A simulation framework to balance radiation induced risk against image quality in radiology

Sunay Rodríguez Pérez


Bosmans Hilde, (KULeuven), hilde.bosmans@uzleuven.be

SCK•CEN Mentor

Struelens Lara
+32 14 33 28 85

SCK•CEN Co-mentor

Vanhavere Filip
+32 14 33 28 59

Expert group

RP Dosimetry and Calibration

PhD started


Short project description

Medical imaging represents the largest source of manmade radiation and is subject to an increasing public awareness. Dose reduction can be achieved by lowering the number of exposures or by lowering the exposure per image. The first part is linked to strictly medical decisions, with justification obligatory for each individual patient. The second part requires ‘optimization’. The Euratom directive 97/43 stipulates that ‘The optimization process shall include the selection of equipment, the consistent production of adequate diagnostic information or therapeutic outcome as well as the practical aspects, quality assurance including quality control and the assessment and evaluation of patient doses or administered activities, taking into account economic and social factors.’ The European medical ALARA network (EMAN) recognized a large variation in doses for nominally the same procedures and made optimization of radiation protection for healthcare workers and patients throughout Europe to its mission (www.eman-network.eu). As far as we know, none of the radiation protection guidance documents have outlined ‘optimization’ in full detail or prepared for practical implementation.


The objective of the study is to perform optimization studies by balancing radiation induced risks against clinically relevant image quality metrics. Image quality metrics with a clear link to radiological tasks will be defined. They should allow a quantitative evaluation and provide reference values. Radiation induced risks will be based upon organ doses rather than effective dose, as the latter concept is of limited use for radiological imaging that is characterized by partial radiation. The study will explore the image quality needed to perform a task with a predefined success rate and will then study which dose is minimally needed to achieve the performance. A fully detailed simulation framework will be used to study current system characteristics as well as new design parameters beyond actual systems in a so-called virtual clinical trial setting. We will focus at X-ray imaging of the chest, as this is a corner stone for routine chest X-ray, pediatric chest X-rays and dynamic, cardiac interventional work.

A simulation framework will be developed that starts from X-ray beam characteristics and specific (anthropomorphic) phantoms up to an image quality metrics for the phantom image.  Specific radiological tasks that have been studied in earlier PhD theses in SCK•CEN are large area contrast optimization in the lung of premature babies (K. Smans) and detectability of small calcifications as well as masses in breast imaging (E. Salvagnini, ongoing work). O. Dragusin has worked on antropomorphic cardiac phantoms. Present study will add more input parameters for optimization, beyond the current system designs, such as mono-energetic or dual energy radiation, new filters, new imaging detectors, new (3D) geometries, new image processing tools and an extended dose range. Antropomorphic phantoms will be completed with inserts that allow detectability studies or any other contrast related measure. Preference will be given to model observers, with some of these still in need for validation with human readings.

The following steps will be required:

  • Select relevant clinical tasks and construct a voxel model to represent a patient (with pathology)
  • Define a quantitative measure for the task (ex: a ‘Model Observer’ using a nonprewhitened filter)
  • Characterize typical imaging modalities  by their MTF, noise power, lag, spectrum, … (to be done in the University Hospital of Leuven)
  • Calculate the quantitative measure from patient images or test object acquisitions for a selected range of settings
  • Complete the SCK•CEN simulation framework of K. Smans for the selected applications.
    This includes (1) adjustments to spectrum, geometry and dose, and imaging detectors, (2) the development of anthropomorphic phantoms for different types of patients and with particular inserts for the tasks under study, and (3) accurate dosimetry modules for specific organs.
  • Simulate the image quality metrices for the available conditions as well as for other specifications and determine optimal values
  • If possible, validate the optimal settings in a (virtual) clinical study

The expected end-results are guidelines and procedures for optimisation in radiology. Such document are not available today.