2518

Stretched-Exponential Diffusion-Weighted Imaging Model for Abdominal MRI

Takeshi Yoshikawa¹, Katsusuke Kyotani², Yoshiharu Ohno¹, Yoshimori Kassai³, Seiya Kai³, Eiji Takeda², Shinichiro Seki¹, and Yuji Kishida⁴

¹Advanced Biomedical Imaging Research Center, Kobe University Graduate School of Medicine, Kobe, Japan, ²Center of Radiology and Radiation Oncology, Kobe University Hospital, Kobe, Japan, ³Toshiba Medical Systems Co., Otawara, Japan, ⁴Radiology, Kobe University Graduate School of Medicine, Kobe, Japan

Synopsis

Stretched-exponential model can be used as an excellent alternative to mono-exponential model in evaluation of abdominal organs and diseases.

Background & Purpose

Bi-exponential model can describe signal decay better than mono-exponential model, but requires many b values and longer acquisition time, causing misregistration, and makes calculation more complex.

Stretched-exponential model can quantify signals arising from a multiplicity of sources with only two parameters.

Distributed diffusion coefficients (DDC)
Water molecular diffusion heterogeneity index (alpha, α)
α=1: heterogeneous intravoxel diffusion
α=0: multi-exponential intravoxel signal decay
Equation: Sb/S0 = exp{-(b×DDC)^α}　　　 cf. Sb/S0 = exp(-b×ADC)

The purpose of this study was to assess capability of stretched-exponential model in evaluation of abdominal organs and diseases.

Methods & Materials

106 patients (68 men and 38 women, mean: 66.9years), who were suspected to have hepato-biliary-pancreatic malignancy and underwent 3T-MRI, were retrospectively analyzed. 62 malignant (HCC: 31, liver meta:8, CCC:2, bile duct Ca:4, panc Ca:10, pNET: 2, panc meta: 2, papillary Ca: 2, RCC: 1 ) and 75 benign (hepatic cyst:18, hepatic hemangioma:5, cholocyctitis:1, IPMN:15, panc cyst:8, pancreatitis:6, panc SCN:1, papillaryadenoma:1, renal cyst:18, hydronephrosis:1, low grade GIST:1) lesions were confirmed and chosen for analysis.

All patients underwent MRI at a 3T scanner (Vantage Titan 3T; Toshiba Medical Systems, Otawara, Japan). Source DWIs were obtained with SE-EPI sequence (TR/TE/FA = 3000-6000/66/90, b values: 0, 500, 1000, matrix: 96 ×128, thickness: 7mm, NEX: 2, scan time: 10-12min, PASTA+SPAIR, PI: 2, MPG: x, y, z) as one of routine sequences in our institution.

ADC, DDC, alpha images were calculated by using mono-exponential and stretched-exponential models on a workstation (Vitrea, Vital Images). Oval ROIs were placed in four liver segments, 3 pancreatic parts, spleen, gallbladder, bilateral kidneys, back muscle, and focal lesions. Maximum of 3 lesions per a patient and lesions with a diameter of >10 mm were chosen for analysis.

Mean ADC, DDC, and alpha values of each organ were calculated. Correlation coefficients among the parameters were assessed for each organ and lesion. Mean values of malignant and benign lesions were compared for each parameter. Lesion contrasts to background organs (=(lesion -background) / (lesion + background)) were calculated and compared among the parameters. Lesion characterization was compared using ROC analysis among the parameters.

Results

DDCs showed similar trends to ADCs in all organs and were significantly correlated (p<0.05, see the Figure). Alphas of gallbladder and muscle were significantly higher than other solid organs (<0.0001). Correlations of focal lesion ADCs and DDCs were significantly lower than solid abdominal organs' ones. Malignant lesion ADC, DDC, and Alpha were significantly lower than benign ones (<0.0001, <0.0001, 0.004). Malignant lesion contrast on DDC was significantly higher than that on ADC (0.027). There was no significant difference for lesion characterization between ADC and DDC.

Discussion

Stretched-exponential model enables diffusion analysis considering diffusion varieties in each voxel. DDC can be used as an excellent alternative to ADC. It has the potential to improve diagnostic performance of abdominal DWI.

This study shows initial experiences. The questions below are not answered.