The correlation of molecular neuroimaging and behavior studies in the preclinical field is of major interest to unlock progress in the understanding of brain processes and assess the validity of preclinical studies in drug development. However, fully achieving this ambition requires to perform molecular images of awake and freely moving animals, whereas most of the preclinical imaging procedures are currently performed under anesthesia. To achieve this combination, the MAPSSIC project aims to develop a pixelated implantable microprobe based on the CMOS Monolithic Active Pixel Sensor (MAPS) technology [1]. This device is set to be used on awake and freely moving rats after the injection of β ⁺ radiotracers used in PET. Thanks to its in situ position, the probe is able to directly detect short range positrons, making it sensitive to volumes corresponding to the structures in which it is implanted. In the other hand micro-PET devices use the coincident detection of γ-rays from annihilation as a relevant signal [2,3]. Here, we present the developments around such device and show the results from Monte Carlo simulations of a typical animal experiment using  C]Raclopride, a commonly used β⁺ radiotracer to confirm the relevance of this imaging device.

The probe relies on two 14700 μm x 700 μm x 200 µm MAPS glued back to back. Each sensor contains 128 x 16 binary pixels of 30 µm x 50 µm and is readout thanks to an adjustable rolling shutter. The 25 µm thickness of the sensitive area ensures a high sensitivity to positrons and an important transparency to γ-rays, allowing the measurement of a local information. The overall thickness of the probe limited to 400 µm ensures the mechanical robustness of the device while limiting the risk of damaging surrounding brain tissues. Simulations on an anatomical rat phantom [4] were performed using Time Activity Curves (TAC) generated thanks to C]Raclopride dynamic micro-PET acquisitions. Additional TACs were generated for the striatum [5] representing uptakes with theoretical decreases of 5% to 30% of the binding potential (BP_ND). Special emphasis was placed on assessing the ability of the probe to record a specific signal in a structure of interest based on local measurements and its capacity to quantify variations in  BP_ND.

More than 93% of the signal appears to be contained within the first 2 mm surrounding the probe. Furtherrmore, using a simple ROI focused on the striatum allowed to isolate the structure (more than 92% of the ROI signal) to accurately record the uptake of the radiotracer in it. Kinetic modeling performed on striatum TAC measured by the probe have demonstrated the ability of the device to consistently quantify the BP_ND. Though, an underestimation of the BP_ND caused by the overestimation of the cerebellum activity due to partial volume effect requires to apply a correction factor to the TAC before the kinetic modeling. Finally, the BP_ND variations implemented in the simulations are accurately reported by the probe measurements with less than 4% error.

Simulations of an in vivo neuroimaging procedure confirm the relevance of the information that the probe is able to provide. The spatial information enabled by the 2048 pixels matrix allowed to explore a first image segmentation approach that has proven to be efficient in targeting specific brain structures and quantifying typical variations of the BP_ND, which allows the opening to longitudinal and comparative behavior neuroimaging applications. The MAPSSIC project is currently undergoing rapid developments as newly manufactured probes have successfully undergone physical tests and are scheduled for biological validation on rodents. This validation is planned based on comparisons with the micro-PET, which is the gold standard.

Image/Figure:

Click to view full size

Image/Figure Caption:

Schematic view of the MAPSSIC device worn by a rat with the probe, the µ-PCB card, the micro controller and RF module communicating with a remote computer.

Author