“Understanding the brain in all its complexity is impossible for any group to accomplish in isolation.”
-Arthur Toga, Director

We’ve built a diverse team of neurobiologists, mathematicians, and computer scientists, and a worldwide network of collaborators sharing data. Our goal is to increase the pace of discovery in neuroscience by better understanding how the brain works when it’s healthy and what goes wrong in disease.

About LONI

Our facility houses two advanced Magnetic Resonance Imaging scanners for data acquisition: a Magnetom Prisma 3T and a Magnetom Terra 7T.

Learn more

LONI’s onsite data center features state-of-the-art security technology and can store more than four petabytes of brain imaging data.

Learn more

Our Scientific Visualization Group creates elegant maps and animations to illustrate brain structure and function.

Learn more

Latest News

Latest News


Join us in welcoming the newest member of the INI faculty, Ioannis Pappas, PhD, who conducts multimodal studies of patients with brain injury and stroke.

Brain imaging and computational models offer insights that may lead to recovery among patients who are comatose or in a persistent vegetative state. But studying such patients is a challenge. By observing individuals under general anesthesia—a more controlled loss of consciousness—Ioannis Pappas, PhD, can begin to understand what’s happening in the brain and build models to test in patients with brain injuries.

Dr. Pappas, who joined the USC Mark and Mary Stevens Neuroimaging and Informatics Institute (INI) as an assistant professor of research neurology in July, conducted that research during his doctoral studies in clinical neuroscience at the University of Cambridge in England. He used functional magnetic resonance imaging (MRI), as well as structural methods including diffusion tensor imaging (DTI), to study how neural connectivity changes in patients who are unconscious and whether those changes can predict recovery over the long term.

During his postdoctoral studies at the University of California Berkeley’s Helen Wills Neuroscience Institute, Dr. Pappas shifted his research focus to stroke recovery, with funding from the U.S. Veterans Administration. Using fMRI, DTI, blood flow imaging, and spectroscopy, he collected longitudinal data of veterans recovering from a stroke in an effort to understand the relationship between behavioral and neurological changes over time.

Because of the nature of that data—stroke patients have lesions in the brain that differ from one person to the next—a big part of Dr. Pappas’ research involves developing methods to standardize and facilitate the processing of MR images. Collecting data with multiple methods, such as combining standard structural brain images with arterial spin labeling, also helps create a more complete picture of stroke recovery. For example, Dr. Pappas found that the amount of residual blood flow in chronic stroke patients who have aphasia relates to their scores on language tests.

At the INI, Dr. Pappas will continue using multimodal methods to characterize clinical data. He is excited to draw on the institute’s vast collections of data and collaborate with its multidisciplinary team of researchers that includes engineers, cognitive neuroscientists, imaging experts, and more.

“We cannot have just one approach—we need multiple ways of collecting, analyzing, and interpreting neuroimaging data,” he said. “For that reason, interdisciplinary research is the way forward for tackling big programs of pathology and disease in the brain.”

Read more about Dr. Pappas’ findings on sedation and disorders of consciousness:

Brain network disintegration during sedation is mediated by the complexity of sparsely connected regions in NeuroImage

Consciousness-specific dynamic interactions of brain integration and functional diversity in Nature Communications

Read more about his research on methods development for studying stroke recovery:

Improved normalization of lesioned brains via cohort-specific templates in Human Brain Mapping

Poster presentation from the 2020 Society for the Neurobiology of Language conference:


The USC+Amazon Center on Secure and Trusted Machine Learning, established in January 2021 to support fundamental research and development of new approaches to machine learning (ML) privacy, security, and trustworthiness, today announced it has selected five research projects for 2021-2022.


A novel non-invasive neuroimaging technique can detect early-stage dysfunction of the blood–brain barrier (BBB) associated with small vessel disease (SVD), according to new research published in Alzheimer’s & Dementia. Cerebral SVD is the most common cause of vascular cognitive impairment, with many cases leading to dementia.


Join us in welcoming the newest member of the INI faculty, Leon Aksman, PhD, who is applying insights from his background in engineering, neuroimaging, and financial modeling to examine the pathology of Alzheimer’s disease.

Dr. Aksman’s career path has been a bit unorthodox. After completing bachelor’s and master’s degrees in engineering, he worked in finance for several years. When he returned to academia to pursue a doctorate in neuroimaging, he applied his knowledge of quantitative modeling to one of the field’s most pressing questions: What causes Alzheimer’s disease (AD) and how does it progress over time?

After completing his doctorate at Kings College in London and postdoctoral training at University College London, Dr. Aksman joined the faculty at the USC Mark and Mary Stevens Neuroimaging and Informatics Institute last fall. Here, he is continuing his work applying statistics and machine learning to multimodal neuroimaging data, as well as cognitive and biofluid measures.

“Because there aren’t yet effective treatments for Alzheimer’s, what really drives me is to understand the disease better,” he said. “We can develop better treatments when we know more about the sequence and heterogeneity of pathologies in AD.”

For example, he used a mathematical model to analyze neuroimaging data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), including subjects in early, middle, and late stages of developing amyloid and tau pathologies. His model revealed two subtypes: one where amyloid pathology develops first and one where tau pathology develops first. This finding challenges the prevailing view among AD researchers that amyloid pathology always accumulates first.

“We got this result in a very data-driven way, but now we have to confirm it using both longitudinal and postmortem data,” Dr. Aksman said.

This research can help answer questions about whether AD manifests differently in certain patients, which can ultimately inform the development of targeted treatments.

Dr. Aksman is also analyzing data from Longitudinal Aging Study in India Diagnostic Assessment of Dementia (LASI-DAD), which includes data on adults in India who have developed dementia, with measures of literacy and urban/rural habitation. He will explore whether AD progression can be classified into various subtypes among this population.

“Most Western studies have not captured the full spectrum of socioeconomic variation in cognitive decline, which may mean that we have yet to explore the full heterogeneity of dementia,” he said.