There are currently no effective treatments for mitochondrial diseases [1, 2]. One major obstacle to clinical trials is a lack of reliable, non-invasive, reproducible biomarkers against which disease progression and treatment response can be measured. Chronic progressive external ophthalmoplegia (CPEO), a common clinical manifestation of mitochondrial disease, is characterised by gradually progressive reduction in the range of eye movements (ROEM). It is usually sporadic, as occurs with single large-scale deletions of mitochondrial DNA (mtDNA) which have a reported point prevalence for clinically affected adults of 1.17 in 100,000 people [3]. However, it can also be inherited in a maternal or Mendelian pattern depending on whether the causative genetic defect resides in the mitochondrial or nuclear genome. Imaging studies have reported extra-ocular muscle (EOM) atrophy, sometimes associated with EOM signal change, as a feature of CPEO. However, investigations have been, for the most part, limited to small, genetically undefined patient series with rather variable radiological findings [4–7]. For instance, CPEO has been associated with normal or near normal volumes despite clinically severe ophthalmoplegia on orbital T1-weighted (T1w) MRI [6], whilst small EOMs are also reported using the same imaging modality [4, 5]. Furthermore, EOM MRI signal abnormality in patients with CPEO, in the absence of orbital or EOM inflammation, has previously been proposed as a possible discriminator from other conditions which cause ophthalmoparesis [6]. However, this finding has not been replicated in independent patient cohorts with CPEO.