cammiller/imagingPC: Analysis of Imaging Mass Spectrometry Data using a Process Convolution Approach

Analysis of imaging data collected on a regular grid using a process convolution (PC) approach. In the discrete PC approach we employ, a zero-centered latent process is convolved with a smoothing kernel function in order to account for spatial information. For computational efficiency with large imaging datasets, this package implements a semivariogram-based approach to estimate and fix the smoothing kernel function. In the context of a Bayesian mixed models framework, this package writes and fits models to estimate the latent process at a limited set of locations called support sites while also incorporating covariates of interest. Furthermore, this package incorporates the PC approach into left-censored models and marginalized two-part models to account for different zero-generating processes. This package was designed for imaging mass spectrometry (IMS) data, but the methods can be extended to imaging data collected over a regular grid.

install.packages("remotes")
remotes::install_github("cammiller/imagingPC")