Abstract
Many studies have built machine-learning (ML)-based prognostic models for glioblastoma (GBM) based on radiological features. We wished to compare the predictive performance of these methods to human knowledge-based approaches. 404 GBM patients were included (311 discovery and 93 validation). 16 morphological and 28 textural descriptors were obtained from pretreatment volumetric postcontrast T1-weighted magnetic resonance images. Different prognostic ML methods were developed. An optimized linear prognostic model (OLPM) was also built using the four significant non-correlated parameters with individual prognosis value. OLPM achieved high prognostic value (validation c-index = 0.817) and outperformed ML models based on either the same parameter set or on the full set of 44 attributes considered. Neural networks with cross-validation-optimized attribute selection achieved comparable results (validation c-index = 0.825). ML models using only the four outstanding parameters obtained better results than their counterparts based on all the attributes, which presented overfitting. In conclusion, OLPM and ML methods studied here provided the most accurate survival predictors for glioblastoma to date, due to a combination of the strength of the methodology, the quality and volume of the data used and the careful attribute selection. The ML methods studied suffered overfitting and lost prognostic value when the number of parameters was increased.
Highlights
Many studies have built machine-learning (ML)-based prognostic models for glioblastoma (GBM) based on radiological features
We constructed ML models using some of the best methods available: artificial neural networks (ANN), support vector machines (SVM), and regression trees (RT)
Thresholds which best split the sample into significant subsets in terms of c-index were: 65 years for age (p-value < 0.001, hazard ratio (HR) = 2.086, c-index = 0.706), 18 cm[3] for CE volume (p-value = 0.062, HR = 1.250, c-index = 0.556), 0.416 cm for CE rim width (p-value = 0.013, HR = 1.378, c-index = 0.590), 5.0405 cm for maximum tumor diameter (p-value = 0.063, HR = 1.246, c-index = 0.572) and 0.509 for surface regularity (p-value = 0.002, HR = 1.476, c-index = 0.578)
Summary
Many studies have built machine-learning (ML)-based prognostic models for glioblastoma (GBM) based on radiological features. Machine learning (ML) techniques have been increasingly used by the radiological research community[4,5] to construct such models[6,7] These methods, when used on sufficiently large data sets, are able to extract hidden information and patterns from data, automatically learning and being able to make predictions about future system behavior. In this study we developed efficient, optimized predictive models using clinical data and high-quality MRI-based morphological information of GBM patients. Our intention was to compare the OLPM with the ML approaches and the best ML-based models recently proposed in the literature to construct accurate prognostic estimators and to show how a non-rigorous use of ML methods in neuro-oncology can lead to misleading results
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
