Scholarlink

Background: <p>Low-dose computed tomography (LDCT) employed in lung cancer screening (LCS) programmes is increasing in uptake worldwide. LCS programmes herald a generational opportunity to simultaneously detect cancer and non-cancer-related early-stage lung disease, yet these efforts are hampered by a shortage of radiologists to interpret scans at scale. Here, we present TANGERINE, a computationally frugal, open-source vision foundation model for volumetric LDCT analysis.</p> Methods: <p>Designed for broad accessibility and rapid adaptation, TANGERINE can be fine-tuned off the shelf for a wide range of disease-specific tasks with limited computational resources and training data. The model is pretrained using self-supervised learning on more than 98,000 thoracic LDCT scans, including the United Kingdom’s largest LCS initiative to date and 27 public datasets. By extending a masked autoencoder framework to three-dimensional imaging, TANGERINE provides a scalable solution for LDCT analysis, combining architectural simplicity, public availability, and modest computational requirements.</p> Results: <p>TANGERINE demonstrates superior computational and data efficiency in a retrospective multi-dataset analysis: it converges rapidly during fine-tuning, requiring significantly fewer graphics processing unit hours than models trained from scratch, and achieves comparable or superior performance using only a fraction of the fine-tuning data. The model achieves strong performance across 14 disease classification tasks, including lung cancer and multiple respiratory diseases, and generalises robustly across diverse clinical centres.</p> Conclusions: <p>TANGERINE’s accessible, open-source, lightweight design lays the foundation for rapid integration into next-generation medical imaging tools, enabling lung cancer screening programmes to pivot from a singular focus on lung cancer detection toward comprehensive respiratory disease management in high-risk populations.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A computationally frugal, open-source chest CT foundation model for thoracic disease detection in lung cancer screening programmes

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年2月4日

查看全文
  • Niccolò McConnell,
  • Pardeep Vasudev,
  • Daisuke Yamada,
  • Daryl Cheng,
  • Mehran Azimbagirad,
  • John McCabe,
  • Shahab Aslani,
  • Ahmed H. Shahin,
  • Yukun Zhou,
  • Sam M. Janes,
  • Jennifer L. Dickson,
  • Carolyn Horst,
  • Sophie Tisi,
  • Helen Hall,
  • Priyam Verghese,
  • Andrew Creamer,
  • Thomas Callender,
  • Ruth Prendecki,
  • Amyn Bhamani,
  • Chuen Khaw,
  • Mamta Ruparel,
  • Monica L. Mullin,
  • Tanya Patrick,
  • Allan Hackshaw,
  • Anne-Marie Hacker,
  • Esther Arthur-Darkwa,
  • Samantha L. Quaife,
  • Arjun Nair,
  • Anand Devaraj,
  • Kylie Gyertson,
  • Vicky Bowyer,
  • Ethaar El-Emir,
  • Judy Airebamen,
  • Alice Cotton,
  • Kaylene Phua,
  • Elodie Murali,
  • Simranjit Mehta,
  • Janine Zylstra,
  • Karen Parry-Billings,
  • Columbus Ife,
  • April Neville,
  • Paul Robinson,
  • Laura Green,
  • Zahra Hanif,
  • Helen Kiconco,
  • Ricardo McEwen,
  • Dominique Arancon,
  • Nicholas Beech,
  • Derya Ovayolu,
  • Christine Hosein,
  • Sylvia Patricia Enes,
  • Jane Rowlands,
  • Sheetal Karavadra,
  • Aashna Samson,
  • Urja Patel,
  • Fahmida Hoque,
  • Hina Pervez,
  • Sofia Nnorom,
  • Moksud Miah,
  • Julian McKee,
  • Mark Clark,
  • Jeannie Eng,
  • Fanta Bojang,
  • Claire Levermore,
  • Anant Patel,
  • Sara Lock,
  • Alan Shaw,
  • Rajesh Banka,
  • Angshu Bhowmik,
  • Ugo Ekeowa,
  • Chris Valerio,
  • William M. Ricketts,
  • Neal Navani,
  • Ali Mohammed,
  • Terry O’Shaughnessy,
  • Charlotte Cash,
  • Magali Taylor,
  • Samanjit Hare,
  • Tunku Aziz,
  • Stephen Ellis,
  • Anthony Edey,
  • Graham Robinson,
  • Alberto Villanueva,
  • Hasti Robbie,
  • Elena Stefan,
  • Charlie Sayer,
  • Nick Screaton,
  • Navinah Nundlall,
  • Lynsey Gallagher,
  • Andrew Crossingham,
  • Thea Buchan,
  • Tanita Limani,
  • Kate Gowers,
  • Kate Davies,
  • John McCabe,
  • Joseph Jacob,
  • Mehran Azimbagirad,
  • Burcu Ozaltin,
  • Tania Anastasiadis,
  • Andrew Perugia,
  • James Rusius,
  • Geoff Bellingan,
  • Maureen Browne,
  • Eleanor Hellier,
  • Catherine Nestor,
  • Andre Altmann,
  • Yipeng Hu,
  • Paul Taylor,
  • Sam M. Janes,
  • Daniel C. Alexander,
  • Joseph Jacob

摘要

Background:

Low-dose computed tomography (LDCT) employed in lung cancer screening (LCS) programmes is increasing in uptake worldwide. LCS programmes herald a generational opportunity to simultaneously detect cancer and non-cancer-related early-stage lung disease, yet these efforts are hampered by a shortage of radiologists to interpret scans at scale. Here, we present TANGERINE, a computationally frugal, open-source vision foundation model for volumetric LDCT analysis.

Methods:

Designed for broad accessibility and rapid adaptation, TANGERINE can be fine-tuned off the shelf for a wide range of disease-specific tasks with limited computational resources and training data. The model is pretrained using self-supervised learning on more than 98,000 thoracic LDCT scans, including the United Kingdom’s largest LCS initiative to date and 27 public datasets. By extending a masked autoencoder framework to three-dimensional imaging, TANGERINE provides a scalable solution for LDCT analysis, combining architectural simplicity, public availability, and modest computational requirements.

Results:

TANGERINE demonstrates superior computational and data efficiency in a retrospective multi-dataset analysis: it converges rapidly during fine-tuning, requiring significantly fewer graphics processing unit hours than models trained from scratch, and achieves comparable or superior performance using only a fraction of the fine-tuning data. The model achieves strong performance across 14 disease classification tasks, including lung cancer and multiple respiratory diseases, and generalises robustly across diverse clinical centres.

Conclusions:

TANGERINE’s accessible, open-source, lightweight design lays the foundation for rapid integration into next-generation medical imaging tools, enabling lung cancer screening programmes to pivot from a singular focus on lung cancer detection toward comprehensive respiratory disease management in high-risk populations.