Transgenic lines provide excellent tools to target defined groups of neurons within the developing nervous system. However, whole mount visualization frequently lacks the context to adequately align and annotate transgene expression. To address this we previously adapted volumetric registration techniques to align patterns from 109 transgenic lines into a common reference space and generated software to survey expression patterns at any point within the brain. This alignment allows us to better characterize expression, predict overlap and identify lines with expression in target regions. Without detailed anatomical annotation, however, the value of the data remains limited. While manual labeling addresses this, it is laborious, highly subjective, and constrained by our understanding of neuroanatomy, especially early in development when many primordial structures lack clear nuclear organization. However, we reasoned that local correlations in transgene expression might delineate emerging structures and allow us to computationally derive neuroanatomical annotations. Using spatially constrained data clustering of 112 transgenic lines, we were able to extract regions corresponding strikingly with previously identified neuroanatomical structures. These data-based annotations are less subjective than those manually defined and may allow identification of regions that would otherwise be impossible to reliably segment. Brain alignment also provides a framework for voxel-based morphometric comparisons, a statistically rigorous method for analyzing brain microstructure. As a proof of principle for the use of voxel-based morphometry in identifying discrete neuromorphological differences, we tested for differences in 10 wildtype brains and their left-right flipped counterparts. We speculated that if the analysis was both sensitive and selective, we should only pull out brain asymmetries. All statistically significant pixels that were found were in fact found in the habenula, the only asymmetric brain structure identified in larval zebrafish. Next, to test whether this analysis could locate novel brain abnormalities in mutant zebrafish, we aligned brains of kctd15a/b knockouts and compared them to their wildtype cousins. While kctd15 has been linked to human obesity, knockout larvae show no gross behavioral or morphological abnormalities yet lag behind wildtypes in their growth into adulthood. However, based on voxel-wise analysis of aligned brains, we identified a midbrain tegmental nucleus, the torus lateralis, that was absent in mutants. This finding was confirmed in six-month old mutants zebrafish using conventional histology. Together these results show that brain alignment can provide a powerful tool to screen for subtle neuroanatomical defects in mutants as well as to segment the larval brain into well-defined anatomical regions.