Poster P-55
S428
Imaging
IDENTIFICATION OF ABDOMINAL ADHESIONS USING DYNAMIC IMAGE REGISTRATION Benjamin P. Wright (1), John W. Fenner (1), Richard Gillott (2), Patricia V. Lawford (1), Paul Spencer (2), Karna Dev Bardhan (3)
1. Medical School - University of Sheffield, England; 2. Clinical Radiology – Rotherham General Hospital, England; 3. Gastroenterology – Rotherham General Hospital, England
Introduction Adhesions are an adverse reaction to the inflammatory response that follows infection or trauma affecting the abdominal viscera or the peritoneum. This may involve attachment of organs to the abdominal wall and/or other internal organs, leading to pain and possible intestinal obstruction. In the worst cases, surgical intervention may be required. Irrespective of the cause, non-invasive diagnosis is difficult and is often a diagnosis of exclusion. A smooth movement of the intraabdominal contents is expected during each respiratory cycle in healthy volunteers. Examination of patients with Crohn’s disease identified subtle disruptions in this pattern of movement. It was hypothesised that adhesions may be detectable by their disrupting influence on smooth visceral movement. Documented methods [Lienemann, 2000] of non-invasive diagnosis involve manual examination of abdominal motion, and highlight areas of abnormal movement in the presence of diaphragmatic and abdominal forces. In the present study we have developed novel image processing and registration methods to support automated identification of disturbances to smooth movement.
Method Standard, dynamic, non-contrast MR images were obtained from healthy volunteers. Patients with Crohn’s disease were also examined, following small intestinal resection and with positively identified adhesions. Application of the registration algorithm (ShIRT [Barber, 2005]) permitted analysis of the images, and computation of vector fields that denote image movement.
Figure 1: Left, MRI Sagittal abdominal section part of a cine sequence of 15 sections. Right, Contour plot displaying movement of the abdominal contents from the left image to the next in the cine sequence; contours display small to large movement using light to dark colours respectively.
Discussion The results of this pilot study are encouraging and suggest that complex patterns of abdominal movement can be quantified and displayed in a manner that can aid adhesion detection. Ideally, such methods would be used to draw the attention of the radiologist to anatomical zones that merit closer inspection. The raw clinical image data is undoubtedly difficult to interpret, but the use of sophisticated mathematical vector operators should facilitate better interpretation. In vitro models have been developed to explore the utility of the algorithms in the context of specific classes of abnormal movement and image distortion. This study demonstrates the feasibility of imaging augmented by mathematical analysis, as a means of characterising abdominal movement in the clinical setting and offers the possibility of a non-invasive method for detection of adhesions.
References Results Sagittal MRI sections from patients and volunteers were examined. Application of the image processing methods permitted visualisation of the data in vector and contour formats. The results show the presence of patterns of movement that invite classification as ‘smooth’ or ‘disrupted’ and may be indicative of pathology.
Journal of Biomechanics 41(S1)
D.Barber and D.R.Hose, J Medical Engineering & Technology, 29(2):53-63, 2005. A.Lienemann et al, Radiology, 217(2):421-425, 2000.
Acknowledgments BRET & EPSRC (UK)
16th ESB Congress, Posters