Ruptured status discrimination performance of aspect ratio, height/width, and bottleneck factor is highly dependent on aneurysm sizing methodology

Abstract

Background: Numerous size and shape parameters have historically been used to describe cerebral aneurysms and to correlate rupture status. These parameters are often inconsistently defined.

Objective: To evaluate the impact of definition variation on rupture status detection performance.

Methods: Catheter rotational angiographic data sets of 134 consecutive aneurysms (60 ruptured) were automatically measured in 3 dimensions with a validated algorithm. According to the literature, aneurysm height was assessed as both maximal and orthogonal distances from dome to neck. Maximal and orthogonal widths were defined perpendicular to height definitions. Neck size was evaluated as minimum, maximum, and average diameter of the neck plane. Aspect ratio (AR; height/neck), height/width ratio (HW), and bottleneck factor (BNF; width/neck) were evaluated for alternative definitions of each size variable. Univariate statistics were used to identify significant features and to compute the area under the curve (AUC) of the receiver-operating characteristic.

Results: The AR, HW, and BNF showed significant dependence on parameter definition. Statistical significance and performance varied widely, depending on alternative definitions: AR, AUC range of 0.59 to 0.75; HW, AUC range of 0.48 to 0.72; and BNF, AUC range of 0.57 to 0.72. Using maximal height, orthogonal width, and minimum neck resulted in the best AR, HW, and BNF performances. Compared with HW, AR and BNF were less sensitive to alternative definitions.

Conclusion: Alternative aneurysm size definitions have a significant impact on prediction performance and optimal threshold values. Adoption of standard methodology and sizing nomenclature appears critical to ensure rupture detection performance and reproducibility across studies.