Image Dataset on Eye Diseases Classification (Uveitis, Conjunctivitis, Cataract, Eyelid) with Symptoms and SMOTE Validation

Dataset Description:
This dataset contains images and corresponding symptom descriptions for five types of eye diseases: Uveitis, Conjunctivitis, Cataract, Eyelid Drooping, and Normal. The dataset is intended for use in medical image analysis and machine learning model development. It includes image data (.jpg format) along with detailed descriptions of the diseases and their symptoms.

Diseases Included:
Normal: No abnormalities, clear vision, no redness or swelling.
Uveitis: Eye redness, pain, blurred vision, sensitivity to light, and floating spots.
Conjunctivitis: Redness, itching, tearing, discharge, and crusting of the eyelids.
Cataract: Cloudy or blurred vision, difficulty seeing at night, sensitivity to glare.
Eyelid Drooping: Drooping eyelids, swelling, irritation, and lumps on more than one eyelid.
Symptoms:
Each disease is associated with its symptoms as listed above. These symptoms are designed to help researchers and practitioners in the classification and diagnosis of these diseases based on visual and textual information. All the symptoms and dataset has been checked and corrected from a Professor from Bangladesh Eye Hospital.

Data Collection:
Images: The images were collected from online sources (via Google search) using disease-specific keywords.
Data Quality Control: Duplicate images were removed, and the dataset was carefully reviewed for accuracy.
Balanced Dataset: To ensure a balanced distribution of images across all diseases, SMOTE (Synthetic Minority Over-sampling Technique) was applied. The final dataset includes an equal number of images for each disease category (649 images per disease).

Dataset Validation:
Before SMOTE:
Cataract: 544 images
Conjunctivitis: 357 images
Eyelid Drooping: 525 images
Normal: 649 images
Uveitis: 223 images

After SMOTE:
All diseases now have 649 images, resulting in a balanced dataset.

The use of SMOTE ensures that all disease categories are equally represented, mitigating the risk of model bias due to class imbalance.

File Formats:
Images: JPEG format (.jpg)

Dataset Usage:
This dataset is intended for research purposes, specifically in medical image classification tasks. It can be used to train, test, and validate machine learning models that aim to identify and diagnose eye diseases based on images and symptoms.

Ethics & Consent:
The images used in this dataset have been gathered from publicly available online sources, and appropriate ethical considerations were taken into account. The dataset has been anonymized, and no personally identifiable information is included.

数据集描述：本数据集汇聚了五种眼科疾病（葡萄膜炎、结膜炎、白内障、眼睑下垂及正常情况）的图像及其相应的症状描述。该数据集旨在应用于医学图像分析和机器学习模型的开发。数据集包含图像数据（JPEG格式）以及疾病及其症状的详细描述。

包含的疾病：
正常：无异常，视力清晰，无红肿。
葡萄膜炎：眼部发红、疼痛、视力模糊、对光敏感及眼前漂浮物。
结膜炎：眼红、瘙痒、流泪、分泌物及眼睑结痂。
白内障：视物模糊或视力模糊、夜间视物困难、对强光敏感。
眼睑下垂：眼睑下垂、肿胀、刺激及多眼睑上的肿块。
症状：每种疾病均与上述症状相关联。这些症状旨在协助研究人员和从业者基于视觉和文本信息对这些疾病进行分类和诊断。所有症状及数据集均经孟加拉国眼科医院教授审核并更正。

数据收集：
图像：通过使用特定疾病的关键词，从网络资源（通过谷歌搜索）收集图像。
数据质量控制：移除重复图像，并仔细审查数据集以确保准确性。
平衡数据集：为确保图像在所有疾病类别中的平衡分布，应用了SMOTE（合成少数类过采样技术）。最终数据集包含每个疾病类别相同数量的图像（每疾病类别649幅图像）。

数据集验证：
在SMOTE之前：
白内障：544幅图像
结膜炎：357幅图像
眼睑下垂：525幅图像
正常：649幅图像
葡萄膜炎：223幅图像

在SMOTE之后：
所有疾病类别现在均包含649幅图像，从而实现了数据集的平衡。
SMOTE的使用确保了所有疾病类别均得到同等程度的代表，从而减轻了因类别不平衡导致的模型偏差风险。

文件格式：
图像：JPEG格式（.jpg）

数据集用途：本数据集旨在用于研究目的，特别是医学图像分类任务。它可以用于训练、测试和验证旨在基于图像和症状识别和诊断眼科疾病的机器学习模型。

伦理与同意：本数据集中使用的图像已从公开可用的网络资源中收集，并已考虑适当的伦理考量。数据集已匿名化，不包含任何个人可识别信息。