Challenges in Building of Deep Learning Models for Glioblastoma Segmentation: Evidence from Clinical Data

Anvar Kurmukov, Aleksandra Dalechina, Talgat Saparov, Mikhail Belyaev, Svetlana Zolotova, Andrey Golanov, Anna Nikolaeva

Research output: Contribution to journalArticlepeer-review

Abstract

In this article, we compare the performance of a state-of-the-art segmentation network (UNet) on two different glioblastoma (GB) segmentation datasets. Our experiments show that the same training procedure yields almost twice as bad results on the retrospective clinical data compared to the BraTS challenge data (in terms of Dice score). We discuss possible reasons for such an outcome, including inter-rater variability and high variability in magnetic resonance imaging (MRI) scanners and scanner settings. The high performance of segmentation models, demonstrated on preselected imaging data, does not bring the community closer to using these algorithms in clinical settings. We believe that a clinically applicable deep learning architecture requires a shift from unified datasets to heterogeneous data.

Original languageEnglish
Pages (from-to)298-302
Number of pages5
JournalStudies in health technology and informatics
Volume281
DOIs
Publication statusPublished - 27 May 2021

Keywords

  • clinical data
  • Deep learning
  • glioblastoma
  • segmentation

Fingerprint

Dive into the research topics of 'Challenges in Building of Deep Learning Models for Glioblastoma Segmentation: Evidence from Clinical Data'. Together they form a unique fingerprint.

Cite this