By Bill Bruton

Titled “Enhancing Biomedical Data Validity With Standardized Segmentation Finite Element Analysis,” it was published June 14.

“Finite element analysis is a powerful computational technique in which 3D models are used to collect biomechanical data. These data are essential for medical and biological research, as well as for prosthetics design,” Wysocki says. “Unfortunately, little information existed about how this research technique was influenced by the CT data segmentation process (a required data processing step).”

The current study tested multiple CT data segmentation approaches through a total of 516 finite element models. The results show that merely a 5 percent change in the CT data segmentation process significantly alters multiple biomechanical measurements.

“This discovery is a game-changer because it influences medical and biological research that uses the technique of finite element analysis,” says Wysocki, lead author on the paper. “The hope is that the methodological insights will lead to biomechanical data that are more accurate and research conclusions that have greater validity.”

Scott T. Doyle, PhD, associate professor of pathology and anatomical sciences, is Wysocki’s mentor and the senior author on the paper.

“The publication of this work signals the importance of computational methods for understanding anatomy. When people think of “quantitative biology” they often think of bioinformatics or electronic health records, but the reality is that computer models, machine learning and artificial intelligence are having an impact in all areas of biology and medicine,” Doyle says. “Our CCBAP graduate program emphasizes the role that these tools have across all length scales and modalities of biomedical practice.”

“Further, we aim to ‘de-mystify’ advanced artificial intelligence for physicians and biomedical students. We hope our students learn to code, design machine learning pipelines and test hypotheses with quantitative modeling tools the same way they learn to run a polymerase chain reaction and measure spinal cord injury in animal models,” Doyle adds. “I hope that the publication of this work leads to more interest in our program in the wider community, and that students look to the University at Buffalo if they are interested in using modern tools to understand the human body, form and function.”

“Congratulations to Matthew and Dr. Doyle. The publication of this paper shines a spotlight on our CCBAP program, which, while relatively new, is already making its mark in the research community,” says Allison Brashear, MD, MBA, UB’s vice president for health sciences and dean of the Jacobs School of Medicine and Biomedical Sciences.