Application of Texture Analysis to Identify Pprognostic Biomarkers for the Optimisatin of Radiotherapy Treatments
Lead Research Organisation:
CARDIFF UNIVERSITY
Department Name: Sch of Engineering
Abstract
EPSRC Portfolio: Healthcare Technologies Theme, specifically Clinical Technologies and Medical Imaging research, both 'Maintain' research areas.
Finding the right treatment for the right patient is essential to improve on the delivery of high-quality, personalised radiation therapy. Medical imaging has a fundamental role in informing diagnosis and treatment and it is a field of rapid development with the introduction of new agents that can picture, mong others, functional micro environmental factors such as metabolic activity, tumour cell proliferation, vascular structure and oxygenation status. Imaging can provide a comprehensive, personalised, non invasive view of tumour characteristics and can be used to monitor tumour progression and response to therapy. Computed Tomography (CT) is an anatomical imaging modality which is widely used in radiotherapy. CT is regularly used to assess the tumour shape and tissue densities and has a key role in the preparation of a radiotherapy plan. Every cancer patient is likely to receive multiple CT scans before, during and after radiotherapy. However, the information provided by this large amount of imaging data, regularly collected and available for each patient, is currently not used at its full potential.
Recent research has shown that computer algorithms and advanced image processing techniques, can be used to extract texture imaging features such as tumour heterogeneity, and entropy that can be associated with morphologic tumour response and overall survival. Indeed, these new image feature extraction techniques offer the opportunity to mine the large amount of image data already collected as standard of care for cancer patients and analyse them in an entire new way, providing more data that can be used to identify prognostic biomarkers for the optimisation and personalisation of radiotherapy treatments.
The main objective of this project is to investigate the feasibility of optimising definitive radiotherapy treatments through the identification of prognostic biomarkers that can can be associated with treatment outcome. The research will develop and implement methods to extract imaging features from oesophageal, head & neck and bladder cancer patients using imaging data acquired throughout the course of treatments carried out at our Centre. The availability of repeated on-treatment imaging scans will allow us to quantify and assess how image characteristics change during radiotherapy delivery, and confirm if this information can be used for individual treatment adaptation.
The project will develop a platform for (a) the automated and reproducible extraction of image-based features from all CT scans acquired for patients undergoing radiotherapy treatment of oesophageal, head and neck cancer and bladder cancer at Velindre (b) the quantitative assessment of the extracted tumour characteristics and (c) the identification of predictive biomarkers through correlation of the image descriptors with clinical outcomes. A workflow will be developed and software will be written to extract imaging data from the radiotherapy Picture Archive and Communication System. Analysis will be carried out enhancing existing open source programs actively developed by our research team.
We expect the technology and methodology developed for this project to be applied to other imaging modalities such as Positron Emission Tomography (PET) and Magnetic Resonance Imaging (MRI) increasing its predictive power. Furthermore, the model could be extended to other tumour types such as lung and anal cancer, and to diagnostic and planning data from other radiotherapy centres in Wales and beyond. The research team has indeed the right national and international links to make this happen. There is a huge opportunity for the findings of this research to be translated into clinical practice and that a further research grant application is made to main Research Councils before the end of the project.
Finding the right treatment for the right patient is essential to improve on the delivery of high-quality, personalised radiation therapy. Medical imaging has a fundamental role in informing diagnosis and treatment and it is a field of rapid development with the introduction of new agents that can picture, mong others, functional micro environmental factors such as metabolic activity, tumour cell proliferation, vascular structure and oxygenation status. Imaging can provide a comprehensive, personalised, non invasive view of tumour characteristics and can be used to monitor tumour progression and response to therapy. Computed Tomography (CT) is an anatomical imaging modality which is widely used in radiotherapy. CT is regularly used to assess the tumour shape and tissue densities and has a key role in the preparation of a radiotherapy plan. Every cancer patient is likely to receive multiple CT scans before, during and after radiotherapy. However, the information provided by this large amount of imaging data, regularly collected and available for each patient, is currently not used at its full potential.
Recent research has shown that computer algorithms and advanced image processing techniques, can be used to extract texture imaging features such as tumour heterogeneity, and entropy that can be associated with morphologic tumour response and overall survival. Indeed, these new image feature extraction techniques offer the opportunity to mine the large amount of image data already collected as standard of care for cancer patients and analyse them in an entire new way, providing more data that can be used to identify prognostic biomarkers for the optimisation and personalisation of radiotherapy treatments.
The main objective of this project is to investigate the feasibility of optimising definitive radiotherapy treatments through the identification of prognostic biomarkers that can can be associated with treatment outcome. The research will develop and implement methods to extract imaging features from oesophageal, head & neck and bladder cancer patients using imaging data acquired throughout the course of treatments carried out at our Centre. The availability of repeated on-treatment imaging scans will allow us to quantify and assess how image characteristics change during radiotherapy delivery, and confirm if this information can be used for individual treatment adaptation.
The project will develop a platform for (a) the automated and reproducible extraction of image-based features from all CT scans acquired for patients undergoing radiotherapy treatment of oesophageal, head and neck cancer and bladder cancer at Velindre (b) the quantitative assessment of the extracted tumour characteristics and (c) the identification of predictive biomarkers through correlation of the image descriptors with clinical outcomes. A workflow will be developed and software will be written to extract imaging data from the radiotherapy Picture Archive and Communication System. Analysis will be carried out enhancing existing open source programs actively developed by our research team.
We expect the technology and methodology developed for this project to be applied to other imaging modalities such as Positron Emission Tomography (PET) and Magnetic Resonance Imaging (MRI) increasing its predictive power. Furthermore, the model could be extended to other tumour types such as lung and anal cancer, and to diagnostic and planning data from other radiotherapy centres in Wales and beyond. The research team has indeed the right national and international links to make this happen. There is a huge opportunity for the findings of this research to be translated into clinical practice and that a further research grant application is made to main Research Councils before the end of the project.
People |
ORCID iD |
Emiliano Spezi (Primary Supervisor) | |
Philip Whybra (Student) |
Publications
Foley KG
(2019)
External validation of a prognostic model incorporating quantitative PET image features in oesophageal cancer.
in Radiotherapy and oncology : journal of the European Society for Therapeutic Radiology and Oncology
Parkinson C
(2018)
PO-0931: Dependency of patient risk stratification on PET target volume definition in Oesophageal cancer
in Radiotherapy and Oncology
Parkinson C
(2018)
Evaluation of prognostic models developed using standardised image features from different PET automated segmentation methods.
in EJNMMI research
Piazzese C
(2019)
PO-0964 Stability and prognostic significance of CT radiomic features from oesophageal cancer patients
in Radiotherapy and Oncology
Piazzese C
(2018)
EP-2141: Evaluation of 2D and 3D radiomics features extracted from CT images of oesophageal cancer patients
in Radiotherapy and Oncology
Piazzese C
(2019)
EP-1926 Radiomics in rectal cancer: prognostic significance of 3D features extracted from diagnostic MRI
in Radiotherapy and Oncology
Shi Z
(2018)
PV-0318: External Validation of Radiation-Induced Dyspnea Models on Esophageal Cancer Radiotherapy Patients
in Radiotherapy and Oncology
Whybra P
(2019)
Assessing radiomic feature robustness to interpolation in 18F-FDG PET imaging.
in Scientific reports
Whybra P
(2018)
EP-2117: Effect of Interpolation on 3D Texture Analysis of PET Imaging in Oesophageal Cancer
in Radiotherapy and Oncology
Whybra P
(2019)
PO-0963 A novel normalisation technique for voxel size dependent radiomic features in oesophageal cancer
in Radiotherapy and Oncology
Studentship Projects
Project Reference | Relationship | Related To | Start | End | Student Name |
---|---|---|---|---|---|
EP/N509449/1 | 30/09/2016 | 29/09/2021 | |||
1804739 | Studentship | EP/N509449/1 | 30/09/2016 | 06/07/2020 | Philip Whybra |
Description | Cardiff university collaboration with Department of Radiation Oncology (MAASTRO Clinic) |
Organisation | MAASTRO clinic |
Country | Netherlands |
Sector | Hospitals |
PI Contribution | For this collaboration, we conducted an external validation of a prognostic model incorporating quantitative PET image features in oesophageal cancer. We aimed to externally validate a published prognostic model from our group with MAASTRO. I was involved in the data analysis of this work. |
Collaborator Contribution | The dataset required for validation was provided by MAASTRO. They also provided expertise in the modelling |
Impact | Paper published (https://doi.org/10.1016/j.radonc.2018.10.033) |
Start Year | 2017 |
Title | Spaarc Pipeline for Automated Analysis and Radiomics Computing (SPAARC). |
Description | Spaarc Pipeline for Automated Analysis and Radiomics Computing (SPAARC). SPAARC implements a number of solutions for processing imaging and data analysis. I developed the standardised radiomics feature extraction built into SPAARC. |
Type Of Technology | Software |
Year Produced | 2018 |
Impact | Software used extensively within our group. Papers that are published/currently under review have use this software for image feature extraction. There is a long term plan to offer it as a web application to others. |
Description | "Prediction and Modeling of response to Molecular and External Beam Radiotherapies" International workshop |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Postgraduate students |
Results and Impact | An international workshop for mainly young researchers (PhD Students and Post-Docs). Fully funded if accepted for presentation. |
Year(s) Of Engagement Activity | 2017 |
URL | https://www.cgo-workshop-vecto.fr |
Description | Postgraduate monthly meeting (PMM) |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | Local |
Primary Audience | Postgraduate students |
Results and Impact | Postgraduate monthly meeting where one speaker presents their research, which it is then evaluated. |
Year(s) Of Engagement Activity | 2017,2018,2019 |