armin3731/phd-dataset-retina-mask

Retina Masked Dataset

Control Non-Healthy

Overview

This dataset contains Retina Multispectral images in Diseased and Control classes that also has masked different parts of Retina.

Already segmented masked are listed below:

• Vessels
• Optic Disk
• Macula
• Background

Region Specific Labels

region_specific_label.csv

Overview

For each image, a region-specific label is generated and stored in a table called region_specific_label.csv. This table contains four distinct columns, each representing a different region. The text values in this table are coded as follows:

• d indicates Diseased
• c indicates Control

TODO:

Dataset module for this dataset is going to be added:

• retina-masked-dataset.py This class will read data from dataset and you can use it in pytorch with ease.

  Metadata Included

Variance Analysis Data

data_variance_analysis.csv

Overview

This dataset is designed for variance analysis of retinal health using phasor analysis of multispectral images. It supports a research project focused on comprehensive region-specific retinal health assessment using machine learning techniques.

Feature Categories

1. Harmonic Features (h1-h4)

Captures phasor analysis (g, s) values across 4 harmonic levels under consistent conditions.

2. Normalization - Phasor Features

Provides normalized representations for baseline comparison.

3. Phasor-RGB

Spectral analysis using RGB channels (bands 2, 5, 9).

4. Phasor-MSI Features

Full multispectral phasor analysis across all 12 spectral bands.

5. Average Features (RGB & MSI)

Direct spectral intensity measurements.

Metadata Included