Triexponential function analysis of diffusion-weighted MRI for diagnosing prostate cancer

• Published in: Journal of magnetic resonance imaging 2016-01, Vol.43 (1), p.138-148
• Main Authors: Ueda, Yu, Takahashi, Satoru, Ohno, Naoki, Kyotani, Katsusuke, Kawamitu, Hideaki, Miyati, Tosiaki, Aoyama, Nobukazu, Ueno, Yoshiko, Kitajima, Kazuhiro, Kawakami, Fumi, Okuaki, Tomoyuki, Tsukamoto, Ryuko, Yanagita, Emmy, Sugimura, Kazuro
• Format: Article
• Language: English
• Subjects: Aged; Algorithms; apparent diffusion coefficient (ADC); Confidence intervals; Diffusion; Diffusion Magnetic Resonance Imaging - methods; diffusion weighted imaging (DWI); Humans; Image Enhancement - methods; Image Interpretation, Computer-Assisted - methods; intravoxel incoherent motion (IVIM); Magnetic resonance imaging; Male; Medical imaging; Numerical Analysis, Computer-Assisted; perfusion; prostate; Prostate cancer; Prostatic Neoplasms - pathology; Reproducibility of Results; Sensitivity and Specificity; triexponential analysis
• ISSN: 1053-1807; 1522-2586
• DOI: 10.1002/jmri.24974

Background To evaluate more detailed information noninvasively through on diffusion and perfusion in prostate cancer (PCa) using triexponential analysis of diffusion‐weighted imaging (DWI). Methods Sixty‐three prostate cancer patients underwent preoperative 3.0 Tesla MRI including eight b‐values DWI. Triexponential analysis was performed to obtain three diffusion coefficients (Dp, Df, Ds), as well as fractions (Fp, Ff, Fs). Each diffusion parameter for cancerous lesions and normal tissues was compared and the relationship between diffusion parameters and Gleason score (GS) was assessed. Ktrans, Ve, and the ratios of intracellular components measured in histopathological specimens were compared with diffusion parameters. Results Dp was significantly greater for cancerous lesions than normal peripheral zone (PZ) (P < 0.001), whereas Dp in transition zone (TZ) showed no significant difference (P = 0.74, 95% confidence interval (CI) = −4.69–6.48). Ds was significantly smaller for each cancerous lesions in PZ and TZ (P < 0.001, respectively). There was no significant difference in Df between cancerous lesions and normal tissues in PZ and TZ (P = 0.07, 95% CI = −0.29–0.12 and P = 0.53, 95% CI = −3.51–2.29, respectively). D obtained with biexponential analysis were significantly smaller in cancerous lesions than in normal tissue in PZ and TZ (P < 0.001 for both), while D* in PZ and TZ showed no significant difference (P = 0.14, 95% CI = −1.60–0.24 and P = 0.31, 95% CI = −3.43–1.16, respectively). Dp in PZ and TZ showed significant correlation with Ktrans (R = 0.85, P < 0.001; R = 0.81, P < 0.001, respectively), while D* in PZ obtained with biexponential analysis showed no such correlation (P = 0.08, 95% CI = −0.14–0.30). Fs was significantly correlated with intracellular space fraction evaluated in histopathological specimens in PZ and TZ cancer (R = 0.41, P < 0.05; R = 0.59, P < 0.001, respectively). Ff and Fs correlated significantly with GS in PZ and TZ cancer (PZ: R = −0.44, P < 0.05; R = 0.37, P < 0.05, TZ: R = −0.59, P < 0.05; R = 0.57, P < 0.05, respectively). Conclusion Triexponential analysis is a noninvasive approach that can provide more detailed information regarding diffusion and perfusion of PCa than biexponential analysis. J. MAGN. RESON. IMAGING 2016;43:138–148.