Related Experiment Video
Updated: Jun 3, 2025
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
6.7K
Deep learning-based lymph node metastasis status predicts prognosis from muscle-invasive bladder cancer
Qingyuan Zheng1,2, Panpan Jiao1,2, Rui Yang1,2
1Department of Urology, Renmin Hospital of Wuhan University, 99 Zhang Zhi-dong Road, Wuhan, Hubei, 430060, P.R. China.
World Journal of Urology
|January 10, 2025
Summary
A deep learning model predicts lymph node metastasis in muscle invasive bladder cancer using primary tumor images. The model’s predicted score also indicates patient survival, offering potential for personalized management.
Area of Science:
- Oncology
- Medical Imaging
- Artificial Intelligence
Background:
- Muscle invasive bladder cancer (MIBC) requires accurate staging for treatment.
- Lymph node metastasis (LNM) is a critical prognostic factor in MIBC.
- Current methods for LNM assessment can be invasive and may not capture all prognostic information.
Purpose of the Study:
- To develop a deep learning (DL) model for predicting LNM status in MIBC using primary tumor histology.
- To validate the prognostic value of the DL model-derived aiN score in MIBC patients.
Main Methods:
- Utilized a dataset of 323 MIBC patients from The Cancer Genome Atlas (TCGA) for training and internal validation.
- Extracted image features using the UNI visual encoder.
- Performed external validation on 139 MIBC patients from Renmin Hospital of Wuhan University (RHWU).
Main Results:
- The DL model achieved an AUC of 0.79 (internal) and 0.72 (external) for LNM prediction.
- The aiN score independently predicted survival in both TCGA (HR=1.608) and RHWU (HR=2.746) cohorts.
- Prognostic value of the aiN score was consistent across various patient subgroups.
Conclusions:
- DL-based analysis of H&E-stained histology can predict LNM status in MIBC.
- The aiN score derived from DL models holds significant prognostic value for MIBC patients.
- This approach may facilitate personalized management strategies for MIBC, pending further prospective validation.
