The dataset used is the OASIS MRI dataset (https://sites.wustl.edu/oasisbrains/), which consists of 80,000 brain MRI images. The images have been divided into four classes based on Alzheimer's progression. The dataset aims to provide a valuable resource for analyzing and detecting early signs of Alzheimer's disease.

To make the dataset accessible, the original .img and .hdr files were converted into Nifti format (.nii) using FSL (FMRIB Software Library). The converted MRI images of 461 patients have been uploaded to a GitHub repository, which can be accessed in multiple parts.

For the neural network training, 2D images were used as input. The brain images were sliced along the z-axis into 256 pieces, and slices ranging from 100 to 160 were selected from each patient. This approach resulted in a comprehensive dataset for analysis.

Patient classification was performed based on the provided metadata and Clinical Dementia Rating (CDR) values, resulting in four classes: demented, very mild demented, mild demented, and non-demented. These classes enable the detection and study of different stages of Alzheimer's disease progression.

During the dataset preparation, the .nii MRI scans were converted to .jpg files. Although this conversion presented some challenges, the files were successfully processed using appropriate tools. The resulting dataset size is 1.3 GB.

With this comprehensive dataset, the project aims to explore various neural network models and achieve optimal results in Alzheimer's disease detection and analysis.

Acknowledgments: “Data were provided 1-12 by OASIS-1: Cross-Sectional: Principal Investigators: D. Marcus, R, Buckner, J, Csernansky J. Morris; P50 AG05681, P01 AG03991, P01 AG026276, R01 AG021910, P20 MH071616, U24 RR021382”

Citation: OASIS-1: Cross-Sectional: https://doi.org/10.1162/jocn.2007.19.9.1498

If you are looking for processed NifTi image version of this dataset please click here

The dataset in the zip file utilized is the OASIS MRI dataset, encompassing 9488 brain MRI images categorized into four classes based on Alzheimer's progression. The primary goal of this dataset is to offer a valuable resource for analyzing and identifying early indicators of Alzheimer's disease. To enhance accessibility, the original .img and .hdr files underwent conversion to Nifti format (.nii) through FSL (FMRIB Software Library). For neural network training, 2D images served as input. Brain images were sliced along the z-axis into 256 pieces, with slices ranging from 100 to 160 selected from each patient. This methodology resulted in a comprehensive dataset conducive to analysis. Patient classification relied on provided metadata and Clinical Dementia Rating (CDR) values, yielding four classes: demented, very mild demented, mild demented, and non-demented. These classes facilitate the examination of various stages of Alzheimer's disease progression. During dataset preparation, the .nii MRI scans were converted to .jpg files. With this extensive dataset, the project aims to explore diverse neural network models and achieve optimal outcomes in Alzheimer's disease detection and analysis. This dataset is user-friendly and easy to manage. If you require additional samples from the OASIS Alzheimer dataset, please follow the provided link below.

credits: https://www.kaggle.com/datasets/ninadaithal/imagesoasis

OASIS is a project aimed at making neuroimaging data sets of the brain freely available to the scientific community. OASIS currently contains 4 datasets (OASIS-1, OASIS-2, OASIS-3, and OASIS-4). OASIS-2 is the "Longitudinal MRI Data in Nondemented and Demented Older Adults" dataset and includes 150 participants aged 60 to 96 years, many of whom were diagnosed with Alzheimer's disease at some point during their participation. Participants were obtained from the longitudinal pool of the Washington University Alzheimer Disease Research Center (ADRC). Each participant was scanned on two or more visits, separated by at least one year, for a total of 373 imaging sessions.

This Dataset consists of OASIS-1 Subject Scans

Detailed preprocessing https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F14462934%2F5021ddb1765cd2fcc88922b80105e47b%2Fpreprocessing.png?generation=1737366916048980&alt=media" alt="">

Link to the CSV file with demographics and metadata https://sites.wustl.edu/oasisbrains/files/2024/04/oasis_cross-sectional-5708aa0a98d82080.xlsx

Please go through the attached PDF file before you begin. There are 2 scans for 20 non-demented participants and they have "MR2" in their subject ID

If you are using the data for your analysis, please cite and credit the original data creation authors Open Access Series of Imaging Studies (OASIS): Cross-Sectional MRI Data in Young, Middle Aged, Nondemented, and Demented Older Adults. Marcus, DS, Wang, TH, Parker, J, Csernansky, JG, Morris, JC, Buckner, RL. Journal of Cognitive Neuroscience, 19, 1498-1507. doi: 10.1162/jocn.2007.19.9.1498

Based on the OASIS-1 Alzheimer's Dataset with 416 subjects.

Most public Alzheimer's datasets exhibit training data leakage, since MRI slices from the same patient are randomly allocated in the test/train split. Due to class imbalance, it become difficult to evenly split the dataset with scikit-learn libraries. This dataset randomly choses patients from each class and allocates them into the training set at an amount deemed sufficient to learn features from. Due to the massive imbalance caused by the "NonDemented" class, which has more images than some other classes combines, we put the majority of these patients in the test set. It is not necessary to have so many examples for successful training.

Acknowledgements: "Yiwei Lu has performed image conversion along with skull striping and other tissue removal with their pre-trained LinkNet3D model."

Citation: Well-Documented Alzheimer's Dataset: https://doi.org/10.34740/kaggle/dsv/10215637

I have cropped and resized the images to 224 x 224 (with padding), making it very easy to just plug this dataset in and get started fine-tuning with pre-trained models. Roboflow is used to augment data, the specifics are provided in a txt file in both the test and train directories. Please read the provenance section for detailed information. There is still a class imbalance, which will probably require weighted sampling or weighted loss during the training process.

Mean and Standard Deviation values for Normalization (Obtained using training set only):

Mean: [0.2682, 0.2682, 0.2682]

Std: [0.3008, 0.3008, 0.3008]

Acknowledgments: “Data were provided 1-12 by OASIS-1: Cross-Sectional: Principal Investigators: D. Marcus, R, Buckner, J, Csernansky J. Morris; P50 AG05681, P01 AG03991, P01 AG026276, R01 AG021910, P20 MH071616, U24 RR021382”

Citation: OASIS-1: Cross-Sectional: https://doi.org/10.1162/jocn.2007.19.9.1498

Introduction: It has been shown that Alzheimer’s disease (AD) is accompanied by marked structural brain changes that can be detected several years before clinical diagnosis via structural magnetic resonance (MR) imaging. In this study, we developed a structural MR-based biomarker for in vivo detection of AD using a supervised machine learning approach. Based on an individual’s pattern of brain atrophy a continuous AD score is assigned which measures the similarity with brain atrophy patterns seen in clinical cases of AD.Methods: The underlying statistical model was trained with MR scans of patients and healthy controls from the Alzheimer’s Disease Neuroimaging Initiative (ADNI-1 screening). Validation was performed within ADNI-1 and in an independent patient sample from the Open Access Series of Imaging Studies (OASIS-1). In addition, our analyses included data from a large general population sample of the Study of Health in Pomerania (SHIP-Trend).Results: Based on the proposed AD score we were able to differentiate patients from healthy controls in ADNI-1 and OASIS-1 with an accuracy of 89% (AUC = 95%) and 87% (AUC = 93%), respectively. Moreover, we found the AD score to be significantly associated with cognitive functioning as assessed by the Mini-Mental State Examination in the OASIS-1 sample after correcting for diagnosis, age, sex, age·sex, and total intracranial volume (Cohen’s f2 = 0.13). Additional analyses showed that the prediction accuracy of AD status based on both the AD score and the MMSE score is significantly higher than when using just one of them. In SHIP-Trend we found the AD score to be weakly but significantly associated with a test of verbal memory consisting of an immediate and a delayed word list recall (again after correcting for age, sex, age·sex, and total intracranial volume, Cohen’s f2 = 0.009). This association was mainly driven by the immediate recall performance.Discussion: In summary, our proposed biomarker well differentiated between patients and healthy controls in an independent test sample. It was associated with measures of cognitive functioning both in a patient sample and a general population sample. Our approach might be useful for defining robust MR-based biomarkers for other neurodegenerative diseases, too.

Completed Open Access Series of Imaging Studies (OASIS-1) dataset containing the preprocessed data from FastSurfer.

This dataset contains the preprocessed 1688 different subjects from the AD, CN, and MCI groups of Alzheimer's disease, and it also contains the filtered CSV file. All of the raw data (3D MRI image) is collected from the Open Access Series of Imaging Studies (OASIS-1) dataset.

Short Introduction

This dataset contains the preprocessed data of the Open Access Series of Imaging Studies (OASIS-1) complete dataset

RAW Dataset source

Open Access Series of Imaging Studies (OASIS-1) is a large dataset that contains various types of data including MRI, and clinical data. The dataset is available at OASIS. The dataset is used for research purposes and is freely available to the public.

Preprocessing

The preprocessing of the raw data is done using the FastSurfer pipeline. All of the details of the preprocessing are available in their official documentation.

Used hardware to preprocess the dataset

Dataset Organization

Coming soon

DICOM Viewer

Freeview software is required to view the DICOM images using all the lookup tables, labels, and so on.

Project aimed at making neuroimaging data sets of brain freely available to scientific community. By compiling and freely distributing neuroimaging data sets, future discoveries in basic and clinical neuroscience are facilitated.

Link to My New Data Set: -

Links and Description for our new data: - MRI Scans - Brain Neurological Classes. For three neurological disorders, released on Kaggle. This includes 24588 Images as follows: -

1. Alzheimer’s Disease (AD). o AD Mild Demented 896 Images. o AD Moderate Demented 503 Images. o AD Very Mild Demented 2240 Images 🧠 The Alzheimer’s Disease Neuroimaging Initiative (ADNI) is a longitudinal, multi-center, observational study. The overall goal of ADNI is to validate biomarkers for Alzheimer’s disease (AD) clinical trials. https://adni.loni.usc.edu/ 🧠 OASIS-1 Cross-sectional MRI Data in Young, Middle Aged, Non-demented and Demented Older Adults This set comprises a cross-sectional collection of 416 subjects, aged 18 to 96. For each subject, 3 or 4 individual T1-weighted MRI scans obtained in single scan sessions are included. The subjects are all right-handed and include both men and women. 100 of the included subjects over the age of 60 have been clinically diagnosed with very mild to moderate Alzheimer’s disease (AD). Additionally, a reliability data set is included, containing 20 non-demented subjects who were imaged on a subsequent visit within 90 days of their initial session. https://sites.wustl.edu/oasisbrains/home/oasis-1/

2. Brain Tumor (BT). o BT Glioma 3000 Images. o BT Meningioma 3160 Images. o BT Pituitary 3490 Images. 🧠 This data is from the BraTS2020 Competition https://www.kaggle.com/datasets/awsaf49/brats2020-training-data 🧠 BRISC 2025 Annotated Dataset for Brain Tumor Image Segmentation and Classification https://www.kaggle.com/datasets/briscdataset/brisc2025?resource=download

3. Multiple Sclerosis (MS) 3195 Images. 🧠 Brain MRI Dataset of Multiple Sclerosis with Consensus Manual Lesion Segmentation and Patient Meta Information: - https://data.mendeley.com/datasets/8bctsm8jz7/1 🧠 MRIs are multi-modal images of the brain, with MS diagnosis being based mainly on : • T1-weighted • FLAIR (Fluid-Attenuated Inversion Recovery) https://shifts.grand-challenge.org/medical-dataset/

4. Normal Brain Images (Healthy) as a control image 8104 Images. 🧠 https://www.kaggle.com/datasets/masoudnickparvar/brain-tumor-mri-dataset/data

A free collection of MRI brain images for testing segmentation algorithms. It is available for download to assess the accuracy, reproducibility and sensitivity of MRI segmentation software. It includes data from infants and adults as well as patients with Alzheimer's disease.

Context:

The Open Access Series of Imaging Studies (OASIS) is a project aimed at making MRI data sets of the brain freely available to the scientific community. By compiling and freely distributing MRI data sets, we hope to facilitate future discoveries in basic and clinical neuroscience. OASIS is made available by the Washington University Alzheimer’s Disease Research Center, Dr. Randy Buckner at the Howard Hughes Medical Institute (HHMI)( at Harvard University, the Neuroinformatics Research Group (NRG) at Washington University School of Medicine, and the Biomedical Informatics Research Network (BIRN).

Content:

• Cross-sectional MRI Data in Young, Middle Aged, Nondemented and Demented Older Adults: This set consists of a cross-sectional collection of 416 subjects aged 18 to 96. For each subject, 3 or 4 individual T1-weighted MRI scans obtained in single scan sessions are included. The subjects are all right-handed and include both men and women. 100 of the included subjects over the age of 60 have been clinically diagnosed with very mild to moderate Alzheimer’s disease (AD). Additionally, a reliability data set is included containing 20 nondemented subjects imaged on a subsequent visit within 90 days of their initial session.
• Longitudinal MRI Data in Nondemented and Demented Older Adults: This set consists of a longitudinal collection of 150 subjects aged 60 to 96. Each subject was scanned on two or more visits, separated by at least one year for a total of 373 imaging sessions. For each subject, 3 or 4 individual T1-weighted MRI scans obtained in single scan sessions are included. The subjects are all right-handed and include both men and women. 72 of the subjects were characterized as nondemented throughout the study. 64 of the included subjects were characterized as demented at the time of their initial visits and remained so for subsequent scans, including 51 individuals with mild to moderate Alzheimer’s disease. Another 14 subjects were characterized as nondemented at the time of their initial visit and were subsequently characterized as demented at a later visit.

Acknowledgements:

When publishing findings that benefit from OASIS data, please include the following grant numbers in the acknowledgements section and in the associated Pubmed Central submission: P50 AG05681, P01 AG03991, R01 AG021910, P20 MH071616, U24 RR0213

Inspiration:

Can you predict dementia? Alzheimer’s?

Ratio pairwise comparisons of pipelines.

I created this dataset because I found that many Alzheimer's MRI datasets on Kaggle are highly repetitive (all based on the 6400-image version, with various augmented datasets), and they lack specific data sources. This causes issues for research and citation. This dataset is sourced from OASIS and includes MRI images (axial slices) of 416 individuals (note that there is a data imbalance issue, please perform upsampling as needed). Each image is specifically named to help you locate the corresponding OASIS research phase and individual. I first extracted MRI images from 416*4 NIfTI files (each person has four MRI scan NIfTI files) and converted them to PNG format. Then, I performed skull stripping on the converted MRIs. Finally, I manually removed images with black regions and incomplete brain displays, which took a lot of time. I hope Kagglers can use it to improve the accuracy of Alzheimer's diagnosis using various deep learning frameworks and contribute to Alzheimer's research.

You can also use the CDR scores from this dataset in association with the ADNI dataset (ADNI also provides documentation based on CDR scores) to build a large-scale dataset. Due to ADNI’s policy restrictions, we are unable to provide processed versions of ADNI images. Thank you for your understanding.

2024-12-1 There are four nii files in the ‘VeryMildDemented’ folder that I forgot to delete. However, this does not affect the images imported using tools like ImageFolder. If you batch convert the images to three channels, it may cause errors. Please search for ‘brain.nii’ and ‘mask.nii’ in the folder and delete them manually.

2026-1-14 Our corresponding paper has been published in npj Digital Medicine: Lu, Y., Yu, H., Li, T. et al. A lightweight CVTC model for accurate Alzheimer’s MRI analysis and lesion annotation. npj Digit. Med. 9, 38 (2026). https://doi.org/10.1038/s41746-025-02212-x Full open-access article: https://www.nature.com/articles/s41746-025-02212-x This dataset contains preprocessed MRI images (enhanced via MBIE, skull-stripped with LinkNet3D, etc.) from the paper, suitable for Alzheimer's disease classification and lesion annotation tasks.

Dice similarity coefficients.