Radiomics is a complex multi-step process that can be considered as part of the more complex world of Artifical Intelligence (AI). The aim of radiomics is aiding clinical decision-making and outcome prediction for more personalized medicine.
Each step of the radiomics process brings challenges that have to be considered; for example, segmentation is challenging because of reproducibility issues. Indeed, there are debates if one should look for the ground truth that is usually considered the manual segmentation made by an expert radiologist, or for the highest reproducibility, that relies on automatic segmentation.
Quantitative features are mathematically extracted by software with different complexity levels. However, we should keep in mind that the numbers extracted and elaborated by software into complex matrices come from our DICOM images, and are therefore influenced by the acquisition parameters. Therefore, differences in numbers may not depend on differences in patho-physiology of a tissue, but only on differences in acquisitions. There are different ways to overcome these limits, and in this article we have faced some of them. Moreover, reproducibility and clinical value of radiomic features should first be tested with internal cross-validation and then validated on independent external cohorts, possibly in large prospective multicentric studies.
- Radiomics is a complex multi-step process aiding clinical decision-making and outcome prediction
- Manual, automatic, and semi-automatic segmentation is challenging because of reproducibility issues
- Quantitative features are mathematically extracted by software, with different complexity levels
- Reproducibility and clinical value of radiomic features should be firstly tested with internal cross-validation and then validated on independent external cohorts
Authors: Stefania Rizzo, Francesca Botta, Sara Raimondi, Daniela Origgi, Cristiana Fanciullo, Alessio Giuseppe Morganti and Massimo Bellomi