Initial Thoughts on the dataset BraTS

link to the dataset: http://braintumorsegmentation.org/

kaggle link to dataset (2017) : https://www.kaggle.com/xxc025/unet-datasets

Google Colab implementation: https://colab.research.google.com/drive/1Bygebxg8NSNiCASi4XJdPLM03fR6SF3M?usp=sharing

Github reference: https://github.com/cv-lee/BraTs

Gliomas are the most common brain tumors that arise from glial cells. They can be categorized into two basic grades: low-grade gliomas (LGG) that tend to exhibit benign tendencies and indicate a better prognosis for the patient, and high-grade gliomas (HGG) that are malignant and more aggressive.

Brain tumors can be imaged by various Magnetic Resonance (MR) sequences. Different sequences can provide complementary information to analyze different subregions of gliomas.

1. T1-weighted
2. Contrast-enhanced T1-weighted (T1c)
3. T2-weighted
4. Fluid Attenuation Inversion Recovery (FLAIR) images.

→ T2 and FLAIR highlight the tumor with peritumoral edema designated “the whole tumor”

→ T1 and T1c highlight the tumor without peritumoral edema, designated “tumor core”

→ Enhancing region of the tumor core with hyper-intensity can be observed in T1c, designated “enhancing tumor core”

The segmentation task is challenging because

→ the size, shape, and localization of brain tumors have considerable variations among patients. This limits the usability and usefulness of prior information about shape and location.

→ The boundaries between adjacent structures are often ambiguous due to the smooth intensity gradients, partial volume effects, and bias field artifacts.