Towards Reinforced Brain Tumor Segmentation on MRI Images Based on Temperature Changes on Pathologic Area

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Computational Biology

Background:

• Brain tumor segmentation from MRI is crucial for clinical diagnosis and treatment planning but remains challenging due to irregular tumor shapes and indistinct boundaries.
• Tumor cells exhibit higher temperatures than normal brain tissue, presenting a potential source of information for segmentation.
• Current segmentation methods can produce false positives and negatives, necessitating improved techniques.

Purpose of the Study:

• To investigate the utility of thermal information from brain tumors to enhance segmentation accuracy in MRI images.
• To reduce false positive and false negative segmentation results by incorporating thermal data.
• To demonstrate that thermal properties of tumors can improve delineation in diagnostic imaging.

Main Methods:

• Simulated brain temperature distribution using Pennes' bioheat equation solved via the finite difference method.
• Introduced ±2% Gaussian noise to simulated temperatures to mimic real-world conditions.
• Applied Canny edge detection to thermal maps to identify tumor contours based on temperature gradients.
• Compared the proposed thermal-based segmentation method against the Chan-Vese level set method using T1-contrast-enhanced and Flair MRI data from phantom and synthetic patients (BRATS 2012/2013).

Main Results:

• The proposed method demonstrated significant improvements in brain tumor segmentation compared to the level set method.
• Segmentation using thermal images alone differentiated an average of 0.8% of the tumor area and 2.48% of healthy tissue.
• The thermal information effectively reduced segmentation errors, enhancing the precision of tumor boundary identification.

Conclusions:

• Tumor contour delineation can be significantly enhanced by exploiting temperature changes within brain tumors.
• Thermal information derived from MRI provides a valuable complementary data source for improving segmentation algorithms.
• This approach holds promise for reinforcing diagnostic accuracy in brain tumor imaging.