Clinically Applicable Pan-Origin Cancer Detection for Lymph Nodes via Artificial Intelligence-Based Pathology

Area of Science:

• Oncology
• Pathology
• Artificial Intelligence

Background:

• Lymph node metastasis is a critical factor influencing cancer staging, treatment, and prognosis.
• Accurate histopathological diagnosis of lymph nodes is essential for effective cancer management.
• There is an urgent need for advanced diagnostic tools, such as artificial intelligence (AI), in lymph node analysis.

Purpose of the Study:

• To develop and validate a pan-origin artificial intelligence system for detecting cancer metastasis in lymph nodes.
• To improve the accuracy and efficiency of lymph node metastasis diagnosis in histopathology.

Main Methods:

• A deep learning-based system was developed using over 700 whole-slide images (WSIs).
• The system comprises two deep learning models designed to identify lymph nodes and detect cancerous presence.
• The models were trained and validated on a large dataset of WSIs from multiple organs.

Main Results:

• The system achieved an area under the receiver operating characteristic curve (AUC) of 0.958 with 95.2% sensitivity and 72.2% specificity on 1,402 WSIs from 49 organs.
• Validation on an independent dataset of 1,051 WSIs from 52 organs from another medical center demonstrated robust performance with an AUC of 0.925.
• The system shows high efficacy in identifying lymph node metastasis across diverse origins.

Conclusions:

• The developed AI system represents a significant advancement in pan-origin lymph node metastasis detection.
• It offers accurate pathological guidance, potentially reducing missed diagnoses in clinical settings.
• This technology can enhance routine clinical practice for cancer diagnosis and patient management.