Using Machine Learning to Predict Response to Image-guided Therapies for Hepatocellular Carcinoma
Celina Hsieh1, Amanda Laguna1, Ian Ikeda1
1From the Department of Diagnostic Imaging (C.H., A.W.P.M., Z.J.) and Warren Alpert Medical School (A.L.), Brown University, Providence, RI; Department of Radiology and Biomedical Imaging, Yale University School of Medicine, New Haven, Conn (I.I., J.C.); Department of Radiology, Hospital of the University of Pennsylvania, Philadelphia, Pa (G.N.); and Department of Radiology and Radiological Sciences, Johns Hopkins University School of Medicine, 601 N Caroline St, Baltimore, MD 21205 (H.X.B.).
Machine learning (ML) can improve hepatocellular carcinoma (HCC) treatment selection. By analyzing quantitative imaging and biomarkers, ML models predict treatment response for personalized minimally invasive therapies.
Area of Science:
- Interventional oncology
- Medical imaging analysis
- Machine learning applications
Background:
- Hepatocellular carcinoma (HCC) treatments are advancing, but current staging systems struggle to optimize patient selection.
- Existing staging relies on serum markers and basic imaging, leading to variable treatment responses.
- A multidimensional approach integrating quantitative imaging, serum markers, and functional biomarkers is needed for better patient triage.
Purpose of the Study:
- To review the basics of machine learning (ML).
- To provide a comprehensive overview of ML's potential in predicting treatment response for HCC patients undergoing minimally invasive image-guided therapy.
- To highlight the need for advanced methods beyond traditional staging systems.
Main Methods:
- Discusses radiomics and artificial intelligence (AI) for extracting multidimensional data from medical images.
- Explains machine learning (ML) as a subset of AI where models learn from data to make predictions.
- Focuses on the application of these image-based methods in interventional oncology for HCC.
Main Results:
- ML models can automatically extract and process vast amounts of data from medical records and images.
- The synthesis of multidimensional data can yield clinically relevant insights for personalized treatment.
- ML shows potential in predicting treatment response, guiding personalized therapy and optimizing resource use.
Conclusions:
- Machine learning offers a powerful tool to overcome limitations of current HCC staging systems.
- Integrating quantitative imaging features and biomarkers with ML can enhance prediction of treatment response.
- This approach promises to optimize patient selection and personalize minimally invasive treatments for HCC.
