Over the past 20 years, research and development in medical physics has improved the accuracy and conformity of radiotherapy tremendously. This includes the development of intensity-modulated radiotherapy (IMRT and VMAT), which allows the delivery of highly conformal dose distributions to complex shaped tumors. More recently, the development of image guided adaptive radiotherapy has provided means to correct for geometric changes and organ motion over the course of therapy. The medical physics group contributes to these technological advances of radiotherapy through both clinically applied and fundamental research projects. In collaboration with industry and radiation oncologists in our department, we work on the integration of new technologies into clinical practice.

In fundamental research we focus on two areas:

The use of radiological images to improve outcome prediction and target delineation in radiotherapy. We work on methods to automate and improve target delineation based on advanced imaging techniques such as MRI. In addition, we work on Radiomics, i.e. computational methods to derive imaging biomarkers that predict patient outcome and response to therapy.

Optimal fractionation in radiotherapy. In most cases, the total radiation dose is not delivered at once. Instead the total dose is divided into many fractions that are delivered over several days or weeks, which allows healthy tissues to recover and tolerate much higher doses. This concept is called fractionation. We work on computational methods to support fractionation decisions and optimize the delivery of radiation over time.

In applied medical physics we are working on various projects but have two main fields of activity:

In dynamic image guided radiation therapy we investigate and implement different motion mitigation techniques. To deal with different aspects of motion compensation performed with robotic couch tracking, i.e. couch control, motion toleration, dosimetric consequences we started a SNF grant project in 2014.

Automated and fast adapted treatment planning is already a key topic in radiation physics. We compared different implementations of contour adaptation and automated plan creation.




Group leader: Stephanie Tanadini-Lang 

Group members: Marta Bogowicz, Alex Vils, Xaver Würms

Cancer is a heterogeneous disease in regard to etiology, pathogenesis, therapy response and prognosis. Tumor response to therapy varies not only among patients but also within the tumor itself. For optimizing treatment strategies, identification of biomarkers may be essential. Imaging biomarkers are of special interest as they provide spatial information on tumor biology and are acquired non-invasively.

In recent years, radiomics has become increasingly important for medical image characterization, both in terms of volume segmentation and prediction of treatment response. Using our in-house developed Software we can extract about 700 radiomic features describing tumor shape, tumor intensity, tumor texture from medical images (figure below).


Based on mathematical definitions we investigate tumor morphology as well as the prominent perceptual texture characteristics such as regularity (or periodicity), directionality and complexity. Altogether, texture features provide much more information about a region of interest than the mean or maximum intensity values, generally used in clinical medicine.

These radiomic features can be used for outcome prognosis or for  correlation to the tumor biology.

Recently we have developed a prognostic model based on 3 radiomic features to assess local tumor control in squamous cell carcinoma of the head and neck treated with definitive radio-chemotherapy. We were able to separate the patients into two groups with excellent and moderate outcome.


Previous publications:

1. Tanadini-Lang S, Bogowicz M, Veit-Haibach P, et al. Exploratory Radiomics in Computed Tomography Perfusion of Prostate Cancer. Anticancer research. 2018;38(2):685-690.

2. Pavic M, Bogowicz M, Würms X, et al. Influence of inter-observer delineation variability on radiomics stability in different tumor sites. Acta Oncologica. 2018:1-5.

3. Leijenaar RT, Bogowicz M, Jochems A, et al. Development and validation of a radiomic signature to predict HPV (p16) status from standard CT imaging: a multicenter study. The British Journal of Radiology. 2018;0(0):20170498.

4. Bogowicz M, Riesterer O, Stark LS, et al. Comparison of PET and CT radiomics for prediction of local tumor control in head and neck squamous cell carcinoma. Acta Oncol. 2017;56(11):1531-1536.

5. Bogowicz M, Riesterer O, Ikenberg K, et al. Computed Tomography Radiomics Predicts HPV Status and Local Tumor Control After Definitive Radiochemotherapy in Head and Neck Squamous Cell Carcinoma. International journal of radiation oncology, biology, physics. 2017;99(4):921-928.

6. Bogowicz M, Leijenaar RTH, Tanadini-Lang S, et al. Post-radiochemotherapy PET radiomics in head and neck cancer - The influence of radiomics implementation on the reproducibility of local control tumor models. Radiother Oncol. 2017;125(3):385-391.

7. Riesterer O, Nesteruk M, Studer G, Guckenberger M, Lang S. Predictive Value of Radiomics Analysis for Local Tumor Control After Radiochemotherapy in Patients With Head and Neck cancer. International Journal of Radiation Oncology • Biology • Physics. 2016;96(2):S117.

8. Bogowicz M, Riesterer O, Bundschuh RA, et al. Stability of radiomic features in CT perfusion maps. Phys Med Biol. 2016;61(24):8736-8749.

9. Nesteruk M, Lang S, Veit-Haibach P, et al. Tumor stage, tumor site and HPV dependent correlation of perfusion CT parameters and [18F]-FDG uptake in head and neck squamous cell carcinoma. Radiotherapy and Oncology. 2015;117(1):125-131.

Automated CTV delineation

Automated CTV delineation

Group Leader: Prof. Dr. Jan Unkelbach

Group Members: Bertrand Pouymayou

Many tumors infiltrate the adjacent normal tissue beyond the mascroscopic tumor mass (GTV) that is visible on today's imaging modalities such as CT, MR, and PET. This represents a challange for defining the clinical target volume (CTV) in radiotherapy, ie the volume that contains microscopic disease and therefore is to be irradiated. While GTV delineation amounts to defining a visible tumor mass on CT, MR, und PET imaging, CTV delineation is not based on visualizing tumor cells directly. It is rather based on antatomical imaging in order to localize the anatomically defined routes of microscopic tumor progression. We work on automated methods for CTV delineation - based on imaging, image processing, and computational models of tumor progression - to consistenty account for complex patient anatomy.

CTV delineation for glioblastoma

Glioblastoma is the most common primary brain tumor. Glioblastoma are known to infiltrate the healthy appearing brain tissue far beyond the GTV that is visible on MRI.  In current clinical practice, many practitioners account for the infiltrative growth by expanding the GTV with a 1-3 centimeter margin to form the CTV which is irradiated to a homogeneous dose of 60 Gy. Target delineation can potentially be improved by accounting the anisotropic spatial growth patterns of gliomas, which are observed in histopathology and MR imaging:

  • Anatomical boundaries: The dura, including its extensions falx cerebri and tentorium cerebelli, represents a boundary for migrating tumor cells. Also, except for rare cases of CSF seeding, gliomas do not infiltrate the ventricles.
  • Tumor cells infiltrate gray matter much less than white matter.
  • Tumor cells seem to migrate primarily along white matter fiber tracts.

Accounting for these growth characteristics requires an interdisciplinary effort involving mathematical modeling techniques, image processing, and analysis of clinical data. We investigate the use of a phenomenological tumor growth model for treatment planning, which replicates these growth patterns. The model is based on the Fisher-Kolmogorov equation, a partial differential equation of reaction-diffusion type.  The model predicts the spatial distribution of tumor cells in regions of the brain that appear normal using current imaging techniques. It is personalized for a given patient using MRI data obtained routinely for glioma patients. More specifically, a segmentation of the brain into white matter, gray matter, cerebrospinal fluid, and tumor. The brain tissue segmentation allows us the solve the model equations on the patient specific geometry. The target volume for radiotherapy planning can be defined as an isoline of the simulated tumor cell density.

In preliminary studies we have identified situations in which the use of the tumor growth model for radiotherapy target definition leads to differences compared to the clinical plan that was actually delivered. This is illustrated in the figures below. Figure 1 shows the T1 post contrast image of a GBM located in the left parietal lobe, close to the falx and the corpus callosum. Figure 4 shows the FLAIR image of the same patient, revealing peritumoral edema surrounding the central tumor mass. Figure 2 shows the segmentation of the brain into white matter, gray matter and CSF as well as the segmentation of the tumor into enhancing core (blue) and peritumoral edema (red).


Figure 3 shows the simulated tumor cell density using the growth model. The model reproduces important spatial growth patterns of glioblastomas: The falx is modeled as an anatomical boundary, which prevents tumor cells from migrating into the contralateral hemisphere. At the same time, the corpus callosum, white matter fiber tracts connecting the cerebral hemispheres, is modeled as a route for contralateral spread of tumor cells. Furthermore, the glioma growth model allows us to describe reduced infiltration of gray matter surrounding major sulci, which can be seen primarily in the region of the lateral sulcus.


Figure 4 illustrates the use of the growth model for target delineation. It compares the manually delineated target (yellow) used in the clinical treatment plan to the target contour derived from the model (red). The red contour corresponds to an isoline of the tumor cell density that encloses the same total volume as the manually defined target. For this patient, the model suggests a further extension of the target into the contralateral hemisphere.

Previous publications:

  1. J. Unkelbach, B. H. Menze, E. Konukoglu, F. Dittmann, M. Le, N. Ayache, and H. Shih. Radiotherapy planning for glioblastoma based on a tumor growth model: improving target volume delineation. Phys. Med. Biol., 2014; 59(3):747-770 

International collaborators:

Helen Shih; Massachusetts General Hospital, Boston, MA

Ender Konukoglu; ETH, Zürich, Switzerland

Group of Nicholas Ayache, INRIA, Sophia Antipolis, France

Bjoern Menze, TU München, Germany

Treatment plan optimization for intensity-modulated radiotherapy (IMRT/IMPT)

Treatment plan optimization for intensity-modulated radiotherapy (IMRT/IMPT)

Group Leader: Prof. Dr. Jan Unkelbach

Treatment planning for radiotherapy is based on two main components: Dose calculation algorithms and mathematical optimization algorithms. Dose calculation algorithms use physical models to describe the interaction of radiation in tissue to calculate the distribution of absorbed radiation dose in the patient. Mathematical optimization methods are used to optimize intensities and incident directions of external radiation fields in order to irradiate the tumor while minimizing the radiation dose to surrounding normal tissues. Our group has worked on many problems related to the further development of optimization algorithms for treatment planning. This includes direct aperture optimization (DAO) [3,4], volumetric modulated arc therapy (VMAT) [2], multi-criteria optimization (MCO) [4], beam angle optimization and non-coplanar VMAT [1], and robust optimization for handling uncertainties in intensity modulated proton therapy [5].

Previous publications:

  1. D. Papp, T. Bortfeld, J. Unkelbach. A modular approach to intensity-modulated arc therapy optimization with noncoplanar trajectories. Phys. Med. Biol., 2015; 60(13):5179-5198
  2. D. Papp and J. Unkelbach. Direct leaf trajectory optimization for volumetric modulated arc therapy with sliding window delivery. Medical Physics, 2014; 41:011701
  3. Cassioli A and Unkelbach J.  Aperture shape optimization for IMRT treatment planning. Phys. Med. Biol. 2013; 58(2):301-18
  4. Salari E and Unkelbach J. A column-generation based technique for multi-criteria direct aperture optimization. Phys. Med. Biol. 2013; 58:621-39
  5. J. Unkelbach, B. Martin, M. Soukup, and T. Bortfeld. Reducing the sensitivity of IMPT treatment plans to setup errors and range uncertainties via probabilistic treatment planning. Medical Physics. 2009; 36:149-163

International collaborators:

David Craft, Thomas Bortfeld; Massachusetts General Hospital, Boston, MA

David Papp; North Carolina State University, USA

Mark Bangert; German Cancer Research Center (DKFZ), Heidelberg, Germany

Optimal fractionation and spatiotemporal fractionation schemes

Optimal fractionation and spatiotemporal fractionation schemes

Group Leader: Prof. Dr. Jan Unkelbach

In current clinical practice, most radiotherapy treatments are fractionated. This is motivated by the observation that most healthy tissues can tolerate a much higher total dose if the radiation is split into small fractions. On the other hand, fractionation typically requires that a higher total dose is delivered to the tumor in order to achieve the same level of response. Fractionation decisions therefore face the tradeoff between increasing the number of fractions to protect normal tissues and increasing the total dose to maintain the same level of tumor control.

In that regard, the ideal treatment would fractionate in normal tissues, and at the same time hypofractionate in the tumor. This appears to be impossible at first glance because the dose to normal tissues is an unavoidable consequence of delivering dose to the tumor. Generally, increasing the dose to the tumor in a given fraction will increase the dose to healthy tissues in that fraction. However, interestingly it is possible to achieve some degree of hypofractionation in parts of the tumor while exploiting the fractionation effect in normal tissues. The latter can be achieved by delivering distinct dose distributions in different fractions, a concept which is referred to a spatiotemporal fractionation.

Figure 1 illustrates the concept of spatiotemporal fractionation for treating a large cerebral arteriovenous malformation (AVM). The treatment consists of 4 fractions delivered with rotation therapy (VMAT or Tomotherapy). Each fraction delivers a high single fraction dose to a distinct part of the target volume.

At the same time, each fraction creates a similar dose bath in the surrounding normal brain and thereby exploits the fractionation effect. Hence, partial hypofractionation in the target volume is achieved with more uniform fractionation in normal tissues, which yields a net improvement of the therapeutic ratio. This demonstrates that there may be a benefit of delivering different dose distributions in different fractions, purely motivated by fractionation effects rather than geometric changes of the patient.

fig.1 phys.jpg 

Figure 1: Spatiotemporal treatment plan for a large cerebral AVM.

Previous publications:

  1. J. Unkelbach, C. Zeng, and M. Engelsman.Simultaneous optimization of dose distributions and fractionation schemes in particle radiotherapy. Med. Phys. 2013; 40(9):091702
  2. J. Unkelbach, D. Papp. The emergence of nonuniform spatiotemporal fractionation schemes within the standard BED model. Med. Phys., 2015;42:2234-2241
  3. J. Unkelbach, M. Bussière, P. Chapman, J. Loeffler, H. Shih. Spatiotemporal Fractionation Schemes for Irradiating Large Cerebral Arteriovenous Malformations. Int. J. Rad. Onc. Biol. Phys., 2016 (in press)

International clinical collaborators:

Helen Shih, Jay Loeffler, Paul Chapman, Ted Hong; Massachusetts General Hospital, Boston, MA

International mathematical collaborators:

David Papp; North Carolina State University, USA

Ehsan Salari; University of Kansas, Wichita, USA

LET in proton therapy planning

LET in proton therapy planning

Group Leader: Prof. Dr. Jan Unkelbach

In-vitro cell survival experiments suggest an increase in proton relative biological effectiveness (RBE) towards the end of range. Although the data from in-vitro experiments varies substantially, it suggests that the RBE might increase from values between 1.0 and 1.1 in the entrance region to values around 1.3 at the Bragg peak and 1.6 in the falloff region [1]. It is typically assumed that this RBE increase is explained by an increase of linear energy transfer (LET) towards the end of range. On the other hand, proton treatment planning and dose reporting has been based on physical dose and a constant RBE of 1.1.

This creates a dilemma for proton therapy planning, especially for IMPT. Underestimation of RBE may lead to underestimation of normal tissue complication probabilities. IMPT treatments with highly modulated fields may deliver highly inhomogeneous LET distributions. This may result in LET hot spots in critical structures within or near the target volume. On the other hand, large uncertainties in RBE, and the fact that dose reporting has historically been based on physical dose, discourage RBE-based IMPT planning approaches that lead to drastic changes compared to current practice.

A possible hybrid approach to address this dilemma consists in LET-guided IMPT planning as recently suggested by our group. In contrast to previous works, our method does not assume knowledge of RBE to perform biological IMPT planning. Instead, it is designed to facilitate IMPT planning in the absence of reliable normal tissue RBE values. We first determine an IMPT plan based on physical dose objectives, as is current clinical practice. In a second step, we modify the LET distribution to avoid high LET in critical structures. This is done using a prioritized optimization scheme, in which LET-based objectives are optimized while limiting the degradation of the physical dose distribution.  In that sense, IMPT treatment plans become safer, while allowing the planning process to be consistent with current dose reporting.

Figure 1a illustrates this approach for a atypical meningioma patient, in whom the target volume (red) overlaps the brainstem (green), the optic nerve, the chiasm, and the pituitary gland (orange). Traditional IMPT planning based on physical dose provides highly conformal dose distributions (1c). Figure 1e shows the spatial distribution of the product of LET and physical dose, which serves as a first order approximation of the additional biological dose that is caused by high LET. In this example, high LET is observed in critical structures in the target volume. After the LET reoptimization step, such LET hot spots in critical structures can be avoided (1f) while minimally compromising the physical dose distribution (1d).

fig2. medphys.jpg 

Figure 1: LET based IMPT reoptimization for a meningioma patient.

It is clear that, in order to modify the LET distribution in critical structures, the dose to these regions has to be delivered by different pencil beams. This is illustrated in Figure 1b, which shows the difference between the physical dose distributions (the reoptimized plan is subtracted from the reference plan). The fluence of pencil beams incident from the patient's left (right side of the image) that stop in the OARs is reduced. Instead, more dose is delivered by pencil beams incident from the patient's right (left side of the image).

Previous publications:

  1. J. Unkelbach, P. Botas, D. Giantsoudi, B. Gorissen, H. Paganetti. Reoptimization of intersity-modulated proton therapy plans based on linear energy transfer. Int. J. Rad. Onc. Biol. Phys., 2016 (in press)

International collaborators:

Group of Harald Paganetti; Massachusetts General Hospital, Boston, MA

Motion compensation through couch tracking

Motion compensation through couch tracking

Group leader: Stephanie Tanadini-Lang

Group members: Stefanie Ehrbar, Alexander Jöhl, Konstantina Karavas

Modern linear accelerators achieve sub-millimeter accuracy in dose delivery. However, not all tumors are stable during the treatment session. Tumors in the lung can move up to 16 mm, in the liver up to 34 mm and the thoracic wall moves up to 14 mm due to respiratory motion. In these cases, tumor motion management is needed to accurately irradiate the tumor while sparing healthy tissue. This becomes important for large treatment volumes, where dose to the surrounding tissue is a limiting factor, and for hypo-fractionated treatments, where high doses are applied in a small number of treatment fractions.
The three motion management techniques currently used in the clinical setting are motion-encompassing treatment, gating and tracking. In comparison to motion-encompassing treatment, gating allows treatment to a smaller volume; however, at the cost of substantially increased treatment time. The most sophisticated treatment technique appears to be tracking because it confines the high dose to the tumor (small volume) and is time efficient.
In Zurich, we have developed and evaluated a couch tracking system to counter-steer the tumor motion during radiotherapy treatments (compare figures below).


In several studies, we have evaluated the performance of this tracking system and the potential clinical benefit of couch tracking for lung, prostate and pancreatic cancer.

fig 10.png


Currently we are working on a study to assess the tolerance of volunteers to the motion of the treatment couch and the integration of different image acquisition methods into the tracking system as motion feedback systems.

Previous publications:

  1. Lang, S., Zeimetz, J., Ochsner, G., Daners, M. S., Riesterer, O., & Klöck, S. (2014). Development and evaluation of a prototype tracking system using the treatment couch. Medical physics, 41(2), 021720.
  2. Jöhl, A., Lang, S., Ehrbar, S., Guckenberger, M., Klöck, S., Meboldt, M., & Schmid Daners, M. (2016). Modeling and performance evaluation of a robotic treatment couch for tumor tracking. Biomedical Engineering/Biomedizinische Technik.
  3. Ehrbar, S., Perrin, R., Peroni, M., Bernatowicz, K., Parkel, T., Pytko, I., Klöck, S., Weber, D., Guckenberger, M., Tanadini-Lang, S., Lomax, A. (2016). Respiratory motion-management in stereotactic body radiation therapy for lung cancer–A dosimetric comparison in an anthropomorphic lung phantom (LuCa). Radiotherapy and Oncology.
  4. Ehrbar, S., Schmid, S., Jöhl, A., Klöck, S., Guckenberger, M., Riesterer, O., & Tanadini-Lang, S. (2017). Validation of Dynamic Treatment-Couch Tracking for Prostate SBRT. Medical Physics.

Collaborations: ETH Zurich, Chair of Product Dev.& Eng. Design

ELPHA: Dynamically deformable liver phantom

ELPHA: Dynamically deformable liver phantom

Group leader: Stephanie Tanadini-Lang

Group members: Stefanie Ehrbar, Alexander Jöhl

Radiotherapy of liver tumors needs to account for motion of the liver. Real-time adaptive radiotherapy is an approach under investigation. In this approach, radiotherapy devices track the liver tumors continuously during treatment and compensate for the motion. To test and improve real-time adaptive radiotherapy, phantoms which represent the liver motions are required.

In a collaboration of the University Hospital Zürich, University of Zürich (USZ) and the Swiss Federal Institute of Technology (ETH), we have developed such an elastic liver phantom (ELPHA), which is dynamically deformable. This liver phantom features realistic liver motion, a time-resolved dosimetry system, and can be imaged using computer-tomography or ultrasound.

ELPHA consists of a deformable silicone torso including a silicone liver and vasculature, which can be imaged with ultrasound and radiography. Respiration is mimicked by compression of the torso. Electromagnetic transponders inside the phantom allow to measure its displacement in real-time. The phantom has been shown to reproduce liver motion traces with submillimeter accuracy. Plastic scintillation dosimeters (PSD) are inserted to the liver for a time-resolved measurement of radiation dose applied to ELPHA. With this dosimetry system and the ability of respiratory deformation, real-time adaptive treatments can be tested, verified and improved.

A video of ELPHA can be found here:



  • ETH Zurich, Product Development Group Zurich, Department of Mechanical and Process Engineering
  • ETH Zurich, Computer-assisted Applications in Medicine, Department of Information Technology and Electrical Engineering


Wir verwenden Cookies, um unsere Website nutzerfreundlich zu gestalten, sie fortlaufend zu verbessern und die Zugriffe auf unsere Website zu analysieren. Mit der Nutzung unserer Dienste erklären Sie sich damit einverstanden. Weitere Informationen finden Sie in unserer Datenschutzerklärung.