CMOS-based Amorphous Silicon Microchannel Plate (CAMCP) detectors are devices built on and read-out by a multi-channel CMOS tile, using a PLANACON photocathode and a hexagonal grid of high aspect ratio micro-channels. Metascintillators are combinatory scintillators comprising more than one type of material, exploiting the electron leakage from a dense scintillator to a fast one. The two structures were simulated together using GATE. The scintillator had dimensions of $24 \times 9.6 \times 7.7 \mathrm{~mm}^{3}$, made of 22 slabs of LYSO:Ce ($24 \times 9.6 \times 0.25 \mathrm{~mm}^{3}$) and PEA2PbBr4perovskites $\left(24 \times 9.6 \times 0.1 \mathrm{~mm}^{3}\right)$. The CAMCP was modeled, based on prototypes, with $1.5 \mu \mathrm{m}$ diameter channels of $40 \mu \mathrm{m}$, organized to be read out by $15 \mathrm{x} 15 \mathrm{~mm}^{2}$ channels, with 6 ps timing resolution. The study focuses on the single photon detection capabilities of the CAMCP and uses machine learning to reconstruct the characteristics of the scintillation event, such as interaction energy and x-y-z-t onset. It is demonstrated that with this structure, it is possible to achieve down to 1 mm spatial, $\sim 8 \mathrm{~mm}$ depth-of-interaction and 89 ps timing resolution for gamma interaction. Event analysis and timewalk corrections further improve the characteristics of this architecture.