Medical imaging has become an essential tool for clinical diagnosis, treatment and monitoring, playing an important role in the improvement of public health. Nowadays a variety of imaging techniques, including x-ray, CT, ultrasound, magnetic resonance imaging (MRI), positron emission tomography (PET) and nuclear medicine, are routinely used in medical practices. However, despite significant advances during the last decades, researches aim to further improve these imaging techniques and to enhance the information and clinical value provided by the images.
Progress in medical imaging research is underpinned by parallel developments in a broad range of complementary disciplines. Advancing research at the intersections of these fields represents one of the most promising strategies for developing imaging technologies that will have a major clinical impact.
Our CDT training programme has three core research themes; from the acquisition, processing and integration of imaging information that can be provided for an individual patient, all the way from cell to tissue, organ and system levels and from anatomical to molecular and cellular targets. Students interested in applying to the CDT should consider their background and select the stream that is most suitable.
This theme aims to fully integrate the latest developments in AI and machine learning into medical imaging pipelines, from acquisition through reconstruction to analysis and interpretation, across the entire breadth of Clinical Imaging Challenges. Research will focus on developing novel machine learning techniques, based on the family of deep learning, such as convolutional deep neural networks, generative adversarial networks, recurrent neural networks, and ensemble solutions, to enable end-to-end optimisation for traditionally modular approaches to image acquisition, reconstruction, image processing and analysis. Designing and optimising network architectures for the clinical tasks at hand will be key to finding robust, generalisable solutions. We will extend these approaches to make use of a wide range of disparate types and sources of data, from textual information in clinical records to imaging studies and clinical examinations, mimicking the way in which trained radiologists/radiographers plan, acquire and interpret medical images. Clinical translation of the developed techniques will be maximised by employing explainable AI to gain clinical acceptance. This theme is closely linked to both the “Emerging Imaging” and “Affordable Imaging” themes, allowing e.g. synergistic, real-time end-to-end image reconstruction of multi-parametric or hybrid imaging; or real-time organ localisation in multiple probe ultrasound. It will also have relevance in the “Smart Imaging Probes” theme, e.g. by enabling AI-driven characterisation of tumour heterogeneity for treatment planning using a combination of molecular features; combined with genetic profiling, this in turn will assist more affordable imaging.
The focus of this theme is to develop and validate smart targeted imaging probes for PET, SPECT, MRI, optical and photoacoustic molecular imaging across a range of clinical imaging challenges. In particular, we will focus on emerging challenges such as disease-related redox changes (hypoxia, reactive oxygen species, nitric oxide), their molecular consequences (aldehyde-containing membrane lipid fragments, protein sulfenylation, mitochondrial function) and drug resistance in cancer. In developing smarter contrast agents, we will seek to complement and maximally exploit emerging imaging devices, including hybrid PET-MR, PET, low-dose CT, ultrasound and fluorescence imaging, directly benefiting from the advances in the “Emerging Imaging” theme. We will also focus on molecular imaging to support emerging therapies including cell-based therapy and targeted drug delivery by tracking cells and drugs and measuring their effects. We will develop contrast agents that use PET to quantify MR regional contrast agent concentration, and combine PET/MR/optical contrast media to harness their complementary benefits and enable real-time visualisation of the trajectory of individual particles to characterise dynamic flows of blood and lymph. We will gain understanding of the biophysical effects of radionuclide therapy to enhance its clinical application. To support these developments, we will develop new radiochemistry platforms with an emphasis on speed and simplicity, aiming to facilitate synthesis of very short half-life tracers and reduce dependency on costly, complex infrastructure (cyclotrons, multiple hot cells, robotic synthesisers) thus facilitating cheaper and wider availability of PET, SPECT and radionuclide therapy both in developed countries and emerging economies, also linking with the “Affordable Imaging” theme.
The primary objective of this theme is to develop innovative disruptive medical imaging technology to improve diagnosis, treatment and prognosis across the clinical imaging challenges. Technological developments will focus on cutting-edge imaging modalities such as clinical 7T MRI including multi-transmission technology, PET-MR hybrid imaging, multinuclear MRI and interventional XMR. King’s leads on the pan-London 7T clinical MR programme, which will unlock new research challenges including the development and testing of transmit and receive coil design, radiofrequency pulse design based on multi-transmit technology and advanced motion correction techniques. For example, for cardiovascular imaging, these developments can be combined to obtain 7T anatomic and metabolic images with unprecedented resolution for assessment of coronary anatomy, cardiac function, viability, fibrosis and tissue perfusion; or to enable investigation of brain development in neonatal infants, pushing spatial resolution limits and exploring novel high field contrast to visualise their small developing structures. The imaging challenges of this theme will also include the development and validation of novel MR, PET and PET-MR acquisition and reconstruction methods for simultaneous tissue characterization and anato-metabolic assessment using existing and novel PET tracers and MR contrast mechanisms, highly benefiting from the advances of the “Smart Imaging Probes” theme. Moreover, the optimal configuration of this CDT will facilitate the integration with the “AI-enabled Imaging” theme research. Finally, there is also crossover to the “Affordable Imaging” theme for developing novel active/passive ultrasound imaging techniques.
The main objective of this theme is to devise and develop pioneering low-cost accurate portable imaging technologies and wearable sensors suited to front-line carers in both the developed and developing world. For example, we will develop automated gestational age estimation with a low-cost 2D ultrasound screening system. Uncertainties regarding accurate gestational age may contribute to the difficulty in accurately assessing the role preterm birth plays in neonatal mortality in the developing world. Ultrasound may help to characterize the true magnitude of this public health concern. We will also investigate the use of scan guidance systems and automatic report generation tools both for obstetric applications and also for a wide range of other medical challenges, e.g. automatic splenic volume measure for patients suffering with sickle cell disease. These tools will embed expert knowledge in the scanning systems and will make crucial biometric evaluation available to non-expert users. Experimental programmable ultrasound equipment at King’s allows us to experiment with novel transmission and reception approaches within ultrasound imaging, linking to the “Emerging Imaging” theme. We will investigate the use of multiprobe ultrasound systems that will address two growing challenges facing diagnostic ultrasound imaging: the increase in physical size of patients, as obesity becomes more widespread; and the expense of training expert operators of ultrasound scanning equipment. We will also build on our expertise in super-resolution ultrasound imaging to develop novel diagnostic aids for detection and characterisation of lesions using ultrasound. All of the developments in this challenge are naturally linked to the “AI-enabled Imaging” theme, for improvement of biometric measurements, analysis and scan guidance systems
The EPSRC Doctoral Training Partnerships (DTP) are a subset of 3 year PhD programs available in Medical Imaging at King's College London School of Biomedical Engineering and Imaging Sciences. They do not include an MRes. Click here for DTP projects.
Applications are invited from candidates with interest in multi-disciplinary research and training in a surgical and interventional engineering related domain, and a 1st class or upper second degree in relevant engineering subjects such as mathematics, optics, computer science, artificial intelligence, physics, mechanical engineering, electronic and electrical engineering.