RETURN TO ABSTRACT LISTING


Objective:
The use of structural MRI automated volumetric analysis allows for a quantitative assessment of brain atrophy to guide the neuroradiologist in evaluating for a specific, etiological neurodegenerative process in patients with cognitive decline. Several such volumetric analysis tools have become available for clinical use; however, intermethod reliability is not always certain. Therefore, this exhibit aims to compare two such algorithms available at our institution, NeuroQuant and IcoMetrix (Icobrain dm).

Materials and Methods:
MR images were obtained from 100 patients ranging between the ages of 24 and 90 (mean 72 years) who underwent 3D volumetric brain MRI protocol scans within our health network from December 16, 2022, to February 10, 2023. Volumetric segmentation and analysis were performed via two separate algorithms on each patient’s scan. These include the NeuroQuant algorithm for automated quantification of regional brain volumes (CorTech Labs, La Jolla, CA), as well as the Icometrix (Icobrain dm, Leuven, Belgium) algorithm for segmentation and volumetric analysis. Both algorithms assign normative percentiles for the volumes of both the hippocampi and the inferior lateral ventricles. Statistical analysis was performed to compare the assigned normative percentiles for both the hippocampal and inferior lateral ventricular volumes via multiple paired t tests, with p values less than 0.05 considered significant (Prism 9.4.0, Graphpad, Boston, MA).

Results:
The mean normative percentile assigned to hippocampal volume for the 100 patients by the NeuroQuant algorithm was the 48.44th percentile. In comparison, the mean normative percentile assigned by the Icometrix algorithm was 36.71st for the same patients (p < 0.0001). Similarly, the mean percentile assigned by NeuroQuant for inferior lateral ventricular volume for 70.48th versus 56.74th as assigned by Icometrix (p = 0.0011).

Conclusion:
Volumetric analysis algorithm comparison between NeuroQuant and Icometrix (Icobrain dm) demonstrated a statistically significant difference regarding mean normative percentile values of both hippocampal and inferior lateral ventricular volumes, with lower overall normative values utilizing the Icometrix algorithm. Explanation for these noticeable normative value differences can be hypothesized to include factors, such as differences in normative brain database utilization, along with differences in segmentation and postprocessing techniques. It remains most important, however, that the neuroradiologist is aware of these normative value differences and considers such upon determining an interpretation.