Correlative Multimodal Imaging for Multiscale Analysis of Complex Biological Systems: Validation Methods and Performance Analysis

Correlative multimodal approaches have the potential to unlock a holistic understanding of complex biological processes with unprecedented detail. Recent advances in imaging modalities allow collecting structural and functional information of the same sample over several spatial and temporal scales, which will enable the quantitative characterization of multiscale metabolic pathways. Nonetheless, with such a disruptive ability, a thorough validation of these approaches is essential to support their findings. Typical methods range from landmarking operations based on physical markers and phantom models or using pairs of points identified after the image acquisition, to synthetic studies that test the reproducibility and reliability of the proposed multimodal approach. This chapter will describe the validation methods applied in correlative multimodal approaches of preclinical models identified in a survey of over 50 studies and analyze their performance to derive practical recommendations that future studies introducing novel pipelines can follow to fulfill this task. A summary of the most used imaging modalities will be presented first, followed by an overview of the preclinical animal models and the biological systems investigated, like mice, rats, or pigs and bone or brain, respectively. Next, the characterization of the validation methods will be complemented with a selection of quantitative metrics used for error calculation, providing insights into how each method performs and how they can help compare different studies and give confidence in the accuracy of the output. Finally, challenges associated with the validation methods will be discussed, including the need to create more annotated open access datasets with multimodal images to foster the development and independent evaluation of novel correlative multimodal approaches.