image: Figure 2: Architectural frameworks for uncertainty quantification in hepatobiliary image processing. (A) T1ρ mapping framework with integrated uncertainty estimation. The probabilistic neural network processes T1ρ-weighted images to simultaneously generate refined T1ρ maps and uncertainty estimates. This framework achieved <3% relative mapping error while reducing scan times from 10 to 6 seconds (Huang et al., 2023). The uncertainty-weighted training enables effective ROI refinement and identification of unreliable regions. (B) UP-Net dual-module physics-driven architecture. The first module employs GAN-based artifact suppression to reduce radial streak artifacts in stack-of-radial MRI data. The second module uses a bifurcated UNet structure for parameter mapping, generating quantitative maps (fat fraction and R2* ) along with uncertainty estimates. This framework dramatically improved processing efficiency (79 ms/slice vs. 3.2 min/slice) while maintaining accuracy through uncertainty-weighted training (Shih et al., 2023). Color coding indicates different processing stages: input data (blue), intermediate processing modules (orange/purple), and output maps (green for quantitative maps, red for uncertainty estimates). Arrows show the data flow through each framework, illustrating how uncertainty quantification is integrated into the processing pipeline. Both architectures demonstrate the evolution toward real-time processing capabilities while maintaining robust uncertainty estimation for clinical applications.
Credit: Copyright: © 2025 Singh et al. This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
“These advancements, by providing more reliable and efficient diagnostic tools, may significantly impact clinical practice by addressing the ever-growing clinical demand and work pressure, while maintaining interpretability and clinical relevance.”
BUFFALO, NY – April 8, 2025 – A new editorial was published in Oncotarget, Volume 16, on April 4, 2025, titled “Deep learning-based uncertainty quantification for quality assurance in hepatobiliary imaging-based techniques.”
Dr. Yashbir Singh from Mayo Clinic and his colleagues discussed how artificial intelligence (AI) can improve liver imaging by recognizing when it might be wrong. This approach, called “uncertainty quantification,” helps clinicians better detect liver cancer and other diseases by pointing out areas in medical scans that need a second look. The authors explain how these AI tools could make imaging results more accurate and reliable, which is especially important when diagnosing serious conditions like liver tumors.
Liver and bile duct imaging is difficult because of the organ’s complex structure and differences in image quality. Even skilled radiologists can struggle to identify small or hidden tumors, especially in patients with liver damage or scarring. The editorial explains how new AI models not only read medical images but also measure their own confidence. When the AI system is unsure, it can alert clinicians to take a closer look. This extra layer of information can reduce missed diagnoses and improve early detection of liver cancer.
One of the most advanced tools described in the editorial is called AHUNet (Anisotropic Hybrid Network). This AI model works with both 2D and 3D images and can highlight which parts of a scan it is most confident about. It performed well when measuring the entire liver and showed how its confidence dropped when scanning smaller or multiple lesions. This feature helps clinicians know when more testing or review is needed.
The authors also looked at other AI models used in liver imaging. Some tools were able to analyze liver fat using ultrasound images and give clinicians both a result and a confidence score. Others improved the speed and accuracy of liver magnetic resonance imaging (MRI) scans, helping to create clear images in less time. These advancements could help hospitals work faster and provide better care.
The editorial highlights how this technology can be especially helpful in smaller clinics. If they do not have liver specialists, they could still use AI systems that flag uncertain results and send them to larger centers for review. Such an approach could improve care in rural or less-resourced areas.
“Radiology departments should develop standardized reporting templates that incorporate uncertainty metrics alongside traditional imaging findings.”
By using AI tools that know when to second-guess themselves, clinicians may soon have more reliable methods for detecting liver cancer and monitoring liver disease. The authors suggest that uncertainty-aware AI may soon become a vital part of everyday medical imaging, supporting faster and more accurate decisions in liver disease care.
Continue reading: DOI: https://doi.org/10.18632/oncotarget.28709
Correspondence to: Yashbir Singh — singh.yashbir@mayo.edu
Keywords: cancer, deep learning, uncertainty quantification, radiology, hepatobiliary imaging
Click here to sign up for free Altmetric alerts about this article.
About Oncotarget:
Oncotarget (a primarily oncology-focused, peer-reviewed, open access journal) aims to maximize research impact through insightful peer-review; eliminate borders between specialties by linking different fields of oncology, cancer research and biomedical sciences; and foster application of basic and clinical science.
Oncotarget is indexed and archived by PubMed/Medline, PubMed Central, Scopus, EMBASE, META (Chan Zuckerberg Initiative) (2018-2022), and Dimensions (Digital Science).
To learn more about Oncotarget, visit Oncotarget.com and connect with us on social media:
X
Facebook
YouTube
Instagram
LinkedIn
Pinterest
Spotify, and available wherever you listen to podcasts
Click here to subscribe to Oncotarget publication updates.
For media inquiries, please contact media@impactjournals.com.
Journal
Oncotarget
Method of Research
Commentary/editorial
Subject of Research
Not applicable
Article Title
Deep learning-based uncertainty quantification for quality assurance in hepatobiliary imaging-based techniques
Article Publication Date
4-Apr-2025
COI Statement
Authors have no conflicts of interest to declare.