Advancing Alzheimer’s Disease Detection in Clinical Settings: MRI Image Data
Description
Dataset Overview Dataset Name: Alzheimer’s Disease Detection Dataset Purpose: To facilitate the development of AI and deep learning models for detecting Alzheimer’s Disease using MRI images. Data Type: MRI brain scans in JPG format. Labels: Four classes of Alzheimer’s Disease progression: NonDemented: No signs of dementia. VeryMildDemented: Very mild cognitive impairment. MildDemented: Mild cognitive impairment. ModerateDemented: Moderate cognitive impairment. Dataset Size: Train Set: 5,121 images NonDemented: 2,560 VeryMildDemented: 1,792 MildDemented: 717 ModerateDemented: 52 Test Set: 1,279 images NonDemented: 640 VeryMildDemented: 448 MildDemented: 179 ModerateDemented: 12 License: CC BY 4.0 Folder Structure The dataset is organized into the following folder structure: Alzheimer_Dataset/ │ ├── images/ │ ├── train/ │ │ ├── NonDemented/ │ │ │ ├── image1.jpg │ │ │ ├── image2.jpg │ │ │ └── ... │ │ ├── VeryMildDemented/ │ │ │ ├── image1.jpg │ │ │ ├── image2.jpg │ │ │ └── ... │ │ ├── MildDemented/ │ │ │ ├── image1.jpg │ │ │ ├── image2.jpg │ │ │ └── ... │ │ └── ModerateDemented/ │ │ ├── image1.jpg │ │ ├── image2.jpg │ │ └── ... │ └── test/ │ ├── NonDemented/ │ │ ├── image1.jpg │ │ ├── image2.jpg │ │ └── ... │ ├── VeryMildDemented/ │ │ ├── image1.jpg │ │ ├── image2.jpg │ │ └── ... │ ├── MildDemented/ │ │ ├── image1.jpg │ │ ├── image2.jpg │ │ └── ... │ └── ModerateDemented/ │ ├── image1.jpg │ ├── image2.jpg │ └── ... │ └── labels/ ├── train_labels.csv └── test_labels.csv Dataset Details 3.1 MRI Images Description: Brain MRI scans (e.g., T1-weighted, T2-weighted) from participants. Format: JPG Resolution: 256x256 pixels Source: ADNI dataset, and local hospital Preprocessing: Converted from DICOM to JPG. Normalized for brightness and contrast. Resized to a consistent resolution. Train Labels (train_labels.csv) The train_labels.csv file contains metadata for the training set of MRI images. Each row corresponds to an image and includes the following fields: Image_Path: The relative path to the image file (e.g., train/NonDemented/image1.jpg). Label: The classification label for the image, which can be NonDemented, VeryMildDemented, MildDemented, or ModerateDemented. Participant_ID: A unique identifier for the participant (e.g., 001). Age: The age of the participant at the time of diagnosis (e.g., 72). Gender: The gender of the participant (e.g., Male or Female). Diagnosis_Date: The date of diagnosis in YYYY-MM-DD format (e.g., 2022-01-15). Notes: Additional notes or comments (e.g., N/A if no notes are available).
Files
Steps to reproduce
Test Labels (test_labels.csv) The test_labels.csv file contains metadata for the test set of MRI images. It follows the same structure as the training labels, with each row including: Image_Path: The relative path to the image file (e.g., test/NonDemented/image1.jpg). Label: The classification label for the image, which can be NonDemented, VeryMildDemented, MildDemented, or ModerateDemented. Participant_ID: A unique identifier for the participant (e.g., 101). Age: The age of the participant at the time of diagnosis (e.g., 70). Gender: The gender of the participant (e.g., Male or Female). Diagnosis_Date: The date of diagnosis in YYYY-MM-DD format (e.g., 2022-05-15). Notes: Additional notes or comments (e.g., N/A if no notes are available). For example, the first row in the file might look like this: test/NonDemented/image1.jpg, NonDemented, 101, 70, Male, 2022-05-15, N/A. Usage Instructions Download the Dataset: Ensure the dataset is downloaded and extracted into the correct folder structure. Load Images: Use libraries like PIL (Python Imaging Library) or OpenCV to load JPG images. Load Labels: Use pandas to read the CSV label files. Preprocessing: Apply any additional preprocessing steps (e.g., normalization, augmentation) as needed. Model Training: Use the train folder for training AI models and the test folder for evaluation Example Code for Loading Dataset import os import pandas as pd from PIL import Image # Load labels train_labels = pd.read_csv('Alzheimer_Dataset/labels/train_labels.csv') # Load images train_images = [] train_classes = [] for index, row in train_labels.iterrows(): img_path = os.path.join('Alzheimer_Dataset/images', row['Image_Path']) img = Image.open(img_path) train_images.append(img) train_classes.append(row['Label']) # Display first image train_images[0].show() print("Label:", train_classes[0]) Ethical Considerations Informed Consent: Ensure all participants provided informed consent for data collection. Privacy: Anonymize all data to protect participant identities. Bias: Address potential biases in data collection (e.g., age, gender, ethnicity).