Processed Prostate MRI Dataset for Early Cancer Detection using Machine Learning

This dataset supports the development of an automated prostate cancer detection model using machine learning. Pre-processing steps include:

a) Resizing to 224×224 pixels

b) Grayscale conversion

c) Noise reduction using a median filter

d) Contrast stretching for enhancement

e) Segmentation using Otsu thresholding and ROI contour detection

Feature extraction was performed using a Gray Level Co-Occurrence Matrix (GLCM) in four orientations (0°, 45°, 90°, and 135°), resulting in 16 statistical texture features (contrast, correlation, energy, and homogeneity). These features were normalized using Min-Max scaling.

Dataset Composition:

1. Number of images: 961

a) 424 positive cases (Prostate cancer detected)

b) 537 negative cases (Cancer not detected)

2. Tabular feature dataset (CSV) containing:

a) 16 GLCM features

b) Class labels (0 = Negative, 1 = Positive)

File Structure:

a) /images/positive/ → MRI scans with prostate cancer

b) /images/negative/ → MRI scans without prostate cancer

c) metadata.csv → Tabular feature dataset with extracted GLCM features and labels

d) readme.txt → Detailed preprocessing and extraction steps

Applications:

a) Machine learning classification (XGBoost, Random Forest, CNN, etc.)

b) Image analysis Medical

c)  Research on early diagnosis of prostate cancer