The ME SPUR Experience: Smith models the human abdomen for ultrasound simulation
Mechanical engineering undergraduate researcher, Ryan Smith.
The ME SPUR Program, modeled after CU Summer Program for Undergraduate Research, enabled undergraduate students to work with mechanical engineering faculty during summer 2020 on research that could be conducted remotely. As a participant, Ryan Smith worked with Assistant Professor Nick Bottenus to use medical image data to develop 3D finite element models of the abdominal wall and perform various compressions to mimic clinical practice. His summer research project was titled, Modeling the Human Abdomen for Ultrasound Simulation.
Smith will be finishing his final semester in mechanical engineering at 兔子先生传媒文化作品 this fall. His interests include numerical methods/analysis, computer modeling and fluid mechanics. His insights below provide a window into his research experience with ME SPUR.
Describe your summer research.
Ultrasound is commonly used to image structures within the abdomen; however, these images are often degraded by acoustic clutter such as aberration and reverberation. Interestingly, it is observed that compressing the ultrasound transducer into the abdomen improves image quality. However, it is not fully understood why. Simulations may help to better understand this effect. The goal of this project is to predict the deformation of the abdomen under compression using FEA. These results can then be used in Fullwave, a tool for modeling nonlinear propagation and multiple scattering of ultrasound through heterogeneous tissues.
Top: Cross section of an abdominal wall.
Bottom: The results of the finite element analysis are used to construct an image of the compressed abdomen.
What was it like working remotely?
This project was well suited for remote work. I was able to interact with my research supervisor, Professor Bottenus during weekly meetings which helped to guide my efforts and at certain points, received feedback and improvements on my code. Moreover, I received some helpful tips on the FE part of the problem from Professor Mark Palmeri at Duke and Boyuan Liu, a member of Professor Maureen Lynch鈥檚 lab.
What challenges did you encounter and work through as part of your project?
Converting a 3D medical image into a mesh is no simple task. Before coming across iso2mesh, I had multiple attempts at this problem. At first, I tried constructing a surface mesh for the outer shape of the abdomen, generating an internal mesh based on this surface mesh, and then assigning materials to elements in the mesh based on the 3D image. The problem with this method is that the internal boundaries separating the various tissue domains are not modeled well, as the mesh was generated with no knowledge of the internal structures. In an attempt to improve upon this, I developed two additional methods, each with its own issues before finding iso2mesh.
What about this project was rewarding?
For me, the most rewarding part of the project was writing the code to map between the original and deformed images. I was able to use my knowledge of shape functions which I learned about in Professor Debanjan Mukherjee鈥檚 CFD class, to map coordinates between the undeformed and deformed meshes. Moreover, this part of the project was a lesson in the computational expense of code. Originally, using a test mesh, the mapping took over eight hours to complete. With the aid of Professor Bottenus, and by making some modifications to the algorithm myself, this time was reduced to around three hours.
What advice would you share with other students considering getting involved in research?
Learn how to find and use manuals/documentation for software you are unfamiliar with. Most software will have these sorts of resources if you look. Moreover, create the same resources for anything you come up with so that others can easily learn and follow what you have done.