{"id":21741374,"url":"https://github.com/keplerc/mri-skull-stripping","last_synced_at":"2026-05-19T15:32:43.064Z","repository":{"id":97925891,"uuid":"133207303","full_name":"KeplerC/MRI-Skull-Stripping","owner":"KeplerC","description":"A series of machine learning models that preprocess MRI images by removing unnecessary brain structrues","archived":false,"fork":false,"pushed_at":"2018-06-08T23:40:17.000Z","size":18825,"stargazers_count":1,"open_issues_count":0,"forks_count":1,"subscribers_count":1,"default_branch":"master","last_synced_at":"2025-01-25T21:55:33.587Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":"http://kychen.xyz/2018/06/02/MRI-2018/","language":"Jupyter Notebook","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/KeplerC.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2018-05-13T05:06:17.000Z","updated_at":"2021-05-16T15:35:44.000Z","dependencies_parsed_at":null,"dependency_job_id":"6023506a-e50a-4323-9c65-c36fcc219997","html_url":"https://github.com/KeplerC/MRI-Skull-Stripping","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/KeplerC%2FMRI-Skull-Stripping","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/KeplerC%2FMRI-Skull-Stripping/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/KeplerC%2FMRI-Skull-Stripping/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/KeplerC%2FMRI-Skull-Stripping/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/KeplerC","download_url":"https://codeload.github.com/KeplerC/MRI-Skull-Stripping/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":244717341,"owners_count":20498284,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2024-11-26T06:18:30.777Z","updated_at":"2026-05-19T15:32:38.020Z","avatar_url":"https://github.com/KeplerC.png","language":"Jupyter Notebook","funding_links":[],"categories":[],"sub_categories":[],"readme":"# MRI Skull Stripping\n\nKaiyuan Chen(chenkaiyuan@ucla.edu)\n\nJingyue Shen(brianshen@ucla.edu)\n\nMore documents about this project is on http://kychen.xyz/2018/06/02/MRI-2018/. \n\n### Introduction \n\nPlease refer to\nhttp://kychen.xyz/2018/05/16/skullreview-2018/\nfor a comprehensive review on\n* what is MRI\n* why do we need skull stripping\n* how does this work\n* a review of recent works\n* bridging skull stripping to machine learning\n* a plan of this work\n\n### Outline\n\n1) The baseline model should be sklearn-based supervised learning. These models are very easy to implement, so I will try out different supervised learning approach like random forest, decision trees and other linear models to see which one works better. \n\n2) Autoencoder. We plan to stack CNNs and calculate RMSE on recovered image in Tensorflow. \n\n3) Other Compressed Sensing approaches. Other compressed sensing generative models like VAE. \n\n### I/O\n#### Feature Selection\n\nFor sklearn models, we choose \nP(this pixel should be removed | (position x, position y), color, surrounding pixels as a local patch) \nand currently, because of the limitation of my computer memory, the patch size is 4 * 4. One can update it by changing m in my source code. \n\nFor CNN Autoencoder model, we feed the entire image to CNN, since we think dealing with matrices of size 256x256 should work fine on a laptop computer. Due to the limited amount of data (about 660 images of brains), we currently set batch size = 15\nand epoch time = 100 for training. One can modify the configuration by changing the hyperparameters in code/config.py.\n\n#### Preprocessing\nFor sklearn models,  we perform a comparison on local patch or other features.\n\nFor CNN model, we now simply divide each image's pixel value by 255. We are going to explore batch normalization technique later.\n\n### Code Snippets \n\nFor all baseline codes and experiments, you can go to a blog post(http://kychen.xyz/2018/05/16/jpskull-2018/) or the **jupyter notebook** in the ./code. \n\nFor CNN Autoencoder code, you can find it in the ./code folder in this repository.\n\n### Dates \n\n2018-5-28 Finished Report\n\n2018-5-26 Finished CNN Autoencoder models\n\n2018-5-16 Finished up other sklearn models in jupyternotebook\n\n2018-5-13 Finished Baseline model(random forest)\n\n2018-5-12 Creating Starter codes \n\n2018-5-4 Writing Introduction/Literature Reviews \n\n2018-5-1 Selecting Topics \n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fkeplerc%2Fmri-skull-stripping","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fkeplerc%2Fmri-skull-stripping","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fkeplerc%2Fmri-skull-stripping/lists"}