Anonymous Data Sharing for Small Bowel

Overview

Crohn's disease is characterised by an abnormal immunological response within the bowel wall leading to abnormal wall thickening, stricturing (narrowing), fistulation (abnormal connections) to adjacent organs and strictures (narrowing), abnormal motility, and local sepsis (infection). Radiological imaging of the small bowel defines diagnosis, disease extent, biological activity and complications and is vital for timely and efficacious clinical management. Small bowel magnetic resonance imaging (MRI) does not impart ionising radiation and is proving to be a safe, well tolerated and robust method of assessing the small bowel and is widely implemented in the NHS and Europe. MRI evaluates multiple disease related features such as bowel wall thickness and motility which are proving increasing reliable for disease identification, staging, therapeutic guidance and assessment of treatment response. Currently however radiologists must manually make these measurements which are time consuming and difficult. There are no computer tools that can quickly and accurate make relevant measurements on MRI to guide patient management. Radiologists at University College Hospital have formed collaborations with groups within University College London (UCL) and around Europe. The investigators have received grant funding to develop computer software to accurately assess the small bowel using MRI over 3 years. To develop this software, it is necessary to use anonymised datasets from patients with and without Crohn's disease undergoing small bowel MRI. Part of this project will require prospective collection of MRI data which has been granted ethical approval as a major amendment to a currently running project (09/H0714/62). The department of Radiology at UCLH has been running a clinical small bowel MRI service since 2005 and have several hundred datasets on its PACS. This current ethics application seeks permission to datashare with the collaborative partners fully anonymised MRI datasets and relevant clinical data from patients previously undergoing small bowel MRI for clinical indications at UCLH.

The investigators wish to share with collaborative partners selected fully anonymized MRI datasets and associated relevant clinical data (such as blood tests results, endoscopic findings, histology) acquired as part of routine clinical practice at UCH since 2005. The collaborative partners are: 1. The Centre for Medical Image Computing (CMIC) at UCL 2. A European consortium (including UCL as partner and UCLH as subcontractor) which has been awarded an FP7 European grant: University of Delph, Amsterdam medical centre, Eldgenössische Technische Hochschule Zürich, Switzerland, Zuse Institut Berlin, Germany, Biotronics3D Limited UK and Vodera Limited UK. The investigators wish to share a total of 75 datasets with the FP7 consortium and 300 with CMIC. Process for the sharing of anonymized datasets acquired as part of clinical practice. Many patients undergoing small bowel MRI at UCH often also undergo additional tests as part of their usual clinical care which are recognized as good standards of reference against which the investigators can validate our MRI findings. Notable examples are blood tests (eg CRP), and endoscopy and biopsy. Furthermore, many patients have normal examinations and these datasets are also very useful in software development to define a standard of normality. Suitable datasets will be found as follows: 1. Search of the PACS database to identify all small bowel examinations performed at UCH 2. Use clinical sources freely available to hospital clinical staff via the CDR web system at UCH to triage these patients into relevant clinical groups notably those who have normal small bowel examinations, and those with abnormal studies who have a relevant clinical standard of reference performed within 6 weeks (CRP level, endoscopy, biopsy) 3. Select a sample of relevant datasets for use in software development. The total number to be data shared with the FP7 consortium listed above will not exceed 75. The total number of datasets to be data shared with UCL (CMIC) will not exceed 300. The larger number shared with CMIC reflects their need for a large database of normal examinations to develop motility software.