Prostate cancer is the most common cancer in men, and for American men, it is the second leading cause of death.

Some prostate cancers may grow slowly and can be monitored over time, while others require immediate treatment. To determine how severe a person’s cancer is, doctors look for abnormalities in biopsy sections on glass slides. But this two-dimensional method is difficult to correctly diagnose marginal cases.

Now, a team led by the University of Washington has developed a new non-destructive method that can image entire 3D biopsies, not just slices. In a proof-of-principle experiment, researchers imaged 300 3D biopsies taken from 50 patients — 6 biopsies per patient — and let the computer use the 3D and 2D results to predict that the patient had aggressive cancer Possibility. The 3D function makes it easier for the computer to identify cases that are more likely to recur within five years.

The team published these results in Cancer Research on December 1.

"We have demonstrated for the first time that the ability to check 100% of biopsies in 3D is more informative and accurate compared to traditional pathology-examining a small part of each biopsy in 2D on a microscope slide," Senior Author Jonathan Liu, Professor of Mechanical Engineering and Bioengineering at the University of Washington. "This is exciting because it is the first of many clinical studies that promise to prove the value of non-destructive 3D pathology for clinical decision-making, such as determining which patients need active treatment or which patient subgroups are against certain drugs The response is the best."

The researchers used prostate specimens from patients who underwent surgery more than 10 years ago, so the team understands the results of each patient and can use this information to train a computer to predict these results. In this study, half of the samples contained more aggressive cancers.

To create a 3D sample, the researchers extracted a "biopsy core" (cylindrical tissue plug) from the surgically removed prostate, and then stained the biopsy core to mimic the typical staining used in 2D methods. The team then used an open light sheet microscope to image each entire biopsy core, which used a piece of light to optically "slice" and image the tissue sample without destroying it.

3D images provide more information than 2D images—especially details about the complex tree-like structure of glands throughout the tissue. These additional functions increase the likelihood that the computer will correctly predict the aggressiveness of cancer.  

Researchers use new artificial intelligence methods, including deep learning image conversion technology, to help manage and interpret the large data sets generated by the project.

"In the past ten years or so, our laboratory has mainly focused on building optical imaging equipment for various clinical applications, including microscopes. However, we have begun to encounter the next major challenge for clinical adoption: how to manage and interpret large amounts of data The collection is obtained from patient specimens," Liu said. "This paper represents the first research in our laboratory, which developed a new computational pipeline to analyze our feature-rich data set. As we continue to improve our imaging technology and computational analysis methods, as well as our For larger-scale clinical research, we hope that we can help change the field of pathology and benefit many types of patients."

The first author of the paper is Weiss Xie, a PhD student in Mechanical Engineering at the University of Wisconsin. The other co-authors of this paper are Robert Serafin, Gan Gao, and Lindsey Barner, all of whom are PhD students in Mechanical Engineering at the University of Washington; Kevin Bishop, PhD student in Bioengineering at the University of Washington; Nicholas Reder, Clinical Department of Laboratory Medicine and Pathology, University of Washington School of Medicine Lecturer; Hongyi Huang, a researcher in mechanical engineering at the University of Washington; Chenyi Mao, a PhD student in the Department of Chemistry at the University of Washington; Nadia Postupna, a research scientist in the Department of Laboratory Medicine and Pathology at the University of Washington School of Medicine; Soyoung Kang, an assistant teaching professor in the Department of Mechanical Engineering at the University of Washington; Han Tsinghua University, undergraduate in bioengineering, University of Washington; Jonathan Wright, professor of urology at the University of Washington School of Medicine; C. Dirk Keene and Lawrence True, both professors in the Department of Laboratory Medicine and Pathology at the University of Washington School of Medicine; Joshua, associate professor of chemistry at the University of Washington Vaughan; Adam Glaser, a senior scientist at the Allen Institute, completed this research as a postdoctoral researcher in mechanical engineering at UW; Can Koyuncu, Pingfu Fu, Andrew Janowczyk and Anant Madabhushi, all at Case Western Reserve University; Patrick of Genentech Leo, who completed this research as a PhD student at Case Western Reserve University; and Sarah Holly of the Canary Foundation.

This research was funded by the Prostate Cancer Research Program of the Department of Defense; National Cancer Institute; National Heart, Lung, and Blood Institute; National Institute of Biomedical Imaging and Bioengineering; National Institute of Mental Health; Virginia State of Excellence Award; National Science Foundation; The Nancy and Buster Alward Foundation; and the Prostate Cancer Foundation Young Researcher Award.

Nicholas Reder, Adam Glaser, Lawrence True and Jonathan Liu are the co-founders and shareholders of Lightspeed Microscopy Inc., a spin-off of UW. The company has obtained a license for the technology used in this article.

For more information, please contact Liu at jonliu@uw.edu.

Prostate cancer risk stratification through non-destructive 3D pathology with deep learning-assisted gland analysis

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of press releases posted to EurekAlert!! Use any information through contributing agencies or through the EurekAlert system.

Sarah McQuate University of Washington smcquate@uw.edu Office: 206-543-2580

Copyright © 2021 American Association for the Advancement of Science (AAAS)

Copyright © 2021 American Association for the Advancement of Science (AAAS)