โก Quick Summary
The study introduces PRET, a groundbreaking few-shot system for pan-cancer recognition that operates without the need for extensive example training. Evaluated on 23 international benchmarks, PRET achieved over 97% area under the curve on 15 benchmarks, demonstrating its potential to revolutionize cancer diagnostics.
๐ Key Details
- ๐ Dataset: 4,484 whole-slide images across 23 international benchmarks
- ๐งฉ Features used: Pan-cancer recognition without example training
- โ๏ธ Technology: PRET (few-shot learning system)
- ๐ Performance: Over 97% AUC on 15 benchmarks, maximum improvement of 36.76%
๐ Key Takeaways
- ๐ PRET offers a flexible and scalable solution for cancer recognition.
- ๐ก Few-shot learning allows for effective diagnostics without extensive labeled data.
- ๐ฅ Clinical-grade performance achieved in lymph node metastasis detection using only eight slide examples.
- ๐ฉโโ๏ธ Outperformed 11 pathologists in diagnostic accuracy.
- ๐ Aims to improve accessibility of AI-based pathology systems, especially in underserved regions.
- ๐ Significant metrics include a maximum improvement of 36.76% over existing methods.
- ๐ฌ Evaluated on a diverse set of tasks across multiple organs and hospitals.
๐ Background
The demand for accurate cancer diagnostics is ever-increasing, yet there is a global shortage of pathologists. Traditional AI models require large amounts of labeled data for each specific cancer type, which can be impractical and costly. This study addresses these challenges by introducing a novel approach that leverages few-shot learning to enhance cancer recognition capabilities.
๐๏ธ Study
The research team evaluated PRET on a comprehensive dataset of 4,484 whole-slide images from various international benchmarks. The goal was to assess its effectiveness in recognizing different types of cancer without the need for extensive training on labeled examples, thereby streamlining the diagnostic process.
๐ Results
PRET demonstrated remarkable performance, achieving over 97% area under the curve on 15 out of 23 benchmarks. The system showed a maximum improvement of 36.76% compared to existing methods, highlighting its potential as a superior tool for cancer diagnostics. Notably, it excelled in detecting lymph node metastasis with only eight examples, outperforming a panel of 11 pathologists.
๐ Impact and Implications
The introduction of PRET could significantly transform the landscape of cancer diagnostics. By providing a cost-effective and accessible solution, it has the potential to enhance diagnostic accuracy in regions with limited resources. This innovation could lead to improved patient outcomes and greater equity in healthcare access, particularly for minority populations and underserved communities.
๐ฎ Conclusion
The development of PRET marks a significant advancement in the field of cancer recognition. Its ability to operate effectively without extensive training data opens new avenues for AI in pathology. As we move forward, further research and implementation of such technologies could lead to a more equitable healthcare system, ensuring that high-quality cancer diagnostics are available to all.
๐ฌ Your comments
What are your thoughts on the potential of few-shot learning in cancer diagnostics? We would love to hear your insights! ๐ฌ Join the conversation in the comments below or connect with us on social media:
PRET is a few-shot system for pan-cancer recognition without example training.
Abstract
Pathological examination stands as the cornerstone in cancer diagnosis, impacting millions worldwide annually. With the shortage of pathologists globally, artificial intelligence (AI) has emerged rapidly to automate the diagnostics process. However, conventional AI models require substantial labeled data for each disease, posing huge challenges in scalability and practicality. Therefore, we introduce PRET (pan-cancer recognition without examples training), a few-shot system to achieve flexible, scalable, and effective cancer recognition across diverse organs, hospitals and tasks without training. Evaluated on 23 international benchmarks comprising 4,484 whole-slide images, our method outperforms existing approaches across 20 tasks, achieving over 97% area under the curve on 15 benchmarks with a maximum improvement of 36.76%. Notably, PRET delivers clinical-grade diagnostic performance in lymph node metastasis detection using only eight slide examples, outperforming 11 pathologists. By offering a flexible and cost-effective solution for pan-cancer recognition, PRET paves the way for accessible and equitable AI-based pathology systems, particularly benefiting minority populations and underserved regions.
Author: [‘Li Y’, ‘Ning Z’, ‘Xiang T’, ‘Zhang Q’, ‘Lin Z’, ‘Yi M’, ‘Feng F’, ‘Zeng B’, ‘Qian X’, ‘Sun L’, ‘Qin J’, ‘Xiang L’, ‘Fan C’, ‘Qin T’, ‘Wang Q’, ‘Bian XW’, ‘Yu KH’, ‘Zhang K’, ‘Zhang Q’, ‘Li X’]
Journal: Nat Cancer
Citation: Li Y, et al. PRET is a few-shot system for pan-cancer recognition without example training. PRET is a few-shot system for pan-cancer recognition without example training. 2026; (unknown volume):(unknown pages). doi: 10.1038/s43018-026-01141-2