Three-Phase Automatic Brain Tumor Diagnosis System Using Patches Based Updated Run Length Region Growing Technique

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Neuroscience

Background:

• Manual segmentation of brain tumors in MRI is time-consuming and challenging due to variations in tissue appearance.
• Accurate tumor detection and segmentation are crucial for diagnosis and treatment planning.

Purpose of the Study:

• To develop an automated three-phase system for brain tumor diagnosis, segmentation, and visualization in MRI human head volumes.
• To improve the efficiency and accuracy of brain tumor analysis compared to manual methods.

Main Methods:

• A patch-based updated run length region growing (PR2G) technique was employed, involving automatic classification, segmentation, and 3D volume construction.
• Support Vector Machine (SVM) classifier with 8x8 patches and infinite feature selection (IFS) were used for tumorous slice detection.
• Run length region growing was applied for tumor segmentation, followed by 3D tumor volume estimation using Carelieri's estimator.

Main Results:

• The PR2G system achieved 97% classification accuracy in distinguishing normal from tumorous slices.
• Segmentation accuracy was measured using Dice similarity, positive predictive value (PPV), sensitivity, and accuracy, with a Dice similarity of 80% reported.
• Datasets from the whole brain atlas (WBA) and BraTS repositories were utilized for experimental validation.

Conclusions:

• The proposed automated PR2G system effectively detects and segments brain tumors in MRI scans.
• The system demonstrates high classification and segmentation accuracy, offering a promising tool for clinical applications in neuro-oncology.