KCR 2012








Notification of abstract acceptance announced via E-mail on July 31st.

Abstract view

Oct. 18 (Thu) Grand Ballroom 105 Room - SS11 NR

16:40-16:50 [SS 11 NR-05]
Bimodal histogram analysis of Wash-in-Emax Ratio derived from dynamic contrast-enhanced perfusion MR imaging for differentiating primary central nervous system lymphoma from glioblastoma


● Speaker	Shanshan Lu (The first affiliated hospital of Nanjing Medical University)
● Authors	Shanshan Lu,Sang Joon Kim²,Ho Sung Kim²,Choong Gon Choi²
● Affiliation	The first affiliated hospital of Nanjing Medical University¹,Asan Medical Center²
● PDF View	SS 11 NR-05.pdf

PURPOSE:
To investigate the diagnostic value of bimodal histogram annalysis of Wash-in-E_max ratio (WER) derived from dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) in differentiating treatment-naive primary central nervous system lymphoma (PCNSL) and glioblastoma.

MATERIALS AND METHODS:
Thirty-one glioblastoma and eleven PCNSL patients were respectively recruited in this study. All of them were histologically confirmed and treatment-naive.Normalized model-free DCE maps were calculated by dividing wash-in value by E_max value on a voxel by voxel basis. Whole enhancing tumor WER histogramswere calculated from normalized DCE maps with a two-component normal distribution mixture fitting curve. Histogram parameters of WER, including mean WER at higher curve (mWER_H), skewness, kurtosis, standard deviation (SD), and three cumulative histogram parameters (WER₅₀, WER₇₅, and WER₉₀), were generated and compared. The classifiers were selected by t-test and receiver operating characteristic (ROC) curves.

RESULTS:
Significant differences between PCNSL and glioblastomas were found in skewness (P < 0.001), SD (P < 0.05) and WER₉₀ (P < 0.05). Bimodal histogram of 90% (n = 28/31) glioma patients demonstrated positive skew while that of 64% (n = 7/11) PCNSL patients was negative skew.SD and WER₉₀ of glioblastoma were higher than that of PCNSL patients. The mWER_Hof glioblastoma was higher than PCNSL patients but without significant difference (P >0.05). ROC curve analyses showed skewness was the best classifier for differentiating PCNSL from glioblastoma, with a sensitivity of 77%, a specificity of 100% and area under curve (AUC) of 0.918.

CONCLUSIONS:
Bimodal histogram analysis of WER derived from DCE-MRI is effective for differentiation of treatment-naive PCNSL and glioblastoma. Skewness is the best predictor.