2026 ECNP Congress

PS02-2162

Cross-pipeline consistency of automated brain morphometry in mild cognitive impairment and Alzheimer’s disease

Amirmahdi Mojtahedzadeh — M. Magyar, G. Csukly · Semmelweis University- Faculty of Medicine- Medical Imaging Centre, Department of Neuroradiology, Budapest, Hungary

2026 ECNP Congress

Background

MRI-based morphometry is an important tool for detecting neurodegeneration in Alzheimer’s disease, particularly through structural biomarkers such as hippocampal atrophy [1]. However, variability across automated segmentation pipelines remains a noticeable challenge for reproducibility, comparability, and clinical interpretation of quantitative imaging results. Widely used tools such as FreeSurfer and volBrain rely on fundamentally different computational approaches, including surface-based reconstruction and voxel-based probabilistic segmentation, which may influence morphometric estimates [2,3]. Understanding the extent of agreement between these methods is vital for reliable biomarker development and multicentre research. We hypothesised that subcortical volumetric measures would demonstrate higher cross-pipeline consistency than cortical thickness metrics.

Methods

We performed a retrospective analysis of 64 patients with cognitive complaints who underwent high-resolution 3D T1-weighted MRI using a standardised acquisition protocol. All scans were processed using FreeSurfer (v7.4.1) and volBrain (vol2Brain). Bilateral hippocampal and amygdalar volumes, as well as cortical thickness in the precuneus and entorhinal cortex, were extracted from both pipelines. Both absolute volumetric values and intracranial volume (ICV)-normalised values were analysed to account for inter-individual differences in head size. Cross-method relationships were assessed using Pearson correlation coefficients, with statistical significance set at p < 0.05. Correlation strength was interpreted using conventional thresholds, and analyses were performed separately for each region and hemisphere to allow detailed region-specific comparison.

Results

Subcortical structures demonstrated strong cross-method consistency across pipelines. Hippocampal volumes showed the highest correlations (left r = 0.903, right r = 0.958; p < 0.001), consistent with previous segmentation studies [4]. Amygdalar volumes showed moderate-to-strong correlations (left r = 0.726, right r = 0.806; p < 0.001), indicating slightly greater variability compared to hippocampal estimates. Following intracranial volume (ICV) normalisation, correlations remained statistically significant but were modestly reduced (hippocampus: left r = 0.85, right r = 0.89; amygdala: left r = 0.69, right r = 0.72; p < 0.001), consistent with removal of shared variance related to global brain size [5]. In contrast, cortical thickness measures demonstrated only moderate correlations across pipelines (precuneus: r = 0.55-0.56; entorhinal cortex: r = 0.60 bilaterally; p < 0.001), indicating increased methodological variability. Across all regions, correlations were statistically significant, with consistently stronger associations observed for subcortical volumes compared to cortical metrics, highlighting region-specific differences in cross-method consistency.

Conclusions

Subcortical volumetric measures, particularly hippocampal and amygdalar volumes, show strong cross-method consistency between segmentation pipelines in a clinical population. Cortical thickness measures demonstrate lower consistency, likely reflecting methodological differences between surface-based and voxel-based approaches. These findings support the use of subcortical morphometry as a relatively stable imaging measure, while highlighting the need for caution when comparing cortical metrics across pipelines. Further work is required to establish harmonisation strategies for multicentre and longitudinal applications.

Final statement: Subcortical volumetric measures appear robust across pipelines, whereas cortical metrics remain method-dependent, emphasising the importance of methodological consistency in quantitative neuroimaging.

References

  1. Frisoni, G. B., Fox, N. C., Jack, C. R., Scheltens, P., & Thompson, P. M. (2010). The clinical use of structural MRI in Alzheimer disease. Nature Reviews Neurology, 6(2), 67-77. doi:10.1038/nrneurol.2009.215
  2. Manjón, J. V., & Coupé, P. (2016). volBrain: An online MRI brain volumetry system. Frontiers in Neuroinformatics, 10, 30. doi:10.3389/fninf.2016.00030
  3. Fischl, B. (2012). FreeSurfer. NeuroImage, 62(2), 774-781. doi:10.1016/j.neuroimage.2012.01.021
  4. Iglesias, J. E., Augustinack, J. C., Nguyen, K., Player, C. M., Player, A., Wright, M., Roy, N., Frosch, M. P., McKee, A. C., Wald, L. L., Fischl, B., & Van Leemput, K. (2015). A computational atlas of the hippocampal formation using ex vivo, ultra-high resolution MRI: Application to adaptive segmentation of in vivo MRI. NeuroImage, 115, 117-137. doi:10.1016/j.neuroimage.2015.04.042
  5. Voevodskaya, O., Simmons, A., Nordenskjöld, R., Kullberg, J., Ahlström, H., Lind, L., Wahlund, L.-O., Larsson, E.-M., Westman, E., & Alzheimer’s Disease Neuroimaging Initiative. (2014). The effects of intracranial volume adjustment approaches on multiple regional MRI volumes in healthy ageing and Alzheimer’s disease. NeuroImage: Clinical, 4, 524-532. doi:10.1016/j.nicl.2014.03.001