MR spectroscopy of brain metabolism
Kaiser, Inglis, Pelton, Brooks, Horning, Pappas
Metabolic abnormalities may have a fundamental role in brain pathology, including traumatic brain injury and degenerative conditions such as Alzheimer’s disease. We are developing methods to measure compounds including glucose, ATP and lactate on a clinical MRI scanner without special hardware or sequences, by exploiting accurate prior knowledge and advanced spectral fitting.
Blood flow dynamics following stroke
Ivanova, Pappas, Frederick, Dronkers, Inglis, D’Esposito
Strokes are caused by interruption to the blood supply to part of the brain, commonly because of a blocked artery. The blood dynamics following a stroke may be especially important in understanding the recovery potential. We are measuring delays in the blood supply (hemodynamic lags) after stroke, as well as the cerebral blood flow, which is generally regarded as a measure of metabolic demand.
Non-invasive assessment of concussions
Halabi, Smith, Inglis, D’Esposito
We are developing a non-invasive device that detects the blood pressure wave in the brain using accelerometers placed on the scalp. Preliminary tests show that the vibrations detected are a sensitive indicator of concussion. We aim to characterize recovery from sports-related concussion using an approach that integrates our novel device, clinical evaluations, cognitive assessments, symptom inventories, and MRI of three different but complimentary vascular measures to characterize blood dynamics in the brains of concussed subjects.
Non-invasive brain stimulation
Sheltraw, Inglis, in collaboration with Ivry Lab
Pulsed transcranial magnetic stimulation (TMS) is limited in its ability to probe subthreshold neuronal activity. This has led researchers to attempt subthreshold perturbation using electrodes connected directly to the scalp, to deliver either direct (tDCS) or alternating (tACS) currents. We are developing a new form of magnetic stimulation that increases the electric fields in the brain by an order of magnitude compared to what can be attained safely with tDCS or tACS.
Domoic acid toxicity & epilepsy in California sea lions
Inglis, Baraban, in collaboration with The Marine Mammal Center, Sausalito, CA
Domoic acid (DA) is a neurotoxin produced in some oceanic algal blooms. This neurotoxin enters the food web via crustaceans and small bony fish. California sea lions are particularly at risk. Sea lions poisoned with DA often strand on beaches where they may exhibit brain seizure activity. We are developing an MRI paradigm to assess the brain damage produced by DA so that the long-term consequences can be understood. In particular, we want to know how animals poisoned in utero may develop as adults. The temporal lobe epilepsy that is a hallmark of DA toxicosis may also help us understand human epilepsy, and develop treatments for this condition.
Augmented remote neurocognitive rehabilitation & psychotherapy
D’Esposito, Pappas, in collaboration with NeuroHealth Technologies, Inc.
One in two individuals experience a mental illness or neurologic disorder across their lifespan; 73% of these affected individuals cannot access the care they need. A primary solution to this major unmet medical need is remote care provision, which has now been accelerated exponentially due to the COVID pandemic. Most clinicians are dissatisfied with remote neurocognitive rehab and psychotherapy due to the loss of in-person fidelity, resulting in diminished ability to engage the patient throughout a session. We are developing a clinician-centric teletherapy platform, called Neuro-ART (Neuro-Assessment, Rehabilitation, and Therapy), to provide a range of real-time patient insights for clinicians via digital biomarkers and machine learning. We will further investigate the application of this platform to enhance access to mental healthcare for minority and rural populations.
Image registration & normalization in the presence of anatomical abnormalities
Pappas, Kayser, D’Esposito
Modern neuroscience relies heavily on anatomical templates with which locate the same structures across a group of people. It is especially challenging to normalize individuals to a template in the presence of gross pathology, such as a stroke lesion. We are developing custom templates that can be used to locate reliably normal anatomical tissues despite gross abnormalities.
Automated lesion segmentation in MRI using deep learning neural networks
Pappas, Kayser, D’Esposito
Stroke lesions vary considerably in size and shape. They also change over time. This makes it incredibly difficult for software algorithms to determine a lesion boundary automatically. Most lesion segmentation is still done by hand. We are developing deep convolutional neural networks to provide high accuracy lesion segmentation. We use a “neurologist-inspired” approach where the neural network exploits the lesion-induced asymmetries between brain hemispheres.
Functional magnetic particle imaging
The performance limits for functional brain imaging using MRI are now well established. Higher magnetic field, faster gradients, and improved scanner stability, e.g. reducing the effects of magnetic susceptibility of the chest during respiration, all offer incremental benefits. Magnetic Particle Imaging has different physical and biological restrictions than MRI. Magnetic susceptibility is a non-factor in MPI. Gradient speed, while a restriction, imposes itself differently with respect to the maximum attainable spatial resolution. And physiological fluctuations – the bane of fMRI – are expected to be greatly reduced in fMPI compared to fMRI.
Locus coeruleus imaging in aging & dementia
Inglis, in collaboration with Jagust & Walker Labs
Pathological alterations to the locus coeruleus (LC), the major source of noradrenaline in the brain, are evident in early stages of neurodegenerative diseases such as Alzheimer’s disease. Novel MRI approaches provide an opportunity to quantify structural features of the LC during disease progression. We are part of an international consortium seeking to optimize LC visibility in MRI, with the hope that LC imaging might eventually serve as a biomarker for neurodegenerative diseases.
Reproducibility in neuroimaging
Human neuroimaging studies may be strongly affected by problems such as low statistical power, vast flexibility in data acquisition and analysis parameter spaces, and software errors. These issues often lead to results that are spurious and fail to replicate or to generalize. We worked on a set of recommendations and procedures that would produce the most meaningful and reliable answers to neuroscientific questions, including best practices for coding, defining workflow, establishing a robust execution environment, and transparent reporting of methods and results in publications.
Open source computing & data sharing in neuroimaging
Computational techniques are central in many areas of neuroscience, and code is relatively easy to share. We worked with international colleagues to identify barriers to code sharing, then developed a set of recommendations that would allow the full potential of global neuroimaging data to be exploited. We also worked on practical recommendations to enhance reuse of neuroimaging data shared in repositories. Collectively, improved documentation and sharing of code and raw data should enhance reproducibility of research, and improves the chances of making robust, novel discoveries on a range of common brain diseases.
Ultralow field MRI of the brain
The Clarke Group constructed a ULFMRI scanner that is capable of detecting at magnetic fields of 50-250 microtesla. A potential advantage of ULFMRI is the intrinsic tissue contrast compared with that of clinical high-field MRI. Differences in relaxation times at ULF arise from the slow exchange of water molecules trapped in protein folds, together with intermolecular proton exchange between the free water and the OH or NH functional groups on proteins. Sensitivity to slow intermolecular exchange suggests that ULFMRI may be used to image stroke or TBI, where changes in protein conformation are an early indication of pathology. Degenerative conditions such as Alzheimer’s disease, characterized by the abnormal build-up of proteins, are also good candidates for ULFMRI detection.