Related Experiment Video
Updated: Oct 27, 2025
Lung CT Segmentation to Identify Consolidations and Ground Glass Areas for Quantitative Assesment of SARS-CoV Pneumonia
Published on: December 19, 2020
Weakly unsupervised conditional generative adversarial network for image-based prognostic prediction for COVID-19
Tomoki Uemura1, Janne J Näppi2, Chinatsu Watari2
13D Imaging Research, Department of Radiology, Massachusetts General Hospital and Harvard Medical School, Boston, MA 02114, USA; Department of Mechanical and Control Engineering, Kyushu Institute of Technology, Kitakyushu 804-8550, Japan.
A new AI model, pix2surv, uses chest CT scans to predict COVID-19 progression and mortality. This weakly unsupervised method outperforms existing predictors, offering a promising tool for patient management.
Area of Science:
- Artificial Intelligence in Medical Imaging
- Radiology and Medical Diagnostics
- Computational Biology and Bioinformatics
Background:
- Coronavirus disease 2019 (COVID-19) presents diverse clinical manifestations, necessitating rapid prognostic estimation for patient management.
- Current image-based predictors for COVID-19 prognosis often rely on manual features and supervised learning, with survival analysis limited to logistic regression.
- There is a need for advanced, automated methods to accurately predict disease progression and mortality from medical imaging.
Purpose of the Study:
- To develop and evaluate a novel weakly unsupervised conditional generative adversarial network (pix2surv) for direct time-to-event estimation from chest computed tomography (CT) images.
- To assess the performance of pix2surv in predicting disease progression and mortality in COVID-19 patients.
- To compare pix2surv's prognostic capabilities against existing laboratory and image-based predictors.
Main Methods:
- Development of pix2surv, a weakly unsupervised conditional generative adversarial network.
- Training pix2surv to directly estimate time-to-event data for survival analysis using chest CT images.
- Comparative analysis of pix2surv performance against traditional laboratory tests and existing image-based prognostic tools.
Main Results:
- pix2surv demonstrated superior performance in estimating disease progression and mortality compared to current methods.
- The model accurately predicts patient outcomes directly from chest CT imaging data.
- The AI-driven approach shows significant potential for improving prognostic accuracy in COVID-19.
Conclusions:
- pix2surv is a highly promising weakly unsupervised AI model for image-based prognostic prediction in COVID-19.
- The model offers a significant advancement over existing methods for estimating survival and disease progression.
- This approach facilitates more accurate and timely patient management strategies for coronavirus disease 2019.
More Related Videos
04:48Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
04:09Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma
Published on: October 10, 2018
