3D_UNet3+: An Optimized Approach for 3D Brain Tumor Segmentation

The demand for sophisticated instruments in the detection of brain cancers is growing very significant. This work presents the 3D_UNet3+, a novel model particularly developed for the purpose of segmenting brain tumor pictures. The 3D_UNet3+ distinguishes itself from typical models by employing a distinctive approach in its filter size and construction. The architecture we have developed plays a crucial role in simplifying the segmentation process and minimizing the computing burden. This makes our model very suitable for practical applications in medical imaging. The 3D_UNet3+ exhibits its potential in the field when assessed using the BraTS2020 dataset. Furthermore, it shows great potential for advancing brain tumor diagnosis and treatment planning.This model has successfully achieved a significant decrease in computational requirements, as demonstrated by a 98.66% reduction in model parameter size and a 98.1% decrease in FLOPS. This highlights its enhanced effectiveness and makes it suitable for integration into advanced technologies such as edge AI and handheld devices.