P239 X-ray phase contrast imaging for staging oesophageal tumours: preliminary results from the VIOLIN study

Introduction Oesophageal cancer is the 7th commonest cause of cancer death worldwide. Radiological staging of local oesophageal cancer is inaccurate. CT currently relies on attenuation of x-rays to generate contrast. Soft tissues have very similar attenuation properties so minimal contrast is generated. X-ray phase contrast imaging (XPCI) uses refraction of x-rays as they pass through tissue instead of attenuation and provides much higher soft tissue contrast. This technology can be tuned to a resolution of approximately 10 µm. This may allow for easy assessment of extent of disease infiltration. We aimed to use XPCI to image oesophagectomy specimens to assess pathological tumour and nodal stage for oesophageal cancer Methods Following ethical approval, 10 oesophagectomy specimens were obtained from patients having surgery for oesophageal cancers. These included both squamous and adenocarcinomas. Specimens were fixed in formalin for 12 hours. Sutures were placed through tissue to enable co-registration between CT slices and histology sections. For some scans, tissue was then dehydrated with graded ethanol for between 4.5 hours and 72 hours before being imaged. A Rigaku (MicroMax 007) xray source was used at 40 kV and 20 mA; a detector with 50µm pixel size; and sample and detector masks made of graphite substrate with gold overlay. Phase contrast was generated using edge illumination technique. We reconstructed the images using MATLAB® software. Specimens were returned for clinical histopathological assessment allowing correlation between H&E slides and CT images. Results We have performed 25 scans on 10 oesophagectomy samples and correlated them with histology Scans of samples in formalin failed to show adequate contrast between oesophageal layers to enable tumour visualisation and staging. Infiltrating the tissue with ethanol led to much better image contrast. We could easily identify mucosa, submucosa and both layers of muscle in reconstructed CT images. We also identified tumour infiltration through tissue layers and destruction of normal oesophageal morphology (figure 1). This was confirmed histologically and could be recognised by radiologists blinded to pathological staging This is the first time that XPCI has been used to image human oesophageal tissue. We have demonstrated the feasibility of the technique and the possibility of obtaining high resolution images which mimic histology with the extra benefit of demonstrating three dimensional structure.