Predicting NSCLC cox proportial hazard with 3D Transformer and DeepSurv

Guokai Fu; Xia Peng

doi:10.1117/12.3004101

11 October 2023 Predicting NSCLC cox proportial hazard with 3D Transformer and DeepSurv

Guokai Fu, Xia Peng

Proceedings Volume 12800, Sixth International Conference on Computer Information Science and Application Technology (CISAT 2023); 1280058 (2023) https://doi.org/10.1117/12.3004101
Event: 6th International Conference on Computer Information Science and Application Technology (CISAT 2023), 2023, Hangzhou, China

Abstract

Inspired by NLP models that process language sequences, we view three-dimensional medical images as a sequence of images and propose the 3D Transformer, which could provide clinical treatment recommendations for Non-Small Cell Lung Cancer (NSCLC) patients by predicting their Cox Proportional Hazard (CPH), We propose an adaptive QKV attention mechanism applied on Transformer. Exclusive experiments show that the adaptive QKV attention mechanism significantly improves the model's focus on important objects in the image and significantly enhances the C-index. Through experiments on CT data from 1,297 NSCLC patients, our model demonstrates a high C-index, indicating its ability to accurately predict CPH and provide recommendations for distinguishing the severity of NSCLC patients' conditions for medical professionals. Our model enables end-to-end data input and output and could be extended to other medical data processing tasks.

(2023) Published by SPIE. Downloading of the abstract is permitted for personal use only.

Citation Download Citation

Guokai Fu and Xia Peng "Predicting NSCLC cox proportial hazard with 3D Transformer and DeepSurv", Proc. SPIE 12800, Sixth International Conference on Computer Information Science and Application Technology (CISAT 2023), 1280058 (11 October 2023); https://doi.org/10.1117/12.3004101

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available