These images show regions of glioblastoma tumors in women (top) and men (bottom) where the researchers’ AI models predict relatively higher risk and lower risk characteristics. High risk areas are shown in red and low risk areas in blue. Credit: Tiwari Lab/UW–Madison
For years, cancer researchers have noticed that more men than women have a deadly form of brain cancer called glioblastoma. They also found that these tumors are often more aggressive in men. But it has proven difficult to identify features that could help doctors predict which tumors are likely to grow most quickly. Researchers at the University of Wisconsin-Madison are turning to artificial intelligence to reveal these risk factors and their gender differences.
Professor of radiology and biomedical engineering Pallavi Tiwari and colleagues published their first findings in the journal Scientific advanceshinting at the promise of AI to improve medical care for cancer patients.
“There is a ton of data collected during a cancer patient’s journey,” says Tiwari, who is also affiliated with the medical physics department. “Right now, unfortunately, it’s generally studied in silos, and that’s where AI has huge potential.”
Few researchers understand this potential better than Tiwari. Arriving at UW-Madison in 2022 to help lead the university’s new AI initiative in medical imaging, Tiwari co-directs the Imaging and Radiation Sciences program at the Carbone Cancer Center. His research harnesses the computing power of AI models to probe large volumes of medical images and find patterns that could help oncologists and their patients make more informed decisions.
“We want to address all the challenges related to the cancer patient journey, from diagnosis and prognosis to evaluating response to treatment,” explains Tiwari.
In this case, Tiwari and former graduate student Ruchika Verma turned to digital images of pathological slides (thin slices of tumor samples) in search of patterns to predict how quickly a tumor might grow and therefore the survival time of a patient.
Glioblastoma is one of the most aggressive forms of cancer, with a median survival of 15 months after diagnosis.
“Patients often don’t live long after diagnosis,” says Tiwari. “But a big challenge is prognosis: identifying how long patients will actually live and what their outcomes are likely to be. This is important because outcomes ultimately determine the treatments they receive and their quality of life after diagnosis .”
To address this challenge, Tiwari and Verma built an AI model capable of identifying even subtle patterns in pathology slides that might never be apparent to the naked eye. Using data from more than 250 studies of glioblastoma patients, they trained the model to recognize unique characteristics of tumors, such as the abundance of certain cell types and the degree to which they invade surrounding healthy tissue.
Additionally, they trained the model to identify any trends between these characteristics and patients’ survival time, while taking their gender into account.
In doing so, they developed an AI model capable of identifying risk factors for more aggressive tumors strongly associated with each gender. In women, higher risk features included tumors that infiltrated healthy tissue. In men, the presence of certain cells surrounding dying tissue (called pseudopalissading cells) was associated with more aggressive tumors.
The model also identified tumor characteristics that appear to translate into worse prognoses for both men and women.
The study could contribute to more individualized care for patients with glioblastoma.
“By discovering these unique patterns, we hope to inspire new avenues for personalized treatment and encourage further research into the underlying biological differences observed in these tumors,” Verma said.
Tiwari and colleagues are doing similar work using MRI data and have begun using AI to analyze pancreatic and breast cancers with the goal of improving patient outcomes.
In addition to his research, Tiwari is helping shape the university’s RISE-AI and RISE-THRIVE initiatives, which make UW-Madison a leader in interdisciplinary research in artificial intelligence and human health, respectively.
“The UW has rich and diverse expertise across our engineering and medical campuses,” says Tiwari, “and with RISE initiatives, we are well positioned to be at the forefront of research translation on AI in clinical care.
More information:
Ruchika Verma et al, Sexually dimorphic computational histopathological signatures prognostic for overall survival in high-grade gliomas via deep learning, Scientific advances (2024). DOI: 10.1126/sciadv.adi0302
Provided by University of Wisconsin-Madison
Quote: AI model shows potential for identifying gender-specific risks associated with brain tumors (October 7, 2024) retrieved October 7, 2024 from
This document is subject to copyright. Except for fair use for private study or research purposes, no part may be reproduced without written permission. The content is provided for informational purposes only.
