C14-b - Ex Vivo Ultrasound Imaging and 3D Volume Reconstruction of Mouse Organs with Image Filtering

Saturday, October 11, 2025

10:00 AM - 12:00 PM PDT

Location: Exhibit Hall F, G & H

Presenting Author(s)

Jiho Yoon

Intern
University of Oklahoma
Edmond, Oklahoma, United States

Co-Author(s)

Adree Bhattacharjee, n/a

Graduate Research Assistant
University of Oklahoma
Norman, Oklahoma, United States
ST

Sydney Turner

Graduate Research Assistant
OU, United States

Primary Investigator(s)

SY

Sangpil Yoon

Assistant Professor
University of Oklahoma, United States

Introduction: : Ultrasound imaging is widely used for its real-time, non-invasive nature, along with its greater
capabilities in capturing internal structures compared to optical imaging. However, it is still often limited to two-
dimensional (2D) views, making it challenging to capture the full structural details of an organ. Ex vivo ultrasound
imaging combined with 3D reconstruction offers a practical solution, allowing detailed structural information to be
obtained in a controlled setting which plays a key role in understanding organ structure and guiding biomedical
research. However, ex vivo ultrasound data often include noise and background artifacts, which can produce
misleading findings if not properly addressed. Reliable reconstruction requires not only careful image acquisition but
also effective filtering and segmentation to clearly separate the organ from the surroundings. In this project, we
aimed to develop an effective method for creating accurate 3D ultrasound structure of mouse organs. By using
filtering steps to clean up the data and remove background noise, we can produce reliable 3D reconstructions that
make it easier to visualize the organ and analyze its structure.

Materials and

Methods: : Fresh mouse organs (spleen and kidney) were fixed in 4% paraformaldehyde overnight.
For embedding, 2 g of agarose powder was dissolved in 200 mL of Type 1 water. Half the solution was poured into a
plastic dish, solidified at 4°C for 10 minutes, and organs were positioned on top. The remaining agarose was gently
poured to cover the organs and solidified again at 4°C [1] . Ex vivo ultrasound imaging was performed using a
Vantage 256 system (Verasonics, Seattle, WA) with a 128-element L22-16vX linear array transducer (15.6 MHz
center frequency) mounted on a computer-controlled 3D translation stage (ILS150CC, Newport Corp., Irvine, CA).
The transducer was translated in 0.02 mm steps across the organ to acquire image slices: 50 for kidney and 40 for
spleen. RF and IQ data were acquired and reconstructed using the Vantage system. Post-processing of IQ data was
done in Matlab. Each 2D image was log-compressed to enhance contrast, and Otsu’s thresholding was used to
segment the organ from background. Small isolated bright regions were removed using area-based filtering
“bwareaopen” to reduce noise. Internal holes were filled using “imfill” to produce continuous organ masks. These
binary masks were applied to the log-compressed images to isolate organ signals. Finally, the masked 2D slices were
stacked to form a 3D volume, which was visualized using Matlab’s volumeViewer toolbox.

Results, Conclusions, and Discussions:: The original log-compressed spleen and kidney images (Figs. 1A and 1F)
went through Otsu thresholding for these images to be segmented from the background (Figs. 1B and 1G). The
average applied threshold value was 127 for the spleen images and 128 for the kidney images. Then, we applied a
morphological opening operation to remove small bright objects and fill internal holes within the organ to ensure the
image appeared as one continuous structure (Figs. 1C and 1H). Next, a masked image (Figs. 1D and 1I) was created
by combining the original image (Figs. 1A and 1F) with the filtered binary mask (Figs. 1C and 1H). This was done
by applying the binary mask to the original image, preserving only the pixel values within the segmented organ
region, and cleaning up the background. Next, as seen from figures 1E and 1J, all the masked 2D image slices were
stacked to form a 3D volume, 40 slices for the spleen and 50 slides for the kidney. This 3D volume was used to
create the final reconstructions, visualized from the coronal plane (Figs. 1a and 1d), sagittal plane (Figs. 1b and 1e),
and from the top (Figs. 1c and 1f) to highlight the organs’ internal structures and surface features. This study presents
a simple and effective method for 3D ultrasound imaging of ex vivo mouse organs. The final 3D reconstructions
show the shape and structure of the organs in detail. However, using fixed-interval slices may miss some fine details,
and Otsu thresholding may not capture the low-contrast regions. Overall, this approach is easy to apply and can help
with analyzing organ anatomy in 3D using ultrasound imaging. Future studies could focus on improvements by using
finer slices or more advanced segmentation to enhance resolution and detail.

Acknowledgements and/or References (Optional): : Acknowledgments:
This work was supported in part by the National Institute of Health (NIH) grant Nos. GM120493
and GM152704, the National Science Foundation (NSF) grant No. CBET1943852 and 2304485.

References:
1. Kim, S., S. Zhang, and S. Yoon, Multiplexed Ultrasound Imaging Using Spectral Analysis on Gas Vesicles.
Adv Healthc Mater, 2022. 11(17): p. e2200568.