Many types of medical images (e.g. MRI, CT) are produced as slices that are acquired along one principal axis. In this format, left/right orientation is almost always ambiguous, regardless of the principal axis along which the images are viewed. To address this issue, conventions have been adopted to properly orient images (e.g. the radiology convention). However, these conventions have been applied in ways that can lead to additional ambiguity, an issue that has been exacerbated with the proliferation of 3D volume construction from 2D image stacks. In this talk I will:
- Briefly illustrate why accurate orientation of medical images is an important issue for clinical diagnosis and treatment, and why left/right is often ambiguous compared to body-based coordinate directions (i.e. medial/lateral, dorsal/ventral, anterior/posterior, rostral/caudal, superior/inferior).
- Demonstrate how left/right orientation is specified in raw images, and how it can be inadvertently flipped in 2D or 3D renderings when using a variety of open source and commercial software packages.
- Demonstrate practical techniques we have adopted to make sure that left is right and right isn't left behind, pun intended.