Dataset: Performance evaluation of image co-registration methods in photoacoustic mesoscopy of the vasculature

In short, this dataset comprises of 3D mesoscopic photoacoustic imaging (PAI) of breast cancer patient-derived xenografts imaged twice. Images were preprocessed and vascular networks were segmented across the image volume. Pairs of mesoscopic PAI were co-registered using five co-registration methods divided into three categories: intensity-based (labelled MI and NCC), shape-based (labelled ICP and Distance), and deep learning-based co-registration (labelled LocalNet). A pair of fixed and moving images are provided along with the “warped” images following co-registration with each tested technique. Raw reconstructed fixed and moving data are provided (fixed_rawImage.zip and moving_rawImage.zip) along with original preprocessed and segmented data (preprocessedImages.zip). Pairs of fixed and warped intensity images and segmentations are provided for each co-registration method ( i) MI_coRegistered.zip, ii) NCC_coRegistered.zip, iii) ICP_coRegistered.zip, iv) Distance_coRegistered.zip, and v) LocalNet_coRegistered.zip). All data are stored in the NIFTI file format (.nii.gz), except for the raw data (.mat), within the zipped folders (.zip).