• AI
  • Molecular Imaging
  • CT
  • X-Ray
  • Ultrasound
  • MRI
  • Facility Management
  • Mammography

Deep Learning Model with DCE-MRI May Help Predict Proliferative Hepatocellular Carcinoma

News
Article

Incorporating dynamic contrast-enhanced MRI, a deep learning model demonstrated a 20 percent higher AUC in external validation testing than clinical factors alone and over a 17 percent higher AUC than radiological factors alone in predicting proliferative hepatocellular carcinoma (HCC).

An emerging deep learning (DL) model may significantly enhance the prediction of proliferative hepatocellular carcinoma (HCC) over clinical factors and radiological indicators alone, according to new research.

For the retrospective study, recently published in Academic Radiology, researchers examined the use of a DL model to predict proliferative HCC in 353 patients with HCC. The study authors noted that overall cohort was comprised of 251 patients in the training set, 62 patients for internal testing and 42 patients for external validation testing.

The DL model included 85 DL features obtained from dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI), arterial phase hyperenhancement (APHE), positive findings for hepatitis B virus (HBV) and elevated alpha-fetoprotein (AFP) levels, according to the study.

Deep Learning Model with DCE-MRI May Help Predict Proliferative Hepatocellular Carcinoma

The above imaging revealed a proliferative hepatocellular carcinoma (HCC) in a 48-year-old female patient. A deep learning model, incorporating dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI), had an 83 percent area under the curve (AUC) for predicting proliferative HCC. (Images courtesy of Academic Radiology.)

In external validation testing, the DL model had a 20 percent higher area under the curve (AUC) for predicting proliferative HCC than clinical factors alone (82.8 percent vs. 62.8 percent), and over a 17 percent higher AUC than radiological factors alone (65.2 percent). Researchers noted the DL model had 100 percent sensitivity in comparison to 66.7 percent for clinical factors and radiological factors alone. The DL model also had a greater than 10 percent higher negative predictive value (NPV) in comparison to clinical factors and radiological factors alone (100 percent vs. 88.5 percent and 87.5 percent respectively).

In an analysis of 128 patients from the overall cohort examining early HCC recurrence following hepatic resection, the study authors found that patients with proliferative HCC had a nearly 20 percent early recurrence rate at two years in comparison to those with non-proliferative HCC (56 percent vs. 36.1 percent).

“The DL-based model augmented with clinical features presented the optimum predictive performance for early recurrence, aligning closely with histological assessments,” wrote Hui Qu, M.D., who is affiliated with the College of Medicine and Biological Information Engineering at Northeastern University in Shenyang, China, and colleagues.

Three Key Takeaways

1. Enhanced predictive performance. The DL-based model significantly outperformed clinical and radiological factors alone in predicting proliferative HCC, showing a 20% higher AUC compared to clinical factors and a 17% higher AUC compared to radiological factors. Additionally, it achieved 100% sensitivity, markedly higher than the 66.7% sensitivity of clinical and radiological factors alone.

2. Early recurrence prediction: For patients undergoing hepatic resection, the DL model demonstrated strong predictive performance for early HCC recurrence, closely aligning with histological assessments. Patients with proliferative HCC had a higher early recurrence rate at two years (56%) compared to those with non-proliferative HCC (36.1%).

3. Personalized treatment and surveillance. The DL model's ability to identify patients at higher risk for HCC recurrence may guide more tailored treatment plans and surveillance protocols. This stratification can help prioritize intensified therapeutic approaches for those at greater risk, potentially improving patient outcomes by facilitating earlier interventions.

The study authors maintained that the enhanced capability of the DL model could facilitate earlier treatment interventions for patients at risk for HCC recurrence.

“ … The DL-based DCE-MRI model could guide treatment decisions by identifying patients who may benefit from an intensified therapeutic approach to reduce the risk of recurrence. Moreover, the stratification of patients based on their risk of early recurrence could lead to more personalized surveillance protocols,” emphasized Qu and colleagues.

(Editor’s note: For related content, see “Can Non-Contrast Abbreviated MRI be a Viable Alternative for HCC Surveillance?,” “Emerging Model with Key MRI Feature Improves Prediction for Advanced Recurrence of Hepatocellular Carcinoma” and “Meta-Analysis Assesses Impact of Radiomics for Predicting Hepatocellular Carcinoma Recurrence.”)

Beyond the inherent limitations of a retrospective study, the authors conceded that the cohort was drawn from an area with endemic HBV, which may limit broader extrapolation of the study findings to other patients with HCC who have different primary etiologies for the disease.

Related Videos
Where the USPSTF Breast Cancer Screening Recommendations Fall Short: An Interview with Stacy Smith-Foley, MD
A Closer Look at MRI-Guided Transurethral Ultrasound Ablation for Intermediate Risk Prostate Cancer
Improving the Quality of Breast MRI Acquisition and Processing
Can Fiber Optic RealShape (FORS) Technology Provide a Viable Alternative to X-Rays for Aortic Procedures?
Can Diffusion Microstructural Imaging Provide Insights into Long Covid Beyond Conventional MRI?
Emerging MRI and PET Research Reveals Link Between Visceral Abdominal Fat and Early Signs of Alzheimer’s Disease
Nina Kottler, MD, MS
The Executive Order on AI: Promising Development for Radiology or ‘HIPAA for AI’?
Practical Insights on CT and MRI Neuroimaging and Reporting for Stroke Patients
Related Content
© 2024 MJH Life Sciences

All rights reserved.