Jannah Theme License is not validated, Go to the theme options page to validate the license, You need a single license for each domain name.
Health

AI-enhanced imaging: probing the brain’s visual processing


Summary: Researchers used AI to select and generate images to study the brain’s visual processing. Functional MRI (fMRI) recorded increased brain activity in response to these images, outperforming control images.

The approach made it possible to adapt visual patterns to individual responses, improving the study of how the brain responds to visual stimuli. This method, offering an impartial and systematic vision of visual processing, could revolutionize neuroscience and therapeutic approaches.

Highlights:

  1. The curated, AI-generated images were used to systematically study the brain’s visual processing, producing more significant activation in targeted areas compared to control images.
  2. Personalized AI models have been shown to improve the brain’s response to visual stimuli, demonstrating the potential for individualized neurological studies.
  3. The research opens avenues to study other sensory systems and explore therapeutic applications, such as modifying brain connectivity for the treatment of mental health.

Source: Weill Cornell University

Researchers from Weill Cornell Medicine, Cornell Tech and Cornell’s Ithaca campus have demonstrated the use of AI-curated natural images and AI-generated synthetic images as neuroscience tools to probe visual processing areas of the brain.

The goal is to apply a data-driven approach to understanding how vision is organized while potentially removing bias that can arise when examining responses to a more limited set of images selected by researchers.

In the study, published October 23 in Communication biologythe researchers asked volunteers to examine images selected or generated based on an AI model of the human visual system.

Image generated by an AI algorithm called BigGAN-deep, designed to activate a specific part of the brain known to respond to images of faces. Image generated on January 28, 2021. Credit: Weill Cornell Medicine

The images were expected to maximally activate multiple visual processing areas. Using functional magnetic resonance imaging (fMRI) to record the volunteers’ brain activity, the researchers found that the images activated target areas significantly better than control images.

The researchers also showed that they could use this image-response data to fine-tune their vision model for individual volunteers, so that images generated to be maximally activated for a particular individual performed better than images generated based on of a general model.

“We think this is a promising new approach to studying the neuroscience of vision,” said study lead author Dr. Amy Kuceyeski, professor of mathematics in radiology and mathematics in neuroscience. at the Feil Family Brain and Mind Research Institute at Weill Cornell Medicine. .

The study was a collaboration with the laboratory of Dr. Mert Sabuncu, professor of electrical and computer engineering at Cornell Engineering and Cornell Tech, and of electrical engineering in radiology at Weill Cornell Medicine. The first author of the study was Dr. Zijin Gu, a doctoral student co-supervised by Dr. Sabuncu and Dr. Kuceyeski at the time of the study.

Creating an accurate model of the human visual system, in part by mapping brain responses to specific images, is one of the most ambitious goals of modern neuroscience. Researchers have discovered, for example, that one visual processing region can activate strongly in response to the image of a face while another can react to a landscape.

To achieve this goal, scientists must rely primarily on non-invasive methods, given the risk and difficulty of recording brain activity directly with implanted electrodes.

The preferred non-invasive method is fMRI, which essentially records changes in blood flow in small vessels in the brain – an indirect measure of brain activity – when subjects are exposed to sensory stimuli or perform cognitive or physical tasks. . An fMRI machine can read these tiny three-dimensional changes in the brain, with resolution on the order of cubic millimeters.

For their own studies, Drs Kuceyeski and Sabuncu and their teams used an existing dataset comprising tens of thousands of natural images, with corresponding fMRI responses from human subjects, to train an AI-like system called an artificial neural network. (ANN). to model the visual processing system of the human brain.

They then used this model to predict which images, in the dataset, should maximally activate multiple targeted vision areas of the brain. They also coupled the model with an AI-based image generator to generate synthetic images to accomplish the same task.

“Our general idea here has been to map and model the visual system in a systematic and unbiased way, in principle even using images that a person would not normally encounter,” Dr. Kuceyeski said.

The researchers recruited six volunteers and recorded their fMRI responses to these images, focusing on responses across several areas of visual processing.

Results showed that for both natural and synthetic images, the predicted maximal activator images, on average across all subjects, activated the targeted brain regions significantly more than a set of selected images or generated to be only average activators. .

This supports the general validity of the team’s ANN-based model and suggests that even synthetic images can be useful as probes for testing and improving such models.

In a follow-up experiment, the team used image and fMRI response data from the first session to create separate ANN-based visual system models for each of the six subjects. They then used these individualized models to select or generate predicted maximal activator images for each subject.

The fMRI responses to these images showed that, at least for the synthetic images, there was greater activation of the targeted visual region, a face processing region called FFA1, compared to responses to images based on the model of band.

This result suggests that AI and fMRI can be useful for individualized modeling of the visual system, for example to study differences in visual system organization between populations.

The researchers are currently conducting similar experiments using a more advanced version of the image generator, called Stable Diffusion.

The same general approach could be useful for studying other senses such as hearing, they noted.

Dr. Kuceyeski also hopes to eventually study the therapeutic potential of this approach.

“In principle, we could change the connectivity between two parts of the brain using specially designed stimuli, for example to weaken a connection that causes excessive anxiety,” she said.

About this research news in AI and visual neuroscience

Author: Barbara Prempeh
Source: Weill Cornell University
Contact: Barbara Prempeh – Weill Cornell University
Picture: Image is attributed to Weill Cornell Medicine

Original research: Free access.
“Human brain responses are modulated when exposed to optimized natural images or synthetically generated images” by Amy Kuceyeski et al. Communication biology


Abstract

Human brain responses are modulated when exposed to optimized natural images or synthetically generated images

Understanding how the human brain interprets and processes information is important. Here, we investigated the selectivity and interindividual differences in human brain responses to images via functional MRI.

In our first experiment, we found that images predicted to achieve maximum activations using a group-level coding model evoke higher responses than images predicted to achieve average activations, and that the gain activation is positively associated with coding model accuracy.

Additionally, the anterior temporal lobe facial area (aTLfaces) and fusiform body area 1 showed higher activation in response to maximal synthetic images compared to maximal natural images.

In our second experiment, we found that synthetic images derived using a custom coding model elicited higher responses than synthetic images derived from group-level coding models or other subjects. The discovery of aTLfaces favoring synthetic images rather than natural images was also reproduced.

Our results indicate the possibility of using generative and data-driven approaches to modulate responses of brain regions at the macro scale and probe interindividual differences and functional specialization of the human visual system.

Gn Health

Not all news on the site expresses the point of view of the site, but we transmit this news automatically and translate it through programmatic technology on the site and not from a human editor.
Back to top button