Precision Medicine Doctoral Training Programme

Computer-aided CT imaging and integration with molecular endotyping to stratify lung fibrotic disease

Project Details - Computer-aided CT imaging and integration with molecular endotyping to stratify lung fibrotic disease

Supervisor(s): Dr Nik Hirani, Prof Aziz Sheikh, Dr Kev Dhallwal & Prof Edwin Van Beek
Centre/Institute: Centre for Inflammation Research, Queen’s Medical Research Institute

Background

Lung fibrotic conditions are a significant burden of disease worldwide and are a cause of approximately 7000 deaths/year in the UK, most of which are due to idiopathic pulmonary fibrosis (IPF). The incidence of IPF is comparable to stomach, liver and cervical cancers, and the survival worse than for breast, colon and stage II lung cancer.

The fibrotic lung diseases, including IPF are highly heterogeneous, and their current classification is inadequate because: 1. It is overly reliant on lung biopsy, an invasive procedure with a 2-7% mortality that many patients will not undergo, leading to a diagnosis of  ‘unclassifiable disease’; 2. It does not reliably inform of either prognosis or treatment efficacy.

In contrast to lung biopsy, all patients will have high-resolution CT imaging as part of the diagnostic work-up, but the clinical reporting of CT’s is subjective and not quantitative.

We and others have studied automated CT texture analysis platforms, such as the Adaptive Multiple Features Method (AMFM) and the Computer-Aided Lung Informatics for Pathology Evaluation and Rating (CALIPER) and shown their potential clinical utility (1,2). Crucially these platforms have not yet been tested in longitudinal ‘real-world’ cohorts in which ground truth (survival, rate of decline in lung function, response to treatment) is known.

The Edinburgh Lung Fibrosis (ELF) Clinic, image-bank and biobank: We have established a unique prospectively populated database designed to capture the natural history of lung fibrosis allied to a gene- and bio-bank. This is the largest incident cohort of unselected lung fibrosis patients globally. The cohort from 01/01/07-31/12/15 consists of >1100 consecutively presenting patients with lung fibrosis. Less than 1% of our cohort has been lost to follow-up.  All patients have volumetric (high resolution) CT scans and >800 patients have serial scans. CT scans are hosted within National Services Scotland (NSS) and this is co-located with the Farr network in the Edinburgh Farr node, enabling a safe haven analytic environment for imaging, clinical and ‘omic data. This platform is being leveraged by Dr Dhaliwal for lung cancer diagnostics (‘LUNG SOLVE’ http://www.hra.nhs.uk/news/research-summaries/lung-solve-version-3/). 

We have ‘banked’ serum and genomic DNA samples from 1070 subjects from our cohort with longitudinal follow up of >12 months (median 4.8 years) and a complete dataset of variables including disease phenotype according to clinical-,CT-,biopsy-category, serial lung function. We are currently interrogating the serum molecular and genetic signatures from patients, beginning with a semi-biased approach according to known IPF-related targets and our own hypothesis-driven concepts (1,2) and extending to a genome-wide (GWAS and whole exome-sequencing) search of candidate genes. We have established international collaborations with groups that have similar but smaller, less mature datasets in which to validate our findings.

Aims

  1. Develop and test an interactive protocol for classification of CT scans using in house (CALIPER) texture analysis platforms.

  2. Interrogate the Edinburgh lung fibrosis cohort of CTs with a texture analysis platform and integrate with molecular endotype data.

  3. Test and validate key findings in separate datasets globally

Training Outcomes

  1. Methodology to iteratively develop and test automatic and interactive image classification 

  2. Integrate large real world ‘omic datasets through standard statistical and machine learning techniques

  3. Develop collaborative interdisciplinary skills

The project would be particularly suited to candidates with degrees in computational engineering, digital imaging processing, physics, machine learning or allied disciplines. 

References

  1. Idiopathic Pulmonary Fibrosis: Adaptive Multiple Features Method Fibrosis Association with Outcomes. Salisbury ML, Lynch DA, van Beek EJ, Kazerooni EA, Guo J, Xia M, Murray S, Anstrom KJ, Yow E, Martinez FJ, Hoffman EA, Flaherty KR; IPFnet Investigators. Am J Respir Crit Care Med. 2016 Oct 21.
  2. Imaging biomarkers in the clinic. van Beek EJ. Biomark Med. 2016 Oct;10(10):1073-1079
  3. Nicol L, Mills R,  Seth S,  MacKinnon A,  McFarlane P, William W,  Stewart G, Howie S,  Dhaliwal D,  Murchison J,  Hirani N. Prognostically predictive biomarkers for IPF; a longitudinal cohort study of treatment naive patients. ABSTRACT; Quarterly J Med 2016 http://qjmed.oxfordjournals.org/content/109/suppl_1/S38.1
  4. O'Dwyer DN, Armstrong ME, Trujillo G, Cooke G, Keane MP, Fallon PG, Simpson AJ, Millar AB, McGrath EE, Whyte MK, Hirani N, Hogaboam CM, Donnelly SC..The Toll-like receptor 3 L412F polymorphism and disease progression in idiopathic pulmonary fibrosis. Am J Respir Crit Care Med.2013 Dec 15;188 (12):1442-50

Apply Now

Click here to Apply Now

  • The deadline for 17/18 applications will be 5pm on Thursday 23rd March 2017. 
  • Applicants must apply to a specific project, ensure you include details of the project you are applying to in Section 4 of your application. We encourage you to contact the primary supervisor prior to making your application.  
  • As you are applying to a specific project, you are not required to submit a Research Proposal as part of your application. 
  • Please ensure you upload as many of the requested documents as possible at the time of submitting your application.