Research Group Stephanie Tanadini-Lang

Z-Rad is a user-friendly radiomics extraction software written in Python that features a graphical user interface (GUI) and an application programming interface (API).

Aim (short background and aims of the project) Early personalized health initiatives and projects have uni-dimensionally focused on biological sample-based disease phenotyping. However, environmental factors are known to attenuate the association between the patients` genetic background and outcome. Consequently, the need for analysis of “downstream” phenotypes has been recognized and has become a strong field of research within personalized health. In this context, multi-modal medical imaging is an important source of phenotype information and plays a prominent role in the diagnosis, staging, and treatment response monitoring. Despite the fact that imaging is standard practice to characterize the patient’s individual disease and to estimate patient individual outcome, the incorporation of medical images into personalized health has been blocked by the current standard practice of image analysis: image analysis is mostly a manual process resulting in predominantly qualitative image characterization. To unlock the full potential of medical images for personalized health, quantitative image characterization (Radiomics) has become a highly promising field of research.

Immunotherapy has revolutionized cancer treatment and renewed efforts to find therapeutic targets that enhance anti-tumor immunity across all types of malignancies. Checkpoint inhibition (CPI), CAR-T cells, and cytokines are now combined with conventional approaches to treat cancer and stimulate anti-tumor immunity. Clinical trials often measure success through general clinical parameters, with outcomes based on radiological imaging and survival rates. However, this conventional monitoring has a downside: patient responses are often detected long after treatment initiation, causing therapy-resistant patients to experience adverse effects and delaying potentially better-tailored therapies. Predictive and monitoring biomarkers for immunotherapy are just beginning to emerge, while new therapeutic approaches still largely depend on conventional clinical monitoring.

Project Title Radiomics analysis for differentiating between oligometastatic and polymetastatic disease, regardless of the primary cancer type

Treatment with immune-checkpoint inhibitors has significantly improved outcomes in metastatic non-small cell lung cancer (NSCLC), making it a first-line treatment. However, only up to 50% of patients respond to immunotherapy, and many experience disease recurrence. As a result, new treatment strategies are being investigated. Recent randomized studies have demonstrated a survival benefit when local consolidative therapy is added to standard systemic therapy for patients with metastatic NSCLC. These studies involved the local eradication of all cancer lesions in addition to systemic therapy and were limited to patients with low metastatic tumor burden, known as oligometastatic disease.

Synthetic CT (sCT) imaging has emerged as an innovation in radiotherapy planning, offering a promising alternative to simulation CT scans. The generation of synthetic CT images can be performed using neural networks, which have shown superiority compared to previously proposed methods. The general aim is predicting the electron density information required for accurate dose calculations and patient positioning, which are traditionally based on a simulation CT scan. Our group develops neural networks that are trained on extensive datasets, learning to convert MR images into synthetic CT images. Treatments sides such as brain and pelvis have been extensively researched in the last few years and commercial solutions are available for clinical implementation. Further body sides such has the abdomen and the head-neck region still require further developments. Our group focused on the latter, developing and proposing novel algorithms for sCT generation. The promising results have shown applicability of sCT in these body sites. The accuracy and precision of synthetic CT images are critical for their implementation into clinical practice. High fidelity in sCT imaging ensures that dose calculations are as precise as those based on conventional CT scans. Normally, a dose consistency to the 2% level is required, while 1% could be achieved with high quality sCT. Therefore, the neural networks used in sCT generation must be rigorously trained and validated to meet these clinical standards. Moreover, precision in sCT imaging is vital for daily patient positioning. These synthetic images are compared to the CBCT or planar x-rays acquired while the patient is on the treatment couch. A precise alignment of the patient is crucial for targeting tumors and sparing normal tissue. The required level of patient position accuracy depends on the treatment technique and can reach the sum-millimeter level for brain stereotactic treatments. Our research groups evaluate and develop validation approaches for patient positioning based on synthetic images. An example is a novel approach based on contours to CBCT matching. In such a case, the contours delineated on the MR are matched directly to the daily CBCT image, which can then be retrospectively evaluated in terms of MR-CBCT or sCT-CBCT match quality. Finally, further advantages of radiotherapy planning based on sCT include: reduced radiation exposure for the patient, reduction of the hospital visits for the patient, simplification of the workflow for the staff and reduction of the uncertainties in the image to image matching. Our research group works on the developments necessary to bring such potential advantages from the research field into clinical practice.

Synthetic CT (sCT) imaging has emerged as an innovation in radiotherapy planning, offering a promising alternative to simulation CT scans. The generation of synthetic CT images can be performed using neural networks, which have shown superiority compared to previously proposed methods. The general aim is predicting the electron density information required for accurate dose calculations and patient positioning, which are traditionally based on a simulation CT scan. Our group develops neural networks that are trained on extensive datasets, learning to convert MR images into synthetic CT images. Treatments sides such as brain and pelvis have been extensively researched in the last few years and commercial solutions are available for clinical implementation. Further body sides such has the abdomen and the head-neck region still require further developments. Our group focused on the latter, developing and proposing novel algorithms for sCT generation. The promising results have shown applicability of sCT in these body sites. The accuracy and precision of synthetic CT images are critical for their implementation into clinical practice. High fidelity in sCT imaging ensures that dose calculations are as precise as those based on conventional CT scans. Normally, a dose consistency to the 2% level is required, while 1% could be achieved with high quality sCT. Therefore, the neural networks used in sCT generation must be rigorously trained and validated to meet these clinical standards. Moreover, precision in sCT imaging is vital for daily patient positioning. These synthetic images are compared to the CBCT or planar x-rays acquired while the patient is on the treatment couch. A precise alignment of the patient is crucial for targeting tumors and sparing normal tissue. The required level of patient position accuracy depends on the treatment technique and can reach the sum-millimeter level for brain stereotactic treatments. Our research groups evaluate and develop validation approaches for patient positioning based on synthetic images. An example is a novel approach based on contours to CBCT matching. In such a case, the contours delineated on the MR are matched directly to the daily CBCT image, which can then be retrospectively evaluated in terms of MR-CBCT or sCT-CBCT match quality. Finally, further advantages of radiotherapy planning based on sCT include: reduced radiation exposure for the patient, reduction of the hospital visits for the patient, simplification of the workflow for the staff and reduction of the uncertainties in the image to image matching. Our research group works on the developments necessary to bring such potential advantages from the research field into clinical practice.

Researchers / PIs: Klara Kefer, Stefanie Ehrbar, Riccardo Dal Bello

Funding: SASRO Research Grant 2022

Research output:

Original research article:
https://doi.org/10.1016/j.zemedi.2023.02.003

Original research article:
https://doi.org/10.1016/j.phro.2023.100464

Original research article:
https://doi.org/10.1016/j.phro.2020.09.013

Original research article:
https://doi.org/10.1016/j.ctro.2023.100624

FLASH radiotherapy with electron beams represents a potential advancement in the field of cancer treatment. This technique uses ultra-high dose rates to potentially spare normal tissue while maintaining tumor control. Our research group implemented this innovative approach by modifying a decommissioned linear accelerator, which had been in clinical use from 2009 until 2022. We demonstrated that dose rates above 1 Gy per pulse and 200 Gy per second at the isocenter can be achieved for 16 MeV and 9 MeV electron beams. We plan to employ this modified linac for a phase I feasibility and safety clinical trial. Current research work focuses on the technical preparation for the clinical trial. One of the primary goals of current research is to assess the stability and reproducibility of the beam properties on the short and long term. Achieving consistent beam characteristics is crucial for ensuring that FLASH radiotherapy can be reliably used in clinical settings. This involves detailed evaluations of the beam’s output, energy, and profiles. To address these challenges, we are developing new methods to perform quality assurance (QA) of the FLASH beam. Traditional QA techniques used in conventional radiotherapy are not always directly applicable to the ultra-high dose rates characteristic of FLASH radiotherapy. Therefore, innovative approaches are required to ensure the accurate delivery of such high doses. One example is the characterisation of prototype active detectors suitable for ultra-high dose rates and their use in in-house developed phantoms for end-to-end testing. The end-to-end testing are performed to simulate clinical conditions and verify the entire treatment process, from planning to delivery. Moreover, we investigate the use of passive detector in collaboration with the Paul Scherrer Institute (PSI). Also, beam stability and reproducibility over time are critical topics in this project. Long-term studies are being conducted to monitor the performance of the modified linac, identifying any trends or deviations that could impact its clinical use. This involves continuous data collection and analysis, allowing the maintenance of optimal performances. Within this project we modified a decommissioned linac for investigating FLASH radiotherapy with electron beams. Current work focuses on the assessment of the stability and reproducibility of the beam, aiming to use FLASH radiotherapy within a clinical trial.

The delivery of radiotherapy (RT) with ultra-high dose rates (UHDR) offers a promising opportunity to widen the therapeutic window. This is potentially achieved by an increased sparing of normal tissue while maintaining unchanged tumor control, a phenomenon known as the FLASH effect. The field of research is evolving rapidly, with the first in-human applications of electron and proton UHDR paving the way for further clinical trials focusing on patients with melanoma and bone metastasis. These clinical trials are supported by numerous positive experiments demonstrating superior normal tissue sparing in various animal models, including rodents, zebrafish, mini-pigs, and domestic animals. However, some experiments have reported negative results, particularly with rodents, zebrafish, and domestic animals. Understanding the presence or absence of the FLASH effect could provide insights into the underlying mechanisms, with oxygen content and UHDR beam properties playing crucial roles. These properties can be explored with experimental irradiation platforms that offer a wide range of beam parameters. Extending these approaches to even more diverse beam configurations is a key aim of this work. Previous in-vivo studies have relied on animal models that must adhere to strict ethical regulations, limiting the scalability and throughput of experiments. To address this, we are focusing on applying this novel irradiation technique to in-vivo experiments using animal models with looser ethical requirements, allowing for higher throughput. One such model is Drosophila melanogaster, which has been widely used in radiation research but has not been extensively studied in the context of UHDR RT and the FLASH effect. Using Drosophila melanogaster as a model organism offers several technical benefits compared to vertebrate models. These include simplicity and cost-effectiveness in laboratory maintenance, a significantly shorter life cycle, abundant externally laid embryos, and the ability to undergo various genetic modifications. However, a notable drawback is the higher radioresistance of Drosophila melanogaster compared to vertebrate animals. Reducing the lifespan of adult flies requires hundreds to thousands of Grays, while observing endpoints in larvae necessitates tens to hundreds of Grays. Despite this, Drosophila melanogaster presents a promising and novel model for investigating UHDR RT and can significantly increase research throughput. Our research group, in collaboration with the biology research group, aims to investigate the FLASH effect in Drosophila melanogaster. This includes the development of dedicated irradiation setups, their characterisation and the animals follow-up post irradiation.

Researchers / PIs: Marvin Kreuzer (UZH – Biology), Irene Vertugno (UZH – Biology), Martin Pruschy (UZH – Biology), Riccardo Dal Bello

Overview of the experimental setups developed and characterized for delivering radiation to Drosophila melanogaster. Figure from abstract: https://user-swndwmf.cld.bz/ESTRO-2024-Abstract-Book/3330/

Research output / publications:
https://user-swndwmf.cld.bz/ESTRO-2024-Abstract-Book/3330/