The artificial intelligence-based model ANORAK improves histopathological grading of lung adenocarcinoma
Xiaoxi Pan, Khalid AbdulJabbar, Jose Coelho‐Lima, Anca-Ioana Grapa, Hanyun Zhang, Alvin H.K. Cheung, Juvenal Baena, Takahiro Karasaki, Claire Wilson, Marco Sereno, Selvaraju Veeriah, Sarah J. Aitken, Allan Hackshaw, Andrew G. Nicholson, Mariam Jamal‐Hanjani, John Le Quesne, Sam M. Janes, Anne-Marie Hacker, Abigail Sharp, Sean Smith, Harjot Kaur Dhanda, Kitty S. Chan, Camilla Pilotti, Rachel Leslie, David Chuter, Mairead MacKenzie, Serena Chee, Aiman Alzetani, Eric Lim, Paulo De Sousa, Simon Jordan, Alexandra Rice, Hilgardt Raubenheimer, Harshil Bhayani, Lyn Ambrose, Anand Devaraj, Hema Chavan, Sofina Begum, Silviu Buderi, Daniel Kaniu, Mpho Malima, Sarah Booth, Nadia Fernandes, Pratibha Shah, Chiara Proli, Madeleine Hewish, Sarah Danson, Michael Shackcloth, Lily Robinson, Peter Russell, Kevin G. Blyth, Andrew Kidd, Alan Kirk, Mo Asif, Rocco Bilancia, Nikos Kostoulas, Mathew Thomas, Craig Dick, J.F. Lester, Amrita Bajaj, Apostolos Nakas, Azmina Sodha-Ramdeen, Mohamad Tufail, Molly Scotland, Rebecca Boyles, Sridhar Rathinam, Dean A. Fennell, Claire Wilson, Domenic Marrone, Sean Dulloo, Gurdeep Matharu, Jacqui Shaw, Joan Riley, Lindsay Primrose, Ekaterini Boleti, Heather Cheyne, Mohammed Khalil, Shirley Richardson, Tracey Cruickshank, Gillian Price, Keith M. Kerr, Sarah Benafif, Dionysis Papadatos-Pastos, James M. Wilson, Tanya Ahmad, Jack French, Kayleigh Gilbert, Babu Naidu, Akshay J. Patel, Aya Osman, Christer Lacson, Gerald Langman, Helen Shackleford, Madava Djearaman, Salma Kadiri, Gary Middleton, Angela Leek, Jack Davies Hodgkinson, Nicola Totten, Ángeles Montero
Abstract
The introduction of the International Association for the Study of Lung Cancer grading system has furthered interest in histopathological grading for risk stratification in lung adenocarcinoma. Complex morphology and high intratumoral heterogeneity present challenges to pathologists, prompting the development of artificial intelligence (AI) methods. Here we developed ANORAK (pyrAmid pooliNg crOss stReam Attention networK), encoding multiresolution inputs with an attention mechanism, to delineate growth patterns from hematoxylin and eosin-stained slides. In 1,372 lung adenocarcinomas across four independent cohorts, AI-based grading was prognostic of disease-free survival, and further assisted pathologists by consistently improving prognostication in stage I tumors. Tumors with discrepant patterns between AI and pathologists had notably higher intratumoral heterogeneity. Furthermore, ANORAK facilitates the morphological and spatial assessment of the acinar pattern, capturing acinus variations with pattern transition. Collectively, our AI method enabled the precision quantification and morphology investigation of growth patterns, reflecting intratumoral histological transitions in lung adenocarcinoma.