Journal
MEDICAL & BIOLOGICAL ENGINEERING & COMPUTING
Volume 57, Issue 8, Pages 1683-1691Publisher
SPRINGER HEIDELBERG
DOI: 10.1007/s11517-019-01986-z
Keywords
Glioblastoma multiforme; Overall survival prediction; Machine learning; Support vector regression; Shape features
Categories
Funding
- NMRC Bedside Bench [R-397-000-245-511]
- Singapore Ministry of Health's National Medical Research Council under its Translational and Clinical Research Flagship Program-Tier 1 [NMRC/TCR/016-NNI/2016]
Ask authors/readers for more resources
Glioblastoma multiforme (GBM) are malignant brain tumors, associated with poor overall survival (OS). This study aims to predict OS of GBM patients (in days) using a regression framework and assess the impact of tumor shape features on OS prediction. Multi-channel MR image derived texture features, tumor shape, and volumetric features, and patient age were obtained for 163 GBM patients. In order to assess the impact of tumor shape features on OS prediction, two feature sets, with and without tumor shape features, were created. For the feature set with tumor shape features, the mean prediction error (MPE) was 14.6days and its 95% confidence interval (CI) was 195.8days. For the feature set excluding shape features, the MPE was 17.1days and its 95% CI was observed to be 212.7days. The coefficient of determination (R-2) value obtained for the feature set with shape features was 0.92, while it was 0.90 for the feature set excluding shape features. Although marginal, inclusion of shape features improves OS prediction in GBM patients. The proposed OS prediction method using regression provides good accuracy and overcomes the limitations of GBM OS classification, like choosing data-derived or pre-decided thresholds to define the OS groups.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available