A Novel Vascular Fiducials‐based Approach (VASFID) for Co‐registering Multiscale Imaging Data for Microcirculation Systems Biology

Abstract

Image‐based systems biology is emerging as a powerful new paradigm for characterizing the microcirculation because it facilitates the modeling of ‘emergent’ vascular function from underlying microvascular imaging data. Moreover, recent advances in preclinical vascular imaging have demonstrated the feasibility of acquiring multicontrast 3D microvascular data simultaneously from magnetic resonance microscopy (MRM), computed tomography (CT), light‐sheet microscopy (LSM) or multiphoton microscopy (MPM), such as T1‐weighted (T1W) and diffusion weighted (DW) MRI contrasts, collagen fibers, tissue cellularity etc. Therefore, there is an urgent need for methods that facilitate the co‐registration of complementary multiscale images to develop and validate novel systems biology models of the microcirculation in health and disease. However, co‐registration of multiscale imaging data for image‐based systems biology applications has remained challenging due to differences in image contrast mechanisms, field of view and spatial resolution which may preclude the identification of common landmarks in MRI, CT and optical images. Additionally, the region of interest can lack isotropic features or landmarks (e.g. tumor tissue), or the tissue could have undergone irreversible non‐rigid deformations during sample preparation (e.g. many optical clearing protocols are known to cause anisotropic expansion or shrinkage of the sample). To address these challenges, we developed an elastic multiscale image co‐registration method (VASFID) that employs “internal” vascular fiducials made visible via a novel vascular contrast agent combination that facilitates integration of MRI, CT and optical images for image‐based investigations of healthy and pathological tissues. Here, we showcase three applications of VASFID: (1) co‐registration of MRI with CT and MPM data in a preclinical breast cancer model to develop a multiscale “cancer atlas” (Fig. 1); (2) generation of anatomically accurate microvascular maps of different brain regions and leg muscles (Fig. 2a–h); and (3) co‐registration of MRI data with post‐clearing light‐sheet microscopy data from the same murine brain sample that had shrunk by ~10% of its original volume (Fig. 2i–l). Collectively, these results demonstrate the utility of our VASFID co‐registration method for aligning optical imaging derived cellular features (e.g. astrocytes, neurons etc.) with vascular data acquired from CT and complementary MRI contrasts (e.g. DW). We expect our VASFID approach to enable novel systems level characterizations of the vascular microenvironment in preclinical disease models and organs.Support or Funding InformationThis work was supported by NCI 1R01CA196701, 1R21CA175784 and 5R01CA138264.Creation of a multiscale “cancer atlas”.(a) MRI, (b) CT were co‐registered using 18 pair‐wise vascular fiducials. (d) These were aligned with those from MPM to achieve MRI‐MPM co‐registration (e). Sections selected from the same volume illustrate co‐registered collagen fiber data (f) and the corresponding fractional anisotropy (FA) map from DW‐MRI (g). GalRh: vascular contrast, GFP: green fluorescent protein expression from MDA‐MD‐231 cancer cells and Col: collagen fiber signal (d).Figure 1Generation of anatomically accurate high‐resolution vascular maps of murine brain and leg.We demonstrate the feasibility of generating high‐resolution microvascular maps of different brain regions and leg muscles by co‐registering MRI (a, e, i) with CT (b, f) or LSM data (j) as shown in (d, h, l). This was not feasible using an affine transformation as shown in (c, g, k) wherein microvessels and bone remained misalignment despite a good volumetric alignment at the whole‐tissue level.Figure 2