• Kornelia M Kliś, Roger M Krzyżewski, Borys M Kwinta, Krzysztof Stachura, and Jerzy Gąsowski.
• Faculty of Computer Science, Electronics and Telecommunications, AGH University of Science and Technology, Kraków, Poland; Jagiellonian University Medical College, Kraków, Poland.
• World Neurosurg. 2018 Dec 1; 120: e863-e869.

BackgroundShape and density of intracerebral hemorrhage (ICH) are associated with a higher risk of poor treatment outcome. However, methods of assessment for the features are still inconclusive. Therefore, we decided to measure ICH shape irregularity using shape factors to achieve objective results.MethodsWe retrospectively analyzed 48 patients with spontaneous ICH confirmed by head computed tomography (CT) scan. We obtained detailed medical history and blood test results from medical records. On admission patients were assessed using Glasgow Coma Scale score, and on discharge patients were assessed using Glasgow Outcome Scale (GOS) score. GOS score of less than 3 was defined as poor outcome. For each slice of CT scan with visible ICH, we extracted its contour and calculated the fractal dimension (FD), compactness (C), and Fourier factor (FF). We also calculated the circle factor (CF), which was defined as the contour perimeter/perimeter of the biggest circle that can be inscribed into the contour, and density heterogeneity, defined as the variance of pixel density.ResultsA total of 28 patients (58.33%) had poor treatment outcome. Those patients had significantly higher C (0.71 ± 0.09 vs. 0.59 ± 0.09; P < 0.01), FD (1.42 ± 0.12 vs. 1.27 ± 0.09; P < 0.01), and CF (3.59 ± 0.92 vs. 2.63 ± 0.63; P < 0.01). In multivariate logistic regression analysis, FD (odds ratio, 4.176; 95% confidence interval, 1.551-15.577; P = 0.012) remained independently associated with higher risk of poor treatment outcome.ConclusionsEach of the shape descriptors, except FF, was associated with treatment outcome after ICH. FD can be used as an independent predictor of outcome.Copyright © 2018 Elsevier Inc. All rights reserved.

20 keywords...

hide…