Rotating a petavoxel reconstruction exposes the viewing-angle bias inherent to Golgi-Cox and confocal dendritic-spine classification

Abstract

Dendritic spines are the principal postsynaptic sites of excitatory transmission. For over a century, their shape has been sorted into discrete categories such as filopodia, thin, long thin, stubby, mushroom, and branched, largely by Golgi-Cox impregnation and, more recently, confocal microscopy. However, both approaches share a fundamental limitation. The histological sectioning and single-viewpoint imaging that these methods rely on cannot control the orientation of a spine relative to the observer. Because a spine is a three-dimensional object, the projection seen depends on how its parent dendrite lies within the section. Here, using the publicly available H01 petavoxel reconstruction of human temporal cortex imaged by serial-section electron microscopy (EM), we show that spine-shape classification depends strongly on viewing angle. A total of 445 spines on layer 4 basal dendrites of five pyramidal neurons were classified from an initial viewpoint (Angle 1), then reclassified after rotation in Neuroglancer (Angle 2). Only 20.9% kept their category, so chance-corrected agreement was negligible (Cohens kappa = 0.027). These observations provide direct evidence that the rigid Golgi-Cox and confocal taxonomies conflate true spine morphology with the arbitrary angle of view. Our results, therefore, support recasting spine shape as a three-dimensional continuum, measurable in petavoxel reconstructions such as H01 through free rotation in Neuroglancer. Significance statementThe classification of dendritic spines into discrete shape classes underpins a vast literature on synaptic plasticity, development, and disease. Yet it rests on two-dimensional images whose viewing angle is not controlled. By rotating the same human spines in a nanoscale EM reconstruction, this study shows that four out of five spines change category with viewpoint alone. The finding exposes a systematic bias in Golgi-Cox and confocal classifications. It argues that spine morphology should be treated as a measurable three-dimensional continuum rather than a set of fixed labels.