Fig. 1.

Developmental changes in cardiac morphology prevent day-long time-lapse imaging with existing algorithms. a In prospective optical gating, brightfield images (greyscale) are acquired and analysed in real time to assign a phase (temporal position in the cardiac cycle) to each frame. Forward prediction is then used to accurately trigger acquisition of one synchronised fluorescence image (green) at a user-defined target phase. b The sample is scanned through the light sheet to generate a synchronised 3D $z$-stack. c The move from confocal laser scanning microscopy (CLSM; upper) to light sheet fluorescence microscopy (LSFM; middle) reduces phototoxic effects by several orders of magnitude by limiting excitation in space. Similarly, the move from retrospective (upper and middle) to prospective gating (lower) reduces phototoxic effects yet further by limiting excitation in time. d Prospective optical gating relies on periodic changes within heartbeats (upper) and is able to cope with small changes between heartbeats (middle). However, phase-lock cannot be maintained over developmental time scales, because the heart undergoes drastic morphological changes (lower). e These changes mean that image-based similarity metrics are unable to match new brightfield images against the reference heartbeat image sequence recorded at the start of the experiment. f A new “smart microscope” is required that can maintain phase-lock in the face of these drastic changes in cardiac morphology over the course of day-long time-lapse imaging