Texture, Morphology, and Statistical Analysis to Differentiate Primary Brain Tumors on Two-Dimensional Magnetic...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Primary brain tumors originate from brain cells due to DNA errors.
• Early diagnosis is crucial for effective treatment of brain tumors.

Purpose of the Study:

• To analyze 2D texture, morphology, and statistical features of brain tumors.
• To classify primary brain tumors using artificial intelligence (AI) techniques.

Main Methods:

• Utilized deep learning (DL) and machine learning (ML) for classification of glioma, meningioma, and pituitary tumors.
• Analyzed T1-weighted MRI 2D scans for multiclass and binary classification.
• Selected the 20 most significant features from 102 calculated features using a three-step selection process.

Main Results:

• A long short-term memory binary classification model (glioma vs. meningioma) demonstrated superior performance.
• Achieved an average accuracy of 97.7%, recall of 97.2%, precision of 97.5%, F1-score of 97.0%, and kappa coefficient of 94.7%.

Conclusions:

• The developed AI method enables effective early diagnosis of primary brain tumors.
• Accurate classification of tumor types is presented as a key to timely and effective treatment.