The objective of this study was to prospectively verify if diffusion-weighted magnetic resonance (DwMR)–related parameters such as perfusion fraction (f) and slow diffusion coefficient (D), according to Le Bihan theory, are more effective than apparent diffusion coefficient (ADC) for classification and characterization of the more frequent focal liver lesions (FLLs) in noncirrhotic liver.
Sixty-seven patients underwent standard liver magnetic resonance imaging (MRI) and free-breath multi-b DwMR study. Two regions of interest were defined by 2 observers, including 1 FLL for each patient (21 hemangiomas, 21 focal nodular hyperplasias, 25 metastases) and part of surrounding parenchyma, respectively. For every FLL, D, f, and ADC were estimated both as absolute value and as ratio between FLL and surrounding parenchyma by fitting the reduced equation of the bicompartmental model to experimental data; t test, analysis of variance, and receiver operating characteristic analysis were performed.
t Test showed significant differences in ADClesion, f lesion, D lesion, ADCratio, and D ratio values between benign and malignant FLLs, more pronounced for ADClesion (P < 0.0009) and ADCratio (P = 0.001). Applying cutoff values of 1.55 × 10 − 3 mm2/s (ADClesion) and 0.89 (ADCratio), the DwMR study presented sensitivities and specificities, respectively, of 84% and 80% (for ADClesion), 72% and 80% (ADCratio).
Apparent diffusion coefficient (by fitting procedures) better performs than do D and f in FLL classification, especially when its values are less than 1.30 or greater than 2.00 × 10 − 3 mm2/s.
* From the Department of Experimental and Clinical Biomedical Sciences, Radiodiagnostic Unit, University of Florence, and †Health Physics Section, Azienda Ospedaliero–Universitaria Careggi, Florence; ‡Department of Neurosciences and Imaging, Section of Radiological Imaging, “G. d’Annunzio” University, Chieti; and § Department of Radiology, Spedali Civili, Brescia, Italy.
Received for publication March 1, 2013; accepted March 26, 2013.
Reprints: Stefano Colagrande, MD, Università degli Studi di Firenze, Dipartimento di Scienze Biomediche Sperimentali e Cliniche, Unità Radiodiagnostica n. 2, Azienda Ospedaliero–Universitaria Careggi, Largo Brambilla 3, Firenze, Italy 50134 (e-mail: email@example.com).
This work was partially funded by SIRM (Società Italiana Radiologia Medica).
The authors report no conflicts of interest.