Functional Magnetic Resonance Imaging (fMRI) is a widely used neuroimaging technique for the analysis of neural activity and functional connectivity. However, the fMRI signal is inherently noisy and susceptible to various artifacts, compromising the accuracy and reliability of derived analyses. This becomes particularly critical when dealing with stroke patients, given the added complexity associated with their neurological condition. Specific preprocessing and denoising are integral steps to identify the nuisance sources and mitigate their effect on fMRI analysis. To address these challenges, we present fMRIStroke, a dedicated preprocessing pipeline designed specifically for the quality control and preprocessing of fMRI data from stroke patients. fMRIStroke operates as an enhancement to standard preprocessing workflows. Building on the outputs from commonly used tools like fMRIprep, fMRIStroke introduces additional quality control visualizations, computes supplementary confounding variables, and performs confound regression (denoising), resulting in preprocessed fMRI data that is ready for subsequent analysis of neural activity or connectivity.