Team Ai
Datasetpublic

dunktra/dermacheck-temporal-pairs

DermaCheck Temporal Pairs Dataset Dataset Description Synthetic temporal image pairs for training MedGemma to detect changes in dermatoscopic images over time. Created for: MedGemma Impact Challenge 2026 - Novel Task Prize (temporal change detection) Dataset Statistics Total pairs: 900 Train: 630 pairs (70.0%) Validation: 135 pairs (15.0%) Test: 135 pairs (15.0%) Generation Methods Controlled Augmentation (~50%): Original HAM10000… See the full description on the dataset page: https://huggingface.co/datasets/dunktra/dermacheck-temporal-pairs.

sourceHugging Facecc-by-nc-sa-4.0updated 9mo agoView on Hugging Face
1likes44downloads
Dataset Card

DermaCheck Temporal Pairs Dataset

Dataset Description

Synthetic temporal image pairs for training MedGemma to detect changes in dermatoscopic images over time.

Created for: MedGemma Impact Challenge 2026 - Novel Task Prize (temporal change detection)

Dataset Statistics

  • —Total pairs: 900
  • —Train: 630 pairs (70.0%)
  • —Validation: 135 pairs (15.0%)
  • —Test: 135 pairs (15.0%)

Generation Methods

  1. 1.Controlled Augmentation (~50%): Original HAM10000 images augmented to simulate temporal evolution
  2. 2.Size increase: 10-30%
  3. 3.Border irregularity: Elastic transforms
  4. 4.Color variation: HSV adjustments
  5. 5.Based on clinical research: melanoma evolution patterns
  1. 1.Natural Pairing (~50%): Similar lesions from HAM10000 matched as temporal proxies
  2. 2.Feature-based similarity matching
  3. 3.Cosine similarity range: [0.6, 0.85]
  4. 4.Same diagnosis category (melanoma focus)
  1. 1.Unchanged Pairs (~15%): Negative examples for balanced training
  2. 2.Same image duplicated as before/after
  3. 3.Label: 0 (no change)

Dataset Structure

Each example contains:

  • —image_before: PIL Image (before timepoint)
  • —image_after: PIL Image (after timepoint)
  • —text_prompt: ABCDE-focused change detection question
  • —label: 0 (no change) or 1 (change detected)
  • —change_description: Educational explanation of changes
  • —method: Generation method (controlledaugmentation, naturalpairing, unchanged)
  • —pair_id: Unique identifier

Source Data

Based on HAM10000 dataset:

  • —Tschandl, P., Rosendahl, C. & Kittler, H. The HAM10000 dataset. Sci Data 5, 180161 (2018).
  • —License: CC BY-NC-SA 4.0 (Non-commercial use)
  • —URL: https://www.kaggle.com/datasets/kmader/skin-cancer-mnist-ham10000

Intended Use

  • —Fine-tuning MedGemma for temporal change detection
  • —Educational tool development
  • —Research purposes only (not for clinical diagnosis)

Citation

If you use this dataset, please cite both the original HAM10000 dataset and this derived work.

License

CC BY-NC-SA 4.0 (Non-commercial use, per HAM10000 license)