Neuroengineering, also known as neural engineering, is a rapidly expanding interdisciplinary field that bridges neuroscience and engineering. It encompasses research at the cellular, tissue, and systems levels, and has become a core discipline within biomedical engineering and beyond. By applying engineering principles and methods, neuroengineering studies the nervous system and develops techniques for diagnosing, treating, and rehabilitating neurological disorders. This field includes neural sensing, signal processing, interfacing, modulation, computation, imaging, and device development, among others. Over the past decade, significant progress has been made in neuroengineering, or neurotechnology, which focuses more on neural devices or tools. This progress is partly attributable to initiatives such as the Brain Research through Advancing Innovative Neurotechnologies (BRAIN) Initiative in the U.S. and similar efforts worldwide.
Neurotechnology has made significant strides in both implantable and wearable devices. Implantable neurotechnologies, such as deep brain stimulation (DBS) devices for treating Parkinson’s disease etc. or spinal cord stimulation devices for managing pain etc., along with invasive brain–computer interface (BCI) devices for motor or communication prostheses, have demonstrated notable success. DBS devices, known for their effectiveness and reliability, have been implanted in over 160,000 patients to manage various neurological disorders. Conversely, implantable BCI devices have primarily remained within academic research labs for proof-of-concept studies and technological demonstrations, with only several dozen patients so far undergoing implantation. However, there is a growing interest from industry in commercializing such technology.
My journey in neuroengineering began with electroencephalography (EEG) for neural sensing and source localization. This research aims to understand brain function using noninvasive signals. EEG captures the electrical activity of neuron populations in the brain. While most neurons do not directly produce signals detectable at the scalp, synchronized firing of specific neuron sub-populations generates measurable electric potentials that can be detected using scalp electrodes. By employing an array of EEG electrodes, researchers have demonstrated the ability to capture its spatio-temporal distribution. This enables the localization and imaging of dynamic brain electrical activity in three-dimensional space. Today, EEG encompasses not only time series signals (waveforms recorded at specific electrodes) or spatial topographies (equi-potential maps over the scalp at specific time points), but also three-dimensional dynamic source images estimated by solving the source imaging problem. Similar to its magnetic counterpart, magnetoencephalography (MEG), EEG has emerged as a modern functional neuroimaging technology over recent decades, facilitating the imaging of brain activation and functional connectivity.
In addition to its role in understanding the brain, EEG has been instrumental in decoding intentions and controlling external devices, leading to the development of mind-controlled neuro-devices. BCIs decode brain intentions, empowering individuals, including those with motor impairments, to directly control computers or machines using brain signals, bypassing the conventional neuromuscular pathway. BCIs come in various forms, categorized as invasive, minimally invasive, or noninvasive. Invasive BCIs involve electrode implantation into the brain, offering high signal quality but limited by surgical requirements and safety risks. Minimally invasive BCIs utilize electrodes implanted over the brain surface or within brain blood vessels. Noninvasive BCIs, primarily utilizing EEG due to its safety, simplicity, and low cost, hold promise for widespread use among patients and the general population. However, they face technical challenges in extracting weak intention signals amidst background noise, as well as information reduction during transmission from the neuron sub-populations responsible for generating intentions to the scalp electrodes. This reduction of information can hinder the accuracy of noninvasive BCI systems. Engineering and neuroscience research efforts are essential to address these challenges and enhance the precision and reliability of noninvasive BCIs for everyday applications.
Recent advances in artificial intelligence (AI) and machine learning suggest promising avenues for enhancing noninvasive BCI performance, potentially leading to precise and robust brain-controlled neuro-devices. These include machine learning and deep learning-based algorithms aimed at improving the decoding performance to more reliably and precisely extract an individual’s movement intention. Such advancements enable the tracking of a continuously moving cursor, controlling the flight of a drone, maneuvering a wheelchair, or guiding the movement of a robotic arm in three-dimensional space.
Similar to its invasive counterpart, DBS, transcranial focused ultrasound (tFUS) neuromodulation has recently emerged as a neurotechnology capable of selectively exciting or inhibiting specific neural circuits with cell type specificity and deep brain penetration. A recent study from our lab demonstrates that tFUS stimulation applied to V5 can enhance the performance of human visual motion-based BCIs for communication purposes by modulating feature-based attention. This integration of tFUS with EEG-based BCIs suggests a promising avenue for further enhancing noninvasive BCI performance through high-specificity neuromodulation.
In summary, neuroengineering shows great promise in deepening our understanding of the brain and in the development of technologies for diagnosing and managing neurological disorders. Whether through invasive or noninvasive methods, the convergence of engineering and neuroscience will remain pivotal in driving innovations in neuroengineering.