First PET Investigation of the Human Brain at 2 µL Resolution with the Ultra-High-Resolution (UHR) scanner

Introduction: High spatial resolution is essential in brain PET imaging as it mitigates the partial volume effect. The poor contrast-to-noise ratio resulting from partial volume and spillover effects is a major obstacle to accurate delineation and quantification of small cerebral structures such as the brainstem nuclei and the individual gyri of the cortex, or to full recovery of image derived input function from cerebral vessels for tracer kinetic studies. Amid technological advances in PET instrumentation and a global thrust to extend brain PET imaging capabilities, our team has designed and built the UHR dedicated brain PET system. Featuring truly pixelated detectors, the UHR achieves an expected 1.3 mm isotropic spatial resolution, which is about a twofold improvement over the HRRT scanner, considered the state-of-the-art reference for brain PET imaging for the past 20 years. The purpose of this work is to demonstrate the ability of the UHR scanner to delineate small cerebral structures and accurately quantify their in vivo tracer concentrations.

Methods: Three patients prescribed with a medical ¹⁸F-FDG PET scan underwent their clinical evaluation on a whole-body Discovery MI scanner for a period of 10 to 20 min after 45-60 min uptake, followed by a 30- to 60-min acquisition on the partial UHR scanner (11/18 rings of detector modules, 14.3 cm axial FOV)), starting ~90 min post injection. Images from the UHR scanner were reconstructed by an OSEM algorithm implementing analytical PSF modeling with 8 subsets and 12 iterations, and included CT-based attenuation and scatter corrections. Images from both scanners were compared and region identification was performed using atlases. SUV values relative to the cerebellum were extracted by placing atlas-based volumes-of-interest (VOIs) on large structures and 4-mm diameter template VOI on identified nuclei.

Results: Several regions of the brain were easily identified visually in UHR images, particularly in the brainstem, whereas they could not be resolved by the whole-body scanner. The measured relative SUV values in selected structures of one of the patients are reported in the Table. Inferior and superior colliculi as well as the substantia nigra and red nuclei were clearly delineated. Commonly seen as a whole in PET images, the thalamus can now be visually segmented into smaller nuclei thanks to the enhanced resolution and contrast of the UHR images. Circumvolutions of the brain can also be resolved, which allows individual gyri such as the primary motor cortex and somatosensory cortex to be clearly discriminated. In addition to cerebral regions, hypermetabolic regions are noticeable along the cortical surface, a physiological process that can hardly be perceived with lower-resolution PET scanners due to partial volume and spillover effects.

Conclusions: The benefits of ultra-high resolution in the context of brain imaging with PET were demonstrated. With this quantum leap in image resolution, many brain regions that were previously only visible in MRI images, such as the brainstem nuclei and regions of the subcortical anatomy never seen before as separate entities, can now be discernable and quantifiable in UHR PET images. Even smaller structures such as the locus coeruleus and the raphe nuclei, respectively known to be involved in the pathogenesis of Alzheimer’s disease and major depressive disorder, can be expected to be resolved with more specific radiotracers targeting processes like neurotransmitters synthesis, neuroreceptors distribution and pathological protein deposition. The advent of UHR PET also holds promises of breakthrough in many fields like neuro-oncology, image-derived input function modeling and the diagnosis of neurodegenerative and neuropsychiatric diseases.

• Download figure
• Open in new tab
• Download powerpoint