Open AccessDiffusion-weighted MRI in adrenal lesions: A warranted adjunct?
Author(s) -
Mona El-Kalioubie,
Sally EmadEldin,
Omar Abdelaziz
Publication year - 2016
Publication title -
the egyptian journal of radiology and nuclear medicine /the egyptian journal of radiology and nuclear medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.19
H-Index - 13
eISSN - 2090-4762
pISSN - 0378-603X
DOI - 10.1016/j.ejrnm.2016.01.003
Subject(s) - medicine , malignancy , lesion , receiver operating characteristic , diffusion mri , effective diffusion coefficient , radiology , nuclear medicine , predictive value , diagnostic accuracy , magnetic resonance imaging , pathology
AbstractPurposeWe assessed the value of adding Diffusion-weighted (DW) sequences to conventional MR in the diagnostic adrenal algorithm, particularly in patients with indeterminate adrenal mass. We compared the predictive capabilities of apparent diffusion coefficient values (ADC), lesion size and signal drop on chemical shift MRI (CS MRI) for differentiating benign and malignant lesions.Materials and methodsFifty-five patients with 59adrenal lesions were evaluated. Lesion size, signal loss on CSMRI and ADCs were assessed. Sensitivity, specificity and accuracy were calculated for each parameter. We compared the predictive capability of ADC, CS signal drop, and lesion size for malignancy with receiver operating characteristic (ROC) curves.ResultsConsidering all lesion subtypes, ADC values showed overlap. Benign and malignant masses had statistically different mean ADCs, with greater mean ADC values in benign lesions (1.78±0.5 versus 0.97×10−3mm2/s±0.3). ADC had the lowest accuracy (71.2%) but highest specificity (100%). Lack of signal drop had the greatest sensitivity (100%); lesion size was the most accurate (88.1%). On ROC curves, ADC had the lowest prediction for malignancy.ConclusionLesion size and signal drop were stronger predictors of adrenal malignancy than ADC. ADC shows poor significance in individual cases due to overlapping values