DeepGI is an automated approach for segmenting gastrointestinal (GI) tract regions in magnetic resonance imaging (MRI) scans. The method integrates Inception-V4 for initial classification, UNet++ with VGG19 encoder for 2.5D data, and Edge UNet for grayscale data segmentation. The model addresses the manual and time-consuming segmentation process in radiotherapy planning, offering a unified solution that captures intricate anatomical details. The integration of diverse architectures enhances the model's accuracy, robustness, and adaptability. The methodology includes meticulous data preprocessing, incorporating 2.5D and grayscale processing to improve performance. The model's predictions are combined through an ensemble approach, leveraging the strengths of different architectures for accurate GI tract segmentation. The model's performance is evaluated using metrics such as Dice Coefficient and 3D Hausdorff Distance, with experimental results showing that Edge UNet excels in grayscale segmentation and UNet++ with VGG19 performs well in 2.5D segmentation. The proposed model provides an efficient and accurate tool for clinicians, significantly improving the efficiency and accuracy of radiotherapy planning. The work contributes to the field of medical image segmentation by introducing a novel approach that addresses the challenges of GI tract segmentation, offering a promising solution for enhancing radiotherapy planning efficiency.DeepGI is an automated approach for segmenting gastrointestinal (GI) tract regions in magnetic resonance imaging (MRI) scans. The method integrates Inception-V4 for initial classification, UNet++ with VGG19 encoder for 2.5D data, and Edge UNet for grayscale data segmentation. The model addresses the manual and time-consuming segmentation process in radiotherapy planning, offering a unified solution that captures intricate anatomical details. The integration of diverse architectures enhances the model's accuracy, robustness, and adaptability. The methodology includes meticulous data preprocessing, incorporating 2.5D and grayscale processing to improve performance. The model's predictions are combined through an ensemble approach, leveraging the strengths of different architectures for accurate GI tract segmentation. The model's performance is evaluated using metrics such as Dice Coefficient and 3D Hausdorff Distance, with experimental results showing that Edge UNet excels in grayscale segmentation and UNet++ with VGG19 performs well in 2.5D segmentation. The proposed model provides an efficient and accurate tool for clinicians, significantly improving the efficiency and accuracy of radiotherapy planning. The work contributes to the field of medical image segmentation by introducing a novel approach that addresses the challenges of GI tract segmentation, offering a promising solution for enhancing radiotherapy planning efficiency.