This software rapidly creates detailed brain maps from under 30 minutes of fMRI scans by using cross-participant models, helping doctors plan surgeries, diagnose disorders, and guide neurotechnology with less patient-specific data and easy visualization.
Functional brain mapping is a critical field within neuroscience and clinical medicine, aiming to identify which regions of the brain are responsible for processing specific types of information, such as language, sensory input, or higher-level cognitive concepts. This information is essential for applications ranging from basic research to clinical interventions, such as presurgical planning for neurosurgery or the development of brain-computer interfaces. Accurate and individualized maps of brain function can help clinicians avoid damaging critical brain areas during surgery, assess the impact of neurological disorders, and tailor treatments for conditions like aphasia or traumatic brain injury. However, the process of generating these maps has traditionally required extensive data collection, often involving hours of functional MRI (fMRI) scanning, which is time-consuming, costly, and not always feasible for all patients. Current approaches to functional brain mapping face significant limitations that hinder their widespread clinical and research use. Traditional methods rely heavily on collecting large amounts of patient-specific fMRI data, which can be impractical for individuals who are unable to tolerate lengthy scans due to age, illness, or injury. Moreover, these methods often require complex workflows involving multiple software tools, manual intervention, and specialized expertise, making them inaccessible for routine clinical practice. The need for extensive data also restricts the ability to map brain function in acute or time-sensitive settings, such as emergency neurosurgery or rapid assessment of consciousness. Additionally, existing techniques may not generalize well across individuals, leading to variability in mapping accuracy and limiting their utility for personalized medicine. As a result, there is a pressing need for solutions that can streamline the mapping process, reduce data requirements, and provide reliable, high-resolution functional maps with minimal patient burden.
This technology is a software solution for rapid, detailed functional brain mapping using cross-participant modeling of fMRI data. By leveraging less than 30 minutes of a patient’s preprocessed fMRI scans, the system aligns these scans to a standardized cortical surface and applies pre-trained encoding models that map various stimulus features—such as acoustic envelopes, phonemes, lexical, and semantic vectors—to voxel responses via regularized linear regression. The resulting feature weights are assembled into comprehensive surface or volumetric maps, indicating which stimulus feature class drives activity at each cortical location. The entire pipeline, implemented in Python and utilizing the pycortex library for visualization, automates data loading, cortical alignment, model fitting, and map generation, streamlining the process and minimizing the need for extensive individual scan data. What differentiates this technology is its innovative cross-participant transfer approach, which enables accurate functional brain mapping with a fraction of the data traditionally required. By using models trained on data from other individuals, it overcomes the limitations of patient-specific mapping, making the process faster and more accessible without sacrificing accuracy. This capability opens new clinical and neurotechnology applications, such as presurgical planning, assessment of neurological disorders, and optimization of brain-computer interface placement. The software’s open-source nature and compatibility with standard fMRI workflows further enhance its accessibility and adaptability, while ongoing validation and the potential for adaptation to other imaging modalities, like fNIRS, position it as a transformative tool for both clinical practice and research.
This software performs rapid functional brain mapping. It processes under 30 minutes of fMRI scans, aligning them and applying pre-trained cross-participant encoding models. Using regularized linear regression, it estimates feature weights for cortical locations, generating detailed functional maps that identify stimulus feature classes (e.g., sounds, concepts) driving brain activity.
U.S. Provisional application serial no. 63/945,036 was filed on 12/19/2025