Clinical applicability of quantitative atrophy measures on MRI in patients suspected of Alzheimer’s disease
Silvia Ingala,Ingrid S. van Maurik,Daniele Altomare,Raphael Wurm,Ellen Dicks,Ronald A. van Schijndel,Marissa Zwan,Femke Bouwman,Niki Schoonenboom,Leo Boelaarts,Gerwin Roks,Rob van Marum,Barbera van Harten,Inge van Uden,Jules Claus,Viktor Wottschel,Hugo Vrenken,Mike P. Wattjes,Wiesje M. van der Flier,Frederik Barkhof
DOI: https://doi.org/10.1007/s00330-021-08503-7
IF: 7.034
2022-05-31
European Radiology
Abstract:Abstract Objectives Neurodegeneration in suspected Alzheimer’s disease can be determined using visual rating or quantitative volumetric assessments. We examined the feasibility of volumetric measurements of gray matter (GMV) and hippocampal volume (HCV) and compared their diagnostic performance with visual rating scales in academic and non-academic memory clinics. Materials and methods We included 231 patients attending local memory clinics (LMC) in the Netherlands and 501 of the academic Amsterdam Dementia Cohort (ADC). MRI scans were acquired using local protocols, including a T1-weighted sequence. Quantification of GMV and HCV was performed using FSL and FreeSurfer. Medial temporal atrophy and global atrophy were assessed with visual rating scales. ROC curves were derived to determine which measure discriminated best between cognitively normal (CN), mild cognitive impairment (MCI), and Alzheimer’s dementia (AD). Results Patients attending LMC (age 70.9 ± 8.9 years; 47% females; 19% CN; 34% MCI; 47% AD) were older, had more cerebrovascular pathology, and had lower GMV and HCV compared to those of the ADC (age 64.9 ± 8.2 years; 42% females; 35% CN, 43% MCI, 22% AD). While visual ratings were feasible in > 95% of scans in both cohorts, quantification was achieved in 94–98% of ADC, but only 68–85% of LMC scans, depending on the software. Visual ratings and volumetric outcomes performed similarly in discriminating CN vs AD in both cohorts. Conclusion In clinical settings, quantification of GM and hippocampal atrophy currently fails in up to one-third of scans, probably due to lack of standardized acquisition protocols. Diagnostic accuracy is similar for volumetric measures and visual rating scales, making the latter suited for clinical practice. Summary statement In a real-life clinical setting, volumetric assessment of MRI scans in dementia patients may require acquisition protocol optimization and does not outperform visual rating scales. Key Points • In a real-life clinical setting, the diagnostic performance of visual rating scales is similar to that of automatic volumetric quantification and may be sufficient to distinguish Alzheimer’s disease groups . • Volumetric assessment of gray matter and hippocampal volumes from MRI scans of patients attending non-academic memory clinics fails in up to 32% of cases . • Clinical MR acquisition protocols should be optimized to improve the output of quantitative software for segmentation of Alzheimer’s disease–specific outcomes .
radiology, nuclear medicine & medical imaging