Abstract #2552

Prediction of Glioma Patient Survival Using MRI-Derived Features and Automated Machine Learning (AutoML) Methods

Maninder Singh¹, Durgesh Kumar Dwivedi¹, B. V. Rathish Kumar², Sudhir Kumar Pathak³, Ranjeet Ranjan Jha⁴, Siddharth Singh¹, Anit Parihar¹, Chhitij Srivastava⁵, and Bal Krishna Ojha⁵

¹Department of Radiodiagnosis, King George's Medical University, Lucknow, India, ²Department of Mathematics and Statistics, Indian Institute of Technology, Kanpur, India, ³Learning Research and Development Center, University of Pittsburgh, Pittsburgh, PA, United States, ⁴Department of Mathematics, Indian Institute of Technology, Patna, India, ⁵Department of Neurosurgery, King George's Medical University, Lucknow, India

Synopsis

Keywords: Tumors (Pre-Treatment), Brain, Glioma; Automated Machine Learning (AutoML); Hyperopt-sklearn

Motivation: Predicting survival days for glioma patients is crucial for clinical decision-making and optimizing treatment strategies.

Goal(s): This study aims to improve survival prediction for glioma patients by developing optimized machine learning models using a feature set derived from MRI data.

Approach: We preprocessed the features extracted from BraTS2020 dataset, and refined to 170 using correlation analysis. We implemented Hyperopt-sklearn for AutoML-driven model selection and hyperparameter tuning to predict survival days.

Results: The kNN model achieved F1 score of greater than 0.8 and a SGDR model performs best in the 301-400 days range, demonstrating the effectiveness and efficiency of AutoML over traditional methods.

Impact: Our results show that AutoML methods like Hyperopt-sklearn can simplify and accelerate survival prediction for glioma patients, even for researchers with minimal programming skills, thereby broadening access to advanced data-driven tools in clinical research.

How to access this content:

For one year after publication, abstracts and videos are only open to registrants of this annual meeting. Registrants should use their existing login information. Non-registrant access can be purchased via the ISMRM E-Library.

After one year, current ISMRM & ISMRT members get free access to both the abstracts and videos. Non-members and non-registrants must purchase access via the ISMRM E-Library.

After two years, the meeting proceedings (abstracts) are opened to the public and require no login information. Videos remain behind password for access by members, registrants and E-Library customers.

Click here for more information on becoming a member.