MATRIX Fall Seminar Series – Dr. Mark Goldberg

Mapping the Stroke “RE-Connectome”

Mark Goldberg, MD

Professor of Neurology

Associate Vice President for Strategic Research Initiatives

UT Health San Antonio

Friday, September 30, 2022

11 AM CST – 12 PM CST

Join from the webinar link

https://utsa.webex.com/utsa/j.php?MTID=m732b1ddde75e79c34bcb0177fe4e1d57

How can the brain restore its function when parts are irreversibly injured?  Stroke, or “brain attack” occurs when an artery supplying one part of the brain becomes blocked by atherosclerosis or blood clot.  The blockage may resolve spontaneously, or hospital stroke specialists may use drugs or catheters to remove the clot.  But if the artery remains blocked for several hours the tissue will be permanently damaged, and the stroke patient will have deficits that reflect the brain region, such as loss of strength, sensation, vision, speech or memory.   Surprisingly, most stroke patients have at least partial recovery of function over the next 90 days, even though the injured neurons and other brain cells do not regenerate.  This occurs because of neural plasticity, the ability of the nervous system to change its activity in response to intrinsic or extrinsic stimuli by reorganizing its structure, functions, or connections.

Our lab studies a unique form of post-stroke plasticity in a mouse model.   After a targeted injury of the primary motor cortex on one hemisphere, the axons from cortex to spinal cord (corticospinal tract) degenerate and muscle function is impaired.   Within 2-4 weeks, axons from the opposite – uninjured –  corticospinal tract send new collateral sprouts across the spinal cord, where they form synapses that replace the ones lost after stroke.  This is a remarkable form of plasticity because the neurons responsible for repair are far from the location of injury.  Neurons and supporting glial cells activate molecular programs of axon growth that do not normally occur in the adult nervous system.

We’ve developed a microscopy and image analysis pipeline to generate large scale 3D data of the full set of brain and spinal connections that change after injury (which we term the “RE-Connectome”).  Viral vectors and transgenic mice express fluorescent proteins in axons and presynaptic terminals.   Mouse brains and spinal cords are removed and fixed after stroke and imaged using a serial two-photon tomography microscope.  The pipeline for whole brain image analysis includes supervised machine learning (pixel-wise random forest models via the “ilastik” software package) followed by registration to a standardized 3-D atlas of the adult mouse brain or spinal cord.  In this talk, we’ll present some conclusions from our reconnectome data, and review challenges with image segmentation.   This model of nervous system plasticity may provide approaches to promoting resilience in human stroke or in neuromorphic networks.

LINK

MATRIX Fall Seminar Series – Dr. Sek Chai

Dynamically Throttleable Neural Networks

Sek Chai, PhD

Chief Technology Officer and Co-Founder

Latent AI

Friday, October 7 2022

11 AM CST – 12 PM CST

Join from the webinar link

https://utsa.webex.com/utsa/j.php?MTID=m432873fd789a104abcb6de4cd4e6bd57

Deep learning model inferences are typically processed with statically defined allocation of compute and memory resources. However, real-world conditions are often varying and thereby, a static runtime inference is neither optimal from an accuracy or efficiency perspective. The problem lies in the current approaches that build model runtime that occupy a single point in the trade-space between performance and resource use. In this presentation, we present a new way to run your neural network dynamically to minimize the working footprint for both memory and compute horsepower. Such an approach enables a runtime that is highly agile, can self-regulate, and could afford new benefits from a robustness perspective.

LINK

MATRIX Fall Seminar Series – Dr. John Basl

An Ethics Ecosystem for AI and Big Data

John Basl

Northeastern University

Friday, October 14, 2022

11 AM – 12 PM CST

https://utsa.webex.com/utsa/j.php?MTID=mbb2f3f5477b167e739ab94fd44f39f04

Realizing the promise of AI while mitigating against ethical pitfalls requires an ethics ecosystem, an ecosystem that includes robust, interdisciplinary research programs, curricula, and training programs, informed regulatory and oversight infrastructure, broad stakeholder engagement, shared conceptual and linguistic space for overcoming challenges, and AI and data ethicists that can translate across disciplinary divides. This talk motivates the need for such an ecosystem and highlights some work attempting to contribute to it.

MATRIX Fall Seminar Series – Dr. Kal Clark

Eye-AI: A New Approach to Quality, Labeling, and Vetting in Radiology

Dr. Kal Clark

University of Texas Health San Antonio

10/21/2022

11AM – 12PM CST

Abstract: The predominant source of medical errors in radiology is surprisingly not deficient medical knowledge. Rather, errors primarily stem from the methods radiologists use to visually inspect the image, referred to as “perceptual errors.” Perception-related errors are estimated to reach 33% on abnormal chest radiology exams. To mitigate this problem, radiologists employ personalized and high-dimensional visual search strategies, otherwise known as search patterns. Quantitative descriptions of these search patterns via direct observation or self-report are not only unreliable, but also interfere with quality improvement interventions, which therefore negatively impact patient care.  Through the use of eye-tracking technology and feature extraction we can employ novel techniques to improve health care quality, passively label datasets, and vet the next cohort of medical practitioners.

MATRIX Fall Seminar Series – Dr. Soheil Kolouri

Lifelong Learning Machines – Approaches and Challenges

Soheil Kolouri

Vanderbilt University

Friday, October 28, 2022

11 AM – 12 PM CST

WebEx: https://utsa.webex.com/utsa/j.php?MTID=m0d75dea1a1670d0f72425cd9d59073ac

Abstract: The next wave of AI demands new machine learning frameworks that can continually learn and adapt to the stream of nonstationary multimodal information. This challenge is called continual, lifelong, or incremental learning in the ML community. Despite their phenomenal performance in various applications, deep neural networks are prone to forgetting their previously learned information upon learning new ones. This phenomenon is called “catastrophic forgetting” and is deeply rooted in the stability-plasticity dilemma. Overcoming catastrophic forgetting in deep neural networks has become an active field of research. In this talk, I will review some of our recent works on overcoming catastrophic forgetting in continual learning, from novel regularization-based approaches to sparsity and heterogeneous dropout. I conclude my talk by discussing our ongoing research on continual learning and some of the open problems and challenges.

Distinguished Speaker Series – Dr. John Gustafson

MATRIX Fall Seminar Series – Dr. Ananth Grama

Computational Methods for Analysis of Functional Brain Images

Dr. Ananth Grama

Purdue University

11/4/2022

11AM – 12PM CST

WebEx: https://utsa.webex.com/utsa/j.php?MTID=m0dd130eaf6ddc33055fb62e2e1c06923

Abstract:  Rapid advances in neuroimaging have resulted in large repositories of images with high temporal and spatial resolutions. This has motivated complex connectomic analyses aimed at understanding representation and processing of stimuli . Such analyses require novel computational tools that significantly extend the state-of-the-art in machine learning and data science. These studies have far-reaching implications for the fields of precision psychiatry, behavioural analysis, and neurodegeneration, in addition to applications in AR/VR and advanced human interfaces.

In this talk, I will discuss recent work in our group on understanding neuronal response to naturalistic visual stimulus — I present two methods based on archetypal analysis and deep learning to: (a) find interpretable representations of fMRI response, (b) predict objects in visual frames, and (c) reconstruct visual inputs. I will also briefly describe computational methods to characterize individual-level uniqueness of functional connectomes. I will present three methods based on matrix sampling, graph alignment, and linear algebraic techniques to accurately identify the identity of individuals, and the nature of the cognitive task being performed.  I will conclude my talk by highlighting the immense significance and challenges posed by problems in the field of neuroimage analyses.