Medical Image Tamper Detection for Deepfake Analysis

Deepfakes: Medical Image Tamper Detection is a dataset from the UCI repository focused on identifying manipulated medical images. It is intended for developing and benchmarking computer vision models to detect deepfakes and tampering in healthcare imaging contexts. The dataset's creator and specific size are not detailed in the provided metadata.

Use Cases

• Classify images as authentic or tampered using pixel-level features and metadata.
• Train convolutional neural networks to detect specific tampering artifacts like splicing or generative inpainting in medical scans.
• Benchmark model performance against known deepfake generation techniques relevant to radiology or pathology images.
• Develop anomaly detection systems to flag potentially altered MRI or X-ray images before clinical use.

Strengths

• Dataset originates from the authoritative UCI Machine Learning Repository, ensuring a baseline of curation.
• Focuses on the critical and emerging domain of medical image security and deepfake detection.

Limitations

• The total number of image samples, rows, and specific column features are unknown, hindering reproducibility assessment.
• The absence of sample data, file formats, and license information complicates practical usability and legal compliance.
• Unknown temporal range and geographic origin of the source images may introduce unquantified bias.

Provenance

Source

UCI Machine Learning Repository