Reply:

We thank Dirk Smeets for his interest in our article and AJNR for the opportunity to respond. Regarding sample size, our study is in line with previous work comparing the effect of a report with no report on accuracy among clinical readers. Our study had the same or more cognitively healthy (CN) participants as in similar prior studies.^1,-,3 Most interesting, it was the CN group in which the report was associated with lower accuracy. These prior studies had 3–9 readers, including trainees and non-neuroradiologists. To mirror actual practice, we had 4 neuroradiology attendings. Smeets cited an FDA database as the basis for the total number of examinations and readers necessary to demonstrate improved performance, presumably for FDA approval; however, it is not clear whether the FDA measure consists of a comparison of radiologists’ diagnoses with-versus-without a report. Still, we agree that a larger sample is needed. The reality as noted in the article cited by Smeets is that MRMC studies “are quite expensive to conduct, requiring a good deal of readers’ time for image interpretation.”⁴

Regarding the validity of a control cohort of patients with subjective memory symptoms, it is exactly this group that must be differentiated from patients with neurodegenerative disease. Second, volumetric analysis showed that the hippocampal, temporal lobe, parietal lobe, and whole-brain volume normalized volumes were higher in CN relative to patients with AD and FTD, suggesting that they were CN. Last, although it is possible that the CN group might develop AD in the future, it is conjecture at this time.

While our study did not demonstrate a change in the accuracy of neuroradiologists’ diagnoses with the report, we believe that quantitative volumetrics is superior to visual assessment in estimating brain volumes and advocate its use as an adjunct to clinical interpretation.

• © 2025 by American Journal of Neuroradiology