UCL Early Detection Resources

UCL is a global leader in combining deep learning methodologies with dedicated state-of-the-art imaging in order to enhance the sensitivity and specificity of our novel early detection strategies.

Computational resources for early detection research

The UCL ACED Centre has dedicated GPU resources for neural network training. They are accessible through the UCL Computer Science High Performance Computing cluster.

Available GPU resources

  • 24 Nvidia Quatro RTX6000 GPUs with 24 GB GPU memory
  • 8 Nvidia Quadro RTX8000 GPUs with 48 GB GPU memory
  • 12 Nvidia A100 GPUs with 40GB GPU memory
  • 40+ Nvidia RTX A6000 (Ampere Architecture) GPUs with 48GB GPU memory (not yet available)

To use these resources, basic training in BASH and Secure Shell (SSH) are required. Users are also required to attend the cluster induction offered by the UCL CS Technical Support Group. Priority will be given to projects involved in cancer early detection research.

If you are interested in using this resource or wish to discuss your requirements further, please contact Yaozhi Lu (Lu) (yz.lu@ucl.ac.uk).

Imaging technologies and expertise 

Dedicated research scanners 

  • 3T MRI scanners, PET-MRI, SpinLab GE Hyperpolariser, access to London 7T MRI Centre, dedicated GMP facility for manufacture of novel imaging tracers 
  • Dedicated clinical grade MRI scanner available for pre-clinical scientists 

Collaboration with CRUK National Cancer Translational Accelerator (NCITA) infrastructure  

  • Multi-centre image repository to share curated and annotated imaging cohorts 
  • Establish consensus on image acquisition standards and processing 

Virtual Disease Frameworks 

REANIMATE 

The REANIMATE framework combines advanced ex vivo and in vivo imaging with mathematical models to create virtual 3D reconstructions of tumours and the tissue microenvironment. REANIMATE allows the simulation of tumour blood flow and the transport of biological fluids and their complex interaction with tissue microstructure. 

VERDICT-MRI 

The VERDICT (Vascular, Extracellular and Restricted Diffusion for Cytometry in Tumors) framework quantifies and maps histologic features of tumours in vivo. Mathematical modelling relates the MR signal to microstructural tissue features, such as cell size, density and vascularity, providing non-invasive estimates of histological features. Contact Laura Panagiotaki, e.panagiotaki@ucl.ac.uk.

In addition to our imaging expertise, our unique cancer cohorts are characterised by three principle attributes. Each cohort: 

  • Explores a novel early cancer detection strategy 
  • Assesses the incremental utility of imaging and biomarkers 
  • Is linked to downstream clinical outcome. 

Early Detection Cohorts

The UCL ACED Centre will be dedicated to rendering these cohorts FAIR compliant so they can be fully exploited by Alliance Partners to accelerate early detection research.

The PROMIS study showed that Multi-parametric MRI (mp-MRI) was twice as good at detecting clinically significant prostate cancer compared to transrectal ultrasound guided biopsy in men following raised Prostate Specific Antigen levels (Ahmed et al. Lancet 2017).  

PROMIS comprises a unique cohort of men (n = 740) that underwent mp-MRI of the prostate as well as 5mm template prostate biopsy (the reference standard) and 12-core transrectal ultrasound directed biopsy. 

The PROMIS cohort represents the most fully annotated cohort of men histopathologically, which is unlikely to be reproduced given the enormous burden of 5mm template sampling. It represents a very important historical resource in which the association between imaging and pathology (in the process of being made fully digitised) can be explored using AI. The cohort has blood, serum and urine samples from patients recruited at UCLH. 

PROMIS datasets

De-identified and pseudonymised matched datasets are available for 576 individuals from the PROMIS study

  • Prostate mp-MRI (DICOM)
  • Clinical MR imaging reports (pdf)
  • Histopathology reports (pdf)
  • Pseudonymised clinical data (csv)

Access to these datasets for research purposes requires application to the ReIMAGINE Trial Management Group, please contact reimagine@ucl.ac.uk for details. 

INNOVATE comprises a cohort of 365 men recruited over 3 years with suspected prostate cancer. This cohort represents a unique prospective cohort combining serum and urinary biomarkers with detailed histopathology and standard and enhanced MRI sequences (multiparametric MRI, VERDICT MRI and Luminal Index MRI) (Johnston et al. BMC Cancer 2016). 

ReIMAGINE Prostate Cancer Risk is a multi-centre, prospective, observational, longitudinal cohort study of men referred to secondary care with a suspicion of prostate cancer. The aim of the study is to develop a robust baseline risk stratification system for men at risk of prostate cancer. ReIMAGINE will generate a deeply phenotyped cohort of newly diagnosed men undergoing MRI-directed prostate biopsy (n = 1000) with combined MRI, targeted biopsy and whole genome sequencing, in addition to multiple commercial and non-commercial tests (circulating DNA, urinary and blood proteins, circulating antibodies).

The UKCTOCS Longitudinal Women’s Cohort (UKLWC) is the bioresource that has been built during the course of the United Kingdom Collaborative Trial of Ovarian Cancer Screening (UKCTOCS), one of the world’s largest randomised controlled trials. 202,638 women from the general population joined the trial between 2001-2005. All women were free of any active malignancy at randomisation. During the ongoing follow-up using electronic health record linkage for the last 20 years (cancer and death registries, hospital episode admissions), many have developed a number of cancers which are of interest to researchers. All women donated serum samples at baseline and 50,000 participants donated a unique set of longitudinal serum samples (median 9). These samples precede cancer diagnosis. The biobank has supported numerous research collaborations with academic and commercial partners into the early detection of cancer biomarkers.

For further information please contact Dr Sophia Apostolidou: s.apostolidou@ucl.ac.uk. 

The ADEPTS study (Accelerated Diagnosis of neuroEndocrine and Pancreatic TumourS), funded by Pancreatic Cancer UK (CI SP, Nov 2018-Apr 2021) is a national programme aiming at improving the diagnosis of patients with pancreatic cancer by setting up a brand-new biobank of pre-diagnosis samples from patients with symptoms (high-risk cohort) and from those with a genetic risk of developing pancreatic cancer. We aim to identify symptoms earlier and via biomarker discovery and validation, to develop sensitive and specific tools for the early detection of pancreatic cancer using less invasive approaches.  

To date, we have collected more than 1,500 urine and blood samples, alongside clinical data, which has already led to further funding and novel findings (e.g. Immunovia AB, Lund, Sweden. A prospective, multi-center investigational study of IMMrayTM PanCan-d diagnostic platform for early detection of PDAC in high-risk populations). 

For further information contact: Prof Stephen Pereira, stephen.pereira@ucl.ac.uk or Dr Pilar Acedo, p.nunez@ucl.ac.uk. Twitter @PereiraGroup 

The SUMMIT study is a prospective cohort study which aims to recruit 25,000 individuals at high risk of developing lung cancer. The study is a collaboration between University College London (UCL), University College London Hospitals (UCLH) and GRAIL inc. a US healthcare company. Participants will have annual CT scans for two years with blood sampling.