Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
169 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
45 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

A Comprehensive Evaluation of Histopathology Foundation Models for Ovarian Cancer Subtype Classification (2405.09990v2)

Published 16 May 2024 in eess.IV, cs.AI, and cs.CV

Abstract: Large pretrained transformers are increasingly being developed as generalised foundation models which can underpin powerful task-specific artificial intelligence models. Histopathology foundation models show great promise across many tasks, but analyses have typically been limited by arbitrary hyperparameters that were not tuned to the specific task. We report the most rigorous single-task validation of histopathology foundation models to date, specifically in ovarian cancer morphological subtyping. Attention-based multiple instance learning classifiers were compared using three ImageNet-pretrained feature extractors and fourteen histopathology foundation models. The training set consisted of 1864 whole slide images from 434 ovarian carcinoma cases at Leeds Teaching Hospitals NHS Trust. Five-class classification performance was evaluated through five-fold cross-validation, and these cross-validation models were ensembled for hold-out testing and external validation on the Transcanadian Study and OCEAN Challenge datasets. The best-performing model used the H-optimus-0 foundation model, with five-class balanced accuracies of 89%, 97%, and 74% in the test sets. Normalisations and augmentations aided the performance of the ImageNet-pretrained ResNets, but these were still outperformed by 13 of the 14 foundation models. Hyperparameter tuning the downstream classifiers improved performance by a median 1.9% balanced accuracy, with many improvements being statistically significant. Histopathology foundation models offer a clear benefit to ovarian cancer subtyping, improving classification performance to a degree where clinical utility is tangible, albeit with an increased computational burden. Such models could provide a second opinion to histopathologists diagnosing challenging cases and may improve the accuracy, objectivity, and efficiency of pathological diagnoses overall.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (61)
  1. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA: A Cancer Journal for Clinicians. 2024:1-35.
  2. Ovarian carcinoma subtypes are different diseases: implications for biomarker studies. PLoS medicine. 2008;5(12):e232.
  3. Invasive epithelial ovarian cancer survival by histotype and disease stage. JNCI: Journal of the National Cancer Institute. 2019;111(1):60-8.
  4. Moch H. Female genital tumours: WHO Classification of Tumours, Volume 4. WHO Classification of Tumours. 2020;4.
  5. Vroobel K. Overview of Ovarian Tumours: Pathogenesis and General Considerations. In: Pathology of the Ovary, Fallopian Tube and Peritoneum. Springer; 2024. p. 95-113.
  6. Ovarian carcinoma histotype determination is highly reproducible, and is improved through the use of immunohistochemistry. Histopathology. 2014;64(7):1004-13.
  7. Royal College of Pathologists. Meeting pathology demand: Histopathology workforce census. RCPath; 2018. Available from: https://www.rcpath.org/static/952a934d-2ec3-48c9-a8e6e00fcdca700f/Meeting-Pathology-Demand-Histopathology-Workforce-Census-2018.pdf.
  8. Access to pathology and laboratory medicine services: a crucial gap. The Lancet. 2018;391(10133):1927-38.
  9. Mortality due to cancer treatment delay: systematic review and meta-analysis. bmj. 2020;371.
  10. Artificial Intelligence in Ovarian Digital Pathology. In: Pathology of the Ovary, Fallopian Tube and Peritoneum. Springer; 2024. p. 731-49.
  11. Artificial intelligence in ovarian cancer histopathology: a systematic review. NPJ Precision Oncology. 2023;7(1):83.
  12. Public evidence on AI products for digital pathology. medRxiv. 2024:2024-02.
  13. Efficient subtyping of ovarian cancer histopathology whole slide images using active sampling in multiple instance learning. In: Tomaszewski JE, Ward AD, editors. Medical Imaging 2023: Digital and Computational Pathology. vol. 12471. International Society for Optics and Photonics. SPIE; 2023. p. 1247110.
  14. Reducing Histopathology Slide Magnification Improves the Accuracy and Speed of Ovarian Cancer Subtyping. arXiv preprint arXiv:231113956. 2023.
  15. Automatic diagnosis of ovarian carcinomas via sparse multiresolution tissue representation. In: Medical Image Computing and Computer-Assisted Intervention–MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part I 18. Springer; 2015. p. 629-36.
  16. Clinically-inspired automatic classification of ovarian carcinoma subtypes. Journal of pathology informatics. 2016;7(1):28.
  17. A structured latent model for ovarian carcinoma subtyping from histopathology slides. Medical image analysis. 2017;39:194-205.
  18. Synthesis of diagnostic quality cancer pathology images by generative adversarial networks. The Journal of pathology. 2020;252(2):178-88.
  19. The utility of color normalization for ai-based diagnosis of hematoxylin and eosin-stained pathology images. The Journal of Pathology. 2022;256(1):15-24.
  20. Deep learning-based histotype diagnosis of ovarian carcinoma whole-slide pathology images. Modern Pathology. 2022;35(12):1983-90.
  21. GRASP: GRAph-Structured Pyramidal Whole Slide Image Representation. arXiv preprint arXiv:240203592. 2024.
  22. Gadermayr M, Tschuchnig M. Multiple instance learning for digital pathology: A review of the state-of-the-art, limitations & future potential. Computerized Medical Imaging and Graphics. 2024:102337.
  23. Deep residual learning for image recognition. In: Proceedings of the IEEE conference on computer vision and pattern recognition; 2016. p. 770-8.
  24. Data-efficient and weakly supervised computational pathology on whole-slide images. Nature biomedical engineering. 2021;5(6):555-70.
  25. Transmil: Transformer based correlated multiple instance learning for whole slide image classification. Advances in neural information processing systems. 2021;34:2136-47.
  26. Embedding space augmentation for weakly supervised learning in whole-slide images. In: 2023 IEEE 20th International Symposium on Biomedical Imaging (ISBI). IEEE; 2023. p. 1-4.
  27. Immune subtyping of melanoma whole slide images using multiple instance learning. Medical Image Analysis. 2024;93:103097.
  28. Imagenet large scale visual recognition challenge. International journal of computer vision. 2015;115:211-52.
  29. Diagnosis of ovarian carcinoma cell type is highly reproducible: a transcanadian study. The American journal of surgical pathology. 2010;34(7):984-93.
  30. Virchow: A Million-Slide Digital Pathology Foundation Model. arXiv preprint arXiv:230907778. 2023.
  31. Self supervised contrastive learning for digital histopathology. Machine Learning with Applications. 2022;7:100198.
  32. Scaling vision transformers to gigapixel images via hierarchical self-supervised learning. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition; 2022. p. 16144-55.
  33. Transformer-based unsupervised contrastive learning for histopathological image classification. Medical image analysis. 2022;81:102559.
  34. Benchmarking self-supervised learning on diverse pathology datasets. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition; 2023. p. 3344-54.
  35. Scaling self-supervised learning for histopathology with masked image modeling. medRxiv. 2023:2023-07.
  36. When an image is worth 1,024 x 1,024 words: A case study in computational pathology. arXiv preprint arXiv:231203558. 2023.
  37. Robust and data-efficient generalization of self-supervised machine learning for diagnostic imaging. Nature Biomedical Engineering. 2023;7(6):756-79.
  38. Computational Pathology at Health System Scale–Self-Supervised Foundation Models from Three Billion Images. arXiv preprint arXiv:231007033. 2023.
  39. Towards a general-purpose foundation model for computational pathology. Nature Medicine. 2024:1-13.
  40. RudolfV: A Foundation Model by Pathologists for Pathologists. arXiv preprint arXiv:240104079. 2024.
  41. Gpt-4 technical report. arXiv preprint arXiv:230308774. 2023.
  42. Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:230709288. 2023.
  43. ##\##900 Comparative evaluation of ovarian carcinoma subtyping in primary versus interval debulking surgery specimen whole slide images using artificial intelligence. International Journal of Gynecologic Cancer. 2023;33(Suppl 3):A429-30.
  44. Otsu N. A threshold selection method from gray-level histograms. Automatica. 1975;11(285-296):23-7.
  45. HistoQC: an open-source quality control tool for digital pathology slides. JCO clinical cancer informatics. 2019;3:1-7.
  46. Review of artifact detection methods for automated analysis and diagnosis in digital pathology. In: Artificial Intelligence For Disease Diagnosis And Prognosis In Smart Healthcare. CRC Press; 2023. p. 177-202.
  47. Generative Adversarial Networks for Stain Normalisation in Histopathology. In: Applications of Generative AI. Cham: Springer International Publishing; 2024. p. 227-47.
  48. Color transfer between images. IEEE Computer graphics and applications. 2001;21(5):34-41.
  49. A method for normalizing histology slides for quantitative analysis. In: 2009 IEEE international symposium on biomedical imaging: from nano to macro. IEEE; 2009. p. 1107-10.
  50. Gigapixel end-to-end training using streaming and attention. Medical Image Analysis. 2023;88:102881.
  51. Augdiff: Diffusion based feature augmentation for multiple instance learning in whole slide image. arXiv preprint arXiv:230306371. 2023.
  52. Classification of epithelial ovarian carcinoma whole-slide pathology images using deep transfer learning. arXiv preprint arXiv:200510957. 2020.
  53. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:201011929. 2020.
  54. A simple framework for contrastive learning of visual representations. In: International conference on machine learning. PMLR; 2020. p. 1597-607.
  55. Dinov2: Learning robust visual features without supervision. arXiv preprint arXiv:230407193. 2023.
  56. Attention-based deep multiple instance learning. In: International conference on machine learning. PMLR; 2018. p. 2127-36.
  57. Kingma DP, Ba J. Adam: A method for stochastic optimization. arXiv preprint arXiv:14126980. 2014.
  58. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal statistical society: series B (Methodological). 1995;57(1):289-300.
  59. TRIPOD+ AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods. bmj. 2024;385.
  60. Machine Learning-driven Histotype Diagnosis of Ovarian Carcinoma: Insights from the OCEAN AI Challenge. medRxiv. 2024:2024-04.
  61. Is attention explanation? an introduction to the debate. In: Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers); 2022. p. 3889-900.
Citations (1)

Summary

We haven't generated a summary for this paper yet.

Youtube Logo Streamline Icon: https://streamlinehq.com