Use of Capsule network to perform brain tumor segmentation using Brats2019 dataset

I am working on brats2019 dataset and task is to segment tumor regions using capsule network model .I train my capsule model but dice score is low .i need a freelancer who can improve dice score upto 0.90 for brain tumor class ET, 0.90 for WT,0.90 for TC separately and hausdorf distance of 0.324 on testing dataset. And freelancer should use tensorflow, keras libraries for building model and use google colab environment and must use brats2019 dataset. And in the last also save trained model in separate file and display one segmented image with different colors showing 3 clases of tumors. if freelancer not achieve desired results as i mentioned earlier (0.90 dice score ) i will not pay any amount to freelancer.