REport Generation in pathology using Pan-Asia Giga-pixel WSIs Banner
Join

REport Generation in pathology using Pan-Asia Giga-pixel WSIs

🔔 Updates

✅ Now open for registration!

⭐ Important Dates:
  • Training data release: 13/05/2025
  • Registration deadline: 27/06/2025
  • Test Phase 1 opens & Test Phase 1 data release: 27/06/2025
  • Test Phase 1 deadline: 18/07/2025
  • Test Phase 2 opens & Test Phase 2 data release: 19/07/2025
  • Test Phase 2 deadline: 09/08/2025
  • Announcement of winners: 09/09/2025

(All times are 10:00 AM, KST)

🔍 Challenge overview

Recent advances in vision-language foundation models have opened new possibilities in medical applications, particularly in image captioning, which generates textual descriptions from images. When applied to gigapixel-scale pathology images, this task demands advanced image analysis methods like slide-level feature extraction to process and interpret vast visual data. Automated pathology report generation, despite its complexities, has gained attention for its potential to address labor shortages, improve diagnostic accuracy, and enhance patient care. However, current evaluation methods relying on traditional NLP metrics such as BLEU, METEOR, and ROUGE are inadequate for the medical domain, where clinical relevance and content accuracy are paramount.

To address these limitations, this initiative focuses on:
1) evaluating report generation models with standardized datasets encompassing diverse pathological cases.
2) comparing generated reports with expert assessments to measure clinical alignment.
3) dentifying and adopting evaluation metrics tailored to medical standards.
4) exploring the integration of generated reports into diagnostic workflows, informed by clinical feedback.

Our ultimate goal is to enhance the practicality and reliability of pathology report generation models by ensuring they produce clinically meaningful and high-quality content. Furthermore, this initiative aims to address the limitations of current AI models in reflecting racial and ethnic diversity by utilizing a broader dataset that includes both Pan-Asia and European data. The challenge dataset comprises approximately 20,500 cases collected from six medical centers across five countries—Korea, Japan, India, Turkey, and Germany—contributing to the development of multicultural and multiethnic medical AI technologies.