Development and external validation of a transfer learning-based system for the pathological diagnosis of colorectal cancer: a large emulated prospective study

Liuhong Yuan; Henghua Zhou; Xiao Xiao; Xiuqin Zhang; Feier Chen; Lin Liu; Jingjia Liu; Shisan Bao; Kun Tao

doi:10.3389/fonc.2024.1365364

Development and external validation of a transfer learning-based system for the pathological diagnosis of colorectal cancer: a large emulated prospective study

Front Oncol. 2024 Apr 25:14:1365364. doi: 10.3389/fonc.2024.1365364. eCollection 2024.

Authors

Liuhong Yuan^#^{1

2}, Henghua Zhou^#², Xiao Xiao^#³, Xiuqin Zhang^{1

2}, Feier Chen^{1

2}, Lin Liu⁴, Jingjia Liu³, Shisan Bao², Kun Tao^{1

2}

Affiliations

¹ Department of Pathology, Tongji Hospital, School of Medicine, Tongji University, Shanghai, China.
² Department of Pathology, Tongren Hospital, School of Medicine Shanghai Jiaotong University, Shanghai, China.
³ Wision Ltd., Chengdu, China.
⁴ Institute of Natural Sciences, MOE-LSC, School of Mathematical Sciences, CMA-Shanghai, SJTU-Yale Joint Center for Biostatistics and Data Science, Shanghai Jiao Tong University, Shanghai, China.

^# Contributed equally.

Abstract

Background: The progress in Colorectal cancer (CRC) screening and management has resulted in an unprecedented caseload for histopathological diagnosis. While artificial intelligence (AI) presents a potential solution, the predominant emphasis on slide-level aggregation performance without thorough verification of cancer in each location, impedes both explainability and transparency. Effectively addressing these challenges is crucial to ensuring the reliability and efficacy of AI in histology applications.

Method: In this study, we created an innovative AI algorithm using transfer learning from a polyp segmentation model in endoscopy. The algorithm precisely localized CRC targets within 0.25 mm² grids from whole slide imaging (WSI). We assessed the CRC detection capabilities at this fine granularity and examined the influence of AI on the diagnostic behavior of pathologists. The evaluation utilized an extensive dataset comprising 858 consecutive patient cases with 1418 WSIs obtained from an external center.

Results: Our results underscore a notable sensitivity of 90.25% and specificity of 96.60% at the grid level, accompanied by a commendable area under the curve (AUC) of 0.962. This translates to an impressive 99.39% sensitivity at the slide level, coupled with a negative likelihood ratio of <0.01, signifying the dependability of the AI system to preclude diagnostic considerations. The positive likelihood ratio of 26.54, surpassing 10 at the grid level, underscores the imperative for meticulous scrutiny of any AI-generated highlights. Consequently, all four participating pathologists demonstrated statistically significant diagnostic improvements with AI assistance.

Conclusion: Our transfer learning approach has successfully yielded an algorithm that can be validated for CRC histological localizations in whole slide imaging. The outcome advocates for the integration of the AI system into histopathological diagnosis, serving either as a diagnostic exclusion application or a computer-aided detection (CADe) tool. This integration has the potential to alleviate the workload of pathologists and ultimately benefit patients.

Keywords: AI-assisted pathological diagnosis; CRC (colorectal cancer); artificial intelligence (AI); pathological diagnosis; transfer learning.

Grants and funding

The author(s) declare financial support was received for the research, authorship, and/or publication of this article. The current study is supported by a grant from Shanghai Hospital Development Centre, grant number SHDC22023209, China.