Related Experiment Video
Updated: Sep 20, 2025
06:57
Utilizing High Resolution Ultrasound to Monitor Tumor Onset and Growth in Genetically Engineered Pancreatic Cancer Models
Published on: April 7, 2018
11.0K
Pancreatic cancer, radiomics and artificial intelligence
Luis Marti-Bonmati1,2, Leonor Cerdá-Alberich1, Alexandre Pérez-Girbés2
1GIBI230 Research Group on Biomedical Imaging, Instituto de Investigación Sanitaria La Fe, Valencia, Spain.
The British Journal of Radiology
|June 10, 2022
Summary
Quantitative imaging and artificial intelligence show promise for predicting pancreatic ductal adenocarcinoma (PDAC) aggressiveness. However, further validation and large datasets are needed for clinical use.
Area of Science:
- Oncology
- Radiology
- Bioinformatics
Background:
- Pancreatic ductal adenocarcinoma (PDAC) staging relies on contrast-enhanced CT, but lacks detail on tumor aggressiveness and microscopic spread.
- Current radiological templates are limited for precise treatment allocation in PDAC.
- Quantitative imaging offers potential for enhanced prognostic and predictive biomarkers.
Purpose of the Study:
- To explore the potential of quantitative imaging analysis, including radiomics and deep learning, for predicting PDAC aggressiveness and patient outcomes.
- To assess the feasibility of developing integrated radiomics models for personalized PDAC management.
- To identify the requirements for clinical implementation of these advanced imaging techniques.
Main Methods:
- Utilizing quantitative imaging analysis, radiomics, and dynamic imaging features to extract data on tumor characteristics.
- Employing deep learning and convolutional neural networks to identify features associated with PDAC biology and aggressiveness.
- Developing clinical models based on radiomics signatures and imaging phenotypes.
Main Results:
- Quantitative imaging features show potential as prognostic and predictive biomarkers for PDAC.
- Radiomics and deep learning models may enable prediction of tumor phenotype, treatment response, and patient prognosis.
- Current quantitative imaging approaches for PDAC are not yet ready for clinical implementation due to limitations.
Conclusions:
- Integrated radiomics models hold promise for personalized management of advanced PDAC.
- Significant limitations, including metric instability and lack of external validation, hinder clinical adoption.
- Development of trustworthy AI solutions requires large annotated datasets, image harmonization, and independent validation.
