1076. Comparative Analysis of Pre- and Post-Treatment MRI Texture Features in Pediatric Diffuse Midline Glioma Patients for Survival Prediction
Authors * Denotes Presenting Author
  1. Bulent Aslan *; Beth Israel Deaconess Medical Center
  2. Erhan Biyikli; Marmara University
  3. Dilek Gul; Marmara University
  4. Nursah Eker; Marmara University
  5. Beste Atasoy; Marmara University
  6. Gulnur Tokuc; Marmara University
  7. Feyyaz Baltacioglu; American Hospital
The aim of our study was to predict survival in pediatric patients with diffuse midline glioma (DMG) by MRI texture analysis before and after radiation and temozolomide therapy.

Materials and Methods:
Thirty-five children with DMG were included in our study. The patients were divided into two groups according to a median survival time of 13 months as short and long survivors. Pre- and post-treatment MRI texture analysis features of the lesions in both groups were compared.

Twenty-one male and 14 female patients were included in this research. The median age was 8 (range: 1-17) years. Four patients were alive at the time of analysis. In the evaluation made according to those who live less than 13 months, which was the median life expectancy, statistically significant texture analysis parameters were pre-treatment Perc.90, DifVarnc, Contrast on ADC, SumAverg on T1, post-treatment SumEntropy on T1, pre- and post-treatment difference Perc.99, DifVarnc on ADC, SumEntropy on T1, AngScMom and SumEntropy on T2. When the patients were grouped according to median survival time, the only statistically significant parameter in the Cox regression analysis performed to investigate the effect of texture analysis parameters in predicting survival was post-treatment SumEntropy on T1.

Our study highlights the potential of MRI texture analysis as a noninvasive and quantitative imaging biomarker for survival prediction in patients with DMG undergoing chemoradiotherapy.