Aim

To evaluate apparent diffusion coefficient (ADC) derived histogram parameters for lesion characterization in whole-body diffusion-weighted imaging (WB-DWI) of lymphoma.

Methods

Fifteen patients with histopathology proven lymphoma (11 Non-Hodgkin; 4 Hodgkin lymphomas) underwent WB-DWI using 2 b-values (0-1000 s/mm2). On coronal reformatted b1000 WB-DWI images, regions of interest (ROI) were drawn semi-automatically on lymph nodes in all nodal stations (n=267) and in axial and appendicular bone regions (n=53). For each ROI, a histogram was constructed from which volume, mean(ADC), median(ADC), skewness(ADC), and kurtosis(ADC) were calculated. Mann-Whitney-U tests were performed to detect significant differences between malignant and benign ROIs per tissue type. Receiver-operating-characteristic curves (ROC) were constructed from which an optimal threshold was determined as well as sensitivity, specificity and accuracy. PET/CT plus bone marrow biopsy (BMB) served as reference standard.

Results

All parameters were significantly different between malignant and benign lymph nodes (p<0.001) with skewness(ADC) being the most accurate. A positive skewness exceeding 0.3041 mm2/s allowed for detection of malignant lymph nodes with 88% accuracy, 88% sensitivity and 87% specificity compared to 63% accuracy, 61% sensitivity and 64% specificity for mean(ADC). Only kurtosis(ADC) (p<0.001) and skewness(ADC) (p=0.003) were significantly different between malignant bone marrow infiltration and normal bone marrow. Kurtosis(ADC) showed highest accuracy and a threshold exceeding 5.26 allowed for detection of malignant bone marrow infiltration with 89% accuracy, 86% sensitivity and 90% specificity.

Conclusions

ADC histogram analysis is feasible for lesion characterization in WB-DWI of lymphoma. Lymph nodes were most accurately characterized using skewness(ADC) and bone tissue using kurtosis(ADC).