Validating texture analysis of medical imaging in radiotherapy treatment

This project concerns a big data approach in medicine called radiomics, and specifically a medical image analysis method used in radiomics called texture analysis (TA). There is evidence that TA is an imaging biomarker with strong predictive power for prognosis and that it is useful for guiding optimal therapy (in cancer treatments such as radiotherapy), but there is difficulty in reproducing results between centres, hence the need for a thorough quantitative understanding of TA which this project will address by developing standardised methodology including phantoms and analytical methods. This project addresses the healthcare technologies grand challenge area of Optimising Treatment, particularly Patient Specific Predictive Modelling plus the ICT research area of Medical Imaging.
Due to the inconsistent results in literature and the many calls for standardisation there is a strong call for work in this area. With the use of phantoms and standardisation of protocols, both PET and CT texture analysis validation will be investigated. This will allow for further understanding about internal structures that instigate the heterogeneous texture of tumours. Imaging parameters can have large effects on the calculated textural features of images, by reproducing identical imaging protocols across centres the different effects of these can be reduced allowing for a more consistent texture to be obtained. By varying the imaging parameters and using a consistent imaging basis in a phantom, the effects of the imaging parameters can also be investigated.