3D convolutional neural networks for outcome prediction in glioblastoma using methionine PET and T1w MRI

Iram Shahzadi, Annekatrin Seidlitz, Alex Zwanenburg, Bettina Beuthien-Baumann, Ivan Platzek, Jörg Kotzerke, Michael Baumann, Mechthild Krause, Steffen Löck

Keywords: Glioblastoma, survival analysis, 3D-CNN, biomarker

TL;DR: In this study, we used 3D-CNNs to develop imaging biomarker integrated with clinical data to predict time to recurrence and overall survival in patients with glioblastoma.

Abstract:

For treatment personalization of patients with glioblastoma, we investigate three different
3D convolutional neural networks (3D-CNN) for predicting time to recurrence (TTR)
and overall survival (OS) from postoperative [11C] methionine PET (MET-PET) and
gadolinium-enhanced T1-weighted magnetic resonance imaging (T1c-w MRI). The 3D-DenseNet
model on MET-PET integrated with age and MGMT status achieved the best
performance on independent test data (Concordance-Index: TTR=0.68, OS=0.65) with significant
patient stratification (p-value: TTR=0.017, OS=0.039). After prospective validation,
these models may be considered for treatment personalization.

Registration: I acknowledge that acceptance of this work at MIDL requires at least one of the authors to register and present the work during the conference.

Authorship: I confirm that I am the author of this work and that it has not been submitted to another publication before.

Paper Type: novel methodological ideas without extensive validation

Primary Subject Area: Application: Radiology

Secondary Subject Area: Validation Study

Confidentiality And Author Instructions: I read the call for papers and author instructions. I acknowledge that exceeding the page limit and/or altering the latex template can result in desk rejection.