An ensemble learning approach for brain cancer detection exploiting radiomic features

Area of Science:

• Medical Imaging Analysis
• Machine Learning in Oncology
• Radiomics

Background:

• Brain cancer is highly aggressive, with a 70% mortality rate.
• Early detection of brain tumors is crucial for improving survival rates.
• Brain cancers are graded I-IV based on cell appearance.

Purpose of the Study:

• To develop a method for recognizing different brain cancer grades using MRI.
• To improve the accuracy of brain tumor grading through advanced analysis.

Main Methods:

• Proposed an ensemble learner method for brain cancer grade discrimination.
• Utilized non-invasive radiomic features from five categories.
• Evaluated feature effectiveness using statistical tests and performance analysis.

Main Results:

• Tested on 111,205 brain MRIs from two public datasets.
• Achieved 99% accuracy in detecting Grade I (benign) and Grades II, III, IV (malignant) brain cancers.
• Demonstrated high performance in classifying brain tumor grades.

Conclusions:

• The developed ensemble learner outperforms current state-of-the-art methods.
• The proposed method shows significant potential for accurate brain cancer grade detection.
• Magnetic resonance imaging analysis is effective for non-invasive tumor grading.