3D brain slice classification and feature extraction using Deformable Hierarchical Heuristic Model.

Area of Science:

• Medical Imaging
• Artificial Intelligence in Oncology
• Computational Pathology

Background:

• Brain tumors represent a critical oncological challenge with poor prognoses in advanced stages.
• Effective treatment planning is essential for improving patient survival and quality of life.
• Current imaging techniques like CT, MRI, and ultrasound aid in brain tumor assessment.

Purpose of the Study:

• To develop and validate a novel technique for extracting and classifying brain tumor features from 3D slice images.
• To enhance the accuracy of brain tumor identification and characterization for improved therapeutic strategies.

Main Methods:

• Preprocessing of 3D brain images including noise removal, resizing, and smoothening.
• Extraction of tumor features using the Volume of Interest (VOI) method.
• Classification of extracted features using the Deformable Hierarchical Heuristic Model-Deep Deconvolutional Residual Network (DHHM-DDRN), analyzing surfaces, curves, and geometric patterns.

Main Results:

• The proposed DHHM-DDRN approach achieved a classification accuracy of 95%.
• Performance metrics included a Dice Similarity Coefficient (DSC) of 83%, precision of 80%, and recall of 85%.
• An F1 score of 55% was obtained for classifying brain cancer features.

Conclusions:

• The novel technique demonstrates high accuracy and effectiveness in classifying brain tumor features from 3D images.
• This method holds potential for improving the diagnostic accuracy and treatment planning for brain cancer patients.
• Further research can explore integration into clinical workflows for enhanced brain tumor management.