Grant: Research Tools and Instruments

The primary goal of my research program is to develop and improve brain imaging and stimulation techniques. The level of brain fluorodeoxyglucose (FDG; a radioactive glucose analog) uptake estimated by a positron emission tomography (PET) can be used as an indicator of the overall health of different brain regions. My group has developed an automated brain FDG-PET reading program that can classify individuals with abnormal spatial distribution of FDG uptake which was associated with the subjects' cognitive impairment vs. individuals with normal cognition with high accuracy (~90%). However, earlier identification, i.e., classifying individuals with mild cognitive impairment (MCI) who later progress to dementia (pMCI) vs. individuals with MCI who do not (sMCI), is more challenging which resulted in modest accuracy (~70%). We are currently investigating how the program's accuracy can be further improved for early identification of dementia at MCI stage, utilizing more advanced artificial intelligence (AI) program and combination of different types of brain scans. Our preliminary data suggests that these approaches can improve the prognostication accuracy. To develop the proposed AI program with large dataset, the proposed high performance computer workstation with Graphics Processing Units (GPU) is necessary. One of the biggest challenges in neuroimaging studies in dementia prognostication is that a causality cannot be established between brain-behavior relationship. For example, it is unclear whether the altered brain activities and/or connectivity that we observed in pMCI vs. sMCI is directly related with cognitive decline or if it was only a secondary effect (i.e., not physiologically relevant). A novel non-invasive brain stimulation technique has recently been developed that may be able to manipulate the brain activities and/or connectivity among multiple regions (i.e., multi-target approach using high-definition transcranial electrical stimulation; HD-tES). The target brain regions for HD-tES are typically determined based on group-level neuroimaging data analysis. However, this practice ignores the significant inter-individual variability as well as time-varying intra-individual variability. The proposed, time-adaptive stimulation technology implemented based on simultaneous recordings of electroencephalography (EEG) and a novel brain simulator will allow more accurate prediction of HD-tES effects which will be based on individual's brain imaging data. A prototype system will be developed using the proposed fully-integrated EEG/HD-tES machine. The prospective trainees will significantly benefit from the proposed novel equipment which will ensure their guaranteed access time to develop the AI and innovative brain stimulation protocol as one of the first users of the fully integrated EEG-guided HD-tES system, the skillsets which will be highly valued in this rapidly growing field.