Radiomics in clinical practice and research
As Quibim, we are flattered to take part of the Italian Medical Physics Association (AIFM) webinar event that took place on September 24th 2021. The webinar was mainly about: “Radiomics for Medical Physicist Specialists: instruction for use”. Focusing on radiomics integration in divided routine and its main implications, it also welcomed medical physicists who are keen to learn more about different innovations in this field.
Quibim’s R&D engineer, Fuensanta Bellvís Bataller, took the attendees through a session on: Radiomics in Clinical Practice and Research. Where she introduced an industry trending topic which is how can healthcare providers manage an increasing image analysis demand while facing a decrease in capacity and tight recourses. This is a global problem that puts physicians in a situation of needing a solution that can help them work more efficiently, accurately, and quickly. In this context, radiologists are specifically the most affected. As they face increasing workloads, with massive volumes of images to be read every day and challenging staff shortages, this creates a burden of burnout and finally impacts patient care. Optimizing the radiology workflow has become mandatory and a challenge to medical physicists. They need precise tools that reduce reading time and improve clinical accuracy. Freeing up radiologists of time-consuming tasks means the entire organization will also work more efficiently and will expedite the decision process that arises in the patient journey. At the same time, medical physicist specialists position themselves as the ‘owners’ of how new radiomics AI development should be used.
In addition, Clinical trials rely more and more on medical images (and are increasingly dictated and regulated by evidence-based guidelines). Because trials cost time and money, researchers need also strong tools to expedite pipelines and a proactive medical imaging technology partner for fast and cost-efficient results.
But what is Radiomics?
When there’s a need, there’s a cause. This is where the role of Quibim comes to the surface, in Quibim we have developed an ecosystem – Quibim Precision® – a solution specialized to turn the visualization of the medical images into quantitative imaging biomarkers.
With our whole-body imaging ecosystem that combines the best of AI and advanced analysis techniques, we help clinicians detect the tiniest changes induced by diseases and drugs in the timeliest fashion. We mine and automatically process medical images to extract quantitative imaging biomarkers to improve both clinical routine and research. In this context, our radiomics solution analyzes and quantifies the signal intensities and pixel interrelationships, to extract textural features related to the heterogeneity of the lesion, for a wide variety of diseases (solid tumors, prostate cancer, liver cancer, osteopenia, among others). These feature extraction processes can be addressed in the following categories:
- First-order: Distribution of individual voxel values.
- Second-order: Statistical inter-relationships between neighboring voxels.
- Fractal dimension: Lesion irregularity.
- Shape: Geometric properties of the delineated ROI.
- Higher-order statistics such as the Wavelet and Laplacian transforms or the Gaussian-filtered images.
Quibim Precision® ecosystem offers tools for mining all automatically extracted imaging biomarkers to perform statistical analysis on clinical problems and scientific research, through a DataMiner environmental. These tools allow the user to study trends in the data and perform a preliminary descriptive analysis prior to a complete statistical study.
As a real-life example, Fuen presented a European research project in which Quibim leads the technical management, the PRIMAGE project, where the relevant role of radiomics for the study of pediatric cancer is clearly investigated. This H2020 project is backed by a cloud-based platform to support decision making in clinical management of pediatric cancer (Neuroblastoma and Diffuse intrinsic pontine glioma – DIPG), offering predictive tools to assist diagnosis, prognosis, therapies choice, and treatment follow-up, based on the use of novel imaging biomarkers, in-silico tumor growth simulation, advanced visualization tools, and AI-based algorithms to predict disease-specific Clinical Endpoints.