Purpose: The purpose of this retrospective study was to investigate the differentiation of abscess and necrotic tumors, using susceptibility-weighted imaging (SWI) and apparent diffusion coefficients (ADC) either separated or combined.
Methods: Imaging was performed on 26 patients with pyogenic brain abscesses, 31 patients with rim-enhancing glioblastomas, and 21 patients with rim-enhancing metastases. The degree of intralesional susceptibility signal (ILSS) was independently assessed by three observers. Average ADC in the lesion core was calculated. After receiver operating characteristic (ROC) analysis, the area under the ROC curve was compared using three different analytical models (ILSS, ADC, and ILSS-ADC combined) to differentiate abscess from the two rim-enhancing necrotic tumors.
Results: The ILSS-ADC combined model had greater area under the ROC curves than ILSS or ADC used alone. In this study, the ILSS-ADC combined model showed 100% diagnostic accuracy differentiating abscesses from glioblastoma. The ADC model and the ILSS-ADC combined model performed equally well in distinguishing abscesses from metastases.
Conclusion: It is concluded that SWI and ADC are complementary, and the combination of SWI and ADC may improve results compared with the use of only one model. Validation by an independent cohort is the next necessary step to broaden its applicability in routine clinical settings.
Keywords: Apparent diffusion coefficient; Brain abscess; Glioblastoma; Intralesional susceptibility signal; Metastasis; Susceptibility weighted imaging.
Copyright © 2019 Elsevier B.V. All rights reserved.