Value of histogram analysis of CT values in differential diagnosis of odontogenic keratocyst and ameloblastoma of the jaw

Overview

This study examines the utility of histogram analysis of computed tomography (CT) attenuation values for differentiating odontogenic keratocyst from ameloblastoma in jaw lesions. Both entities represent distinct pathological processes requiring different clinical management, yet their radiological differentiation remains challenging using conventional imaging assessment. The investigation applies quantitative histogram-based analysis to multislice CT data from a cohort of pathologically confirmed cases to identify discriminatory imaging parameters. The research addresses the need for objective, reproducible imaging metrics that can support preoperative differential diagnosis in oral and maxillofacial pathology.

Methods and approach

The study enrolled 32 patients with pathologically confirmed odontogenic keratocyst and 59 patients with pathologically confirmed ameloblastoma of the jaw. FireVoxel software was employed for image processing and extraction of histogram parameters from multislice CT datasets. The analysis quantified multiple statistical descriptors of CT attenuation value distribution within lesions, including mean, standard deviation, variance, coefficient of variation, skewness, kurtosis, entropy, and percentile values at the 1st, 5th, 10th, 25th, 50th, 75th, 90th, 95th, and 99th positions. Statistical comparison between the two lesion groups utilized two-independent-samples t-tests or Mann-Whitney U tests as appropriate. Diagnostic performance was evaluated through receiver operating characteristic curve analysis with calculation of area under the curve for each histogram parameter.

Results

Statistically significant differences between odontogenic keratocyst and ameloblastoma were identified in the 5th and 10th percentile histogram parameters, with Z-values of 4.486 and 2.058 respectively. The 5th percentile demonstrated superior discriminatory capacity, achieving the largest area under the receiver operating characteristic curve at 0.835. A cut-off value of -4 Hounsfield units for the 5th percentile provided optimal diagnostic efficacy for distinguishing between the two lesion types. The other histogram parameters examined did not demonstrate statistically significant differences between the diagnostic groups.

Implications

Histogram analysis of CT attenuation values derived from multislice CT imaging provides quantitative support for differential diagnosis of odontogenic keratocyst and ameloblastoma of the jaw. The 5th percentile of the CT value distribution demonstrates the strongest diagnostic performance among the parameters evaluated, offering a reproducible metric for clinical application. This approach represents a form of texture analysis that extracts information beyond visual interpretation, potentially enhancing preoperative diagnostic accuracy. The method may contribute to treatment planning by reducing diagnostic uncertainty in cases where conventional imaging features are equivocal, though the moderate area under the curve suggests limitations in standalone diagnostic utility.