Wearable brain PET aims to support human brain imaging under less static conditions than conventional PET systems. Conventional brain PET usually requires subjects to remain stationary in a sitting or supine position during scanning, which can limit use in some groups, including children and individuals with epilepsy or other neurological disorders. A 2026 evaluation published in the Journal of Nuclear Medicine assessed SmartBrain, a wearable brain PET system developed for human brain imaging. SmartBrain was evaluated using recognised PET performance standards, phantom imaging, dynamic rat imaging and human 18F-FDG brain imaging. Its performance was also compared with the Discovery MI PET/CT system. The evaluation focuses on whether a compact wearable system can meet basic performance expectations for brain-focused PET imaging while allowing data collection under less static conditions than conventional systems.
Compact Wearable System Design
SmartBrain consists of a compact detector ring designed for brain imaging. The system uses multiple detector modules arranged in rings around the head. Each module combines lutetium–yttrium oxyorthosilicate crystals with silicon photomultipliers. A channel-reduction strategy decreases the number of readout channels, helping reduce the size and weight of the electronics. Short crystals are also used to reduce system weight, although this choice lowers sensitivity and signal-to-noise ratio.
Must Read: Deep Learning Enhances Fast Brain MRI
The overall system weighs approximately 6 kg. Two mechanical support approaches were developed: a wearable backpack and a suspension system. The backpack configuration permits data collection in ambulatory states, while the suspension system supports seated scanning. The system also supports list-mode acquisition, time-stamped data, time-of-flight processing and dynamic imaging. Temperature is stabilised using external fans and air-cooled ducts.
The design addresses one limitation of conventional brain PET: the requirement for stationary scanning. Earlier wearable or portable brain PET systems have shown feasibility but have also had limitations in field of view, weight, geometric flexibility or imaging performance. SmartBrain was developed to support imaging during free movement, with a larger axial field of view than some previous wearable prototypes and with time-of-flight capability.
Performance Evaluation and Phantom Imaging
SmartBrain was evaluated using the NEMA NU 2-2018 standard, covering spatial resolution, sensitivity, count rate performance, correction accuracy, image quality, energy resolution and timing resolution. The system demonstrated millimetre-level spatial resolution at the centre of the field of view. Sensitivity was lower than that of larger clinical PET/CT systems, which is consistent with the use of thinner crystals in a lightweight wearable design. Time-of-flight performance supported image reconstruction under low-activity conditions.
Image quality testing used a brain-sized phantom with spheres of different sizes. Contrast recovery was measurable across the spheres, with higher recovery in the largest sphere. A multilayer Derenzo phantom showed that small hot rods could be visually separated, particularly when time-of-flight reconstruction was used. Without time-of-flight reconstruction, the smallest rods remained visible, but image noise increased.
A custom Hoffman brain phantom was also used to assess realistic brain-structure imaging. The phantom contained radioactive grey matter regions, while white matter and ventricular regions were nonradioactive. Reconstructed transaxial, coronal and sagittal slices showed cortical grey matter activity with separation from nonradioactive regions. The images reproduced gyral patterns, supporting the system’s ability to visualise brain structures in a controlled phantom setting. These tests provide evidence of spatial resolution and image-quality performance under experimental conditions.
Human Imaging and Technical Limitations
Human imaging involved a 43-year-old male patient with epilepsy who underwent 18F-FDG brain imaging under an approved protocol. A conventional PET/CT scan was acquired first, followed later by a SmartBrain PET scan after the injected activity had decreased. SmartBrain reconstruction used time-of-flight information, point-spread-function modelling and standard corrections. CT images from the Discovery MI PET/CT system were used for attenuation correction.
SmartBrain images showed cortical uptake patterns, defined grey matter distribution and preserved gyral anatomy. White matter and ventricular regions showed low to no activity, in line with the expected 18F-FDG distribution. Comparison with the Discovery MI PET/CT system showed that both systems displayed cortical uptake patterns. SmartBrain showed detailed cortical features despite the later imaging time point and reduced activity. The comparative images indicated delineation of gyral details with the wearable system, while the conventional scanner produced high-quality images under standard whole-body PET/CT conditions.
Several design choices may contribute to the observed performance. Full-ring coverage supports field-of-view coverage, short crystals reduce depth-of-interaction uncertainty and crystal scatter, and the compact bore reduces the impact of noncollinearity in brain imaging. Time-of-flight reconstruction also helps compensate for reduced sensitivity. Current limitations remain. The thinner crystals reduce sensitivity, the absence of pile-up correction constrains performance under higher count-rate conditions, and fast dynamic imaging remains challenging. Planned optimisation includes detector geometry, correction algorithms and artificial intelligence-assisted reconstruction.
SmartBrain demonstrated brain-level spatial resolution and dynamic imaging capability in the reported evaluation. Its compact and wearable configuration allows brain PET imaging beyond the constraints of conventional static scanning. Phantom and human imaging showed visualisation of cortical structures. Further optimisation is needed for sensitivity, detector coverage, count-rate performance, acquisition time and reconstruction methods. Current limitations include lower sensitivity from thinner crystals, the absence of pile-up correction and challenges for fast dynamic imaging.
Source: Journal of Nuclear Medicine
Image Credit: iStock
References:
Liu H, Qu W, Liang D et al. (2026) Performance Evaluation of SmartBrain: A Wearable PET System for Human Brain Imaging. Journal of Nuclear Medicine: jnumed.125.271350.
