261507.W. Simson, L. Zhuang, B. N. Frey, S. J. Sanabria, J. J. Dahl, D. Hyun, “Ultrasound Autofocusing: Common Midpoint Phase Error Optimization via Differentiable Beamforming,” arXiv preprint arXiv:2410.03008, 2024.Conference Proceedings:S. J. Sanabria, S. Munot, T. Brevett, A. Telichko and J. Dahl, "Speed-of-Sound Dispersion Estimation from Pulse-Echo Data," 2023 IEEE International Ultrasonics Symposium (IUS), Montreal, QC, Canada, 2023, pp. 1-4, doi: 10.1109/IUS51837.2023.10306727.