PTU-024 Photometric stereo reconstruction for surface analysis of mucosal tissue

Introduction The American Society of Gastroenterology endoscopy led Preservation and Incorporation of Valuable endoscopic Innovations initiative has identified real time polyp detection diagnosis as one of the next major technology-driven changes in endoscopy.1We have recently described a novel photometric stereo (PS) imaging sensor for endoscopy imaging in a porcine model.2Following image acquisition, reconstruction of the surface data is necessary to calculate the shape index (SI) to identify regions that are locally spherical, suggestive of polyps to aid polyp detection. Method Using a porcine gut model, photometric images were captured using a six-light source PS setup as previously described.2Surface analysis of the obtained surface data was performed: Derivatives of the height fields arranged on a square lattice were calculated using finite differences, and used to characterise the differential geometry using the principal curvatures.3Surface measures analysis: for each point on the surface, the shape index (SI) was computed and used to measure the local shape: SI = 1/2 – 1/π tan−1 ((K1+K2)/(K1-K2)) ResultsAbstract PTU-024 Figure 1 Porcine colonic data captured using the photometric stereo system. Left to right: Colour image, Normal map, Reconstructed height map, SI image top view, SI image side view Conclusion Using a novel PS image acquisition 3D reconstruction was obtained on colonic mucosa. We observe that the recovered 3D surface retains the surface geometry in the captured areas and important structural information at a fine level of detail, even in the presence of numerous specular reflections. This is highly significant for automated processing and analysis of surface abnormalities. Disclosure of interest None Declared. References Rex, et alet al. Gastrointest Endosc 2011;73:419–22 Poullis, et alet al. Gut 2014;63(Suppl 1):A46 do Carmo M. Differential geometry of curves and surfaces. Prentice Hall, 1976, ISBN:0132125897

Introduction High levels of recurrence and residual tissue have been reported following piecemeal endoscopic resection (pEMR) of large colorectal polyps with figures approaching 30% described. 1 The application of argon plasma coagulation (APC) has been advocated to reduce recurrence following pEMR, with early data appearing to support the use of APC to treat tiny polyp fragments not amenable to endoscopic snare resection. 2 More recent data however, appears to oppose this view, suggesting no change in or even increased recurrence following adjunctive APC use. 1 With a lack of standardised APC application in relevant international case series, many of which include only small sample sizes, the position regarding its benefit remains controversial.
Method Retrospective analysis was conducted on 153 pEMR procedures undertaken in the Bowel Cancer Screening Programme (BCSP) between 2011-12. Recurrence rates on first endoscopic follow-up were compared between 2 groups (APC use vs no APC use). Statistical analysis was performed using the chi-squared test.
Results Recurrence was detected on first endoscopic follow-up in 25.5% of cases. A lower level of recurrence was seen in the APC group compared with the non APC group (17.9% vs 31.4%), with statistical significance seen on 1-sided but not 2sided chi square analysis (p = 0.042 (1-sided), p = 0.064 (2sided)) Abstract PTU-023 Table 1 Piecemeal recurrence on first f/up (%) Total ( Conclusion While statistical significance was only seen on 1sided chi-square analysis, the level of recurrence was over 40% lower in the APC group (17.9% vs 31.4%). It can be argued that our study may be underpowered and that the result indicates likely clinical significance. Our data does not support the data from other studies indicating a possible detrimental effect of APC. Whilst the use of APC is likely to be more standardised within the BCSP, the exact circumstances of its use, such as the amount of residual tissue post snare resection remaining prior to APC use in other studies, is unclear. We believe APC is appropriate therapy for tiny residual polyp fragments post snare resection, but should not be used on larger tissue areas. A large randomised controlled trial with a standardised protocol for APC application would add further evidence. Disclosure of interest None Declared. Introduction The American Society of Gastroenterology endoscopy led Preservation and Incorporation of Valuable endoscopic Innovations initiative has identified real time polyp detection diagnosis as one of the next major technology-driven changes in endoscopy. 1 We have recently described a novel photometric stereo (PS) imaging sensor for endoscopy imaging in a porcine model. 2 Following image acquisition, reconstruction of the surface data is necessary to calculate the shape index (SI) to identify regions that are locally spherical, suggestive of polyps to aid polyp detection.
Method Using a porcine gut model, photometric images were captured using a six-light source PS setup as previously described. 2 Surface analysis of the obtained surface data was performed: Derivatives of the height fields arranged on a square lattice were calculated using finite differences, and used to characterise the differential geometry using the principal curvatures. 3 Surface measures analysis: for each point on the surface, the shape index (SI) was computed and used to measure the local shape: SI = 1/2 -1/p tanÀ1 ((K1+K2)/(K1-K2)) Results Abstract PTU-024 Figure 1 Porcine colonic data captured using the photometric stereo system. Left to right: Colour image, Normal map, Reconstructed height map, SI image top view, SI image side view Conclusion Using a novel PS image acquisition 3D reconstruction was obtained on colonic mucosa. We observe that the recovered 3D surface retains the surface geometry in the captured areas and important structural information at a fine level of detail, even in the presence of numerous specular reflections. This is highly significant for automated processing and analysis of surface abnormalities.

Disclosure of interest None Declared.
Introduction Documentation of Rockall score (RS) in patients with AUGIB should be accurate in order to prioritise patients for gastroscopy. We conducted a pilot study in 2012 which showed that RS was calculated incorrectly in 46% patients. We investigated this further with a larger cohort of patients. Method Information was collected retrospectively on patients who presented with presented with AUGIB from February 2012 to February 2013. Patients who had out-of-hours gastroscopy excluded. Demographics, time to perform gastroscopy, RS documented by the doctor requesting the gastroscopy, RS calculated by analysing patient records (including A&E, paramedic records) were collected. We analysed patients whose RS was either underscored or overscored by the requesting doctor (this was compared to the actual score calculated by us) and correlated the calculated RS with the need for endoscopic intervention. Results There were 248 patients with 149 (60%) males, age 16 to 98 years (mean 69). Presenting symptoms were melaena in 129 (51%), haemetemesis 35 (14%), coffee ground vomiting 64 (25%), rectal bleed 17 (7%), melaena and haematemesis 4.
RS was calculated incorrectly in 194 (78%) patients with 110 (44%) over scored (Group A), 84 (34%) underscored (Group B). Mean RS on request form for Group A patients was 3.64 ± 1.97 while the calculated RS was 1.85 ± 1.61. In Group B, mean RS documented was 1.8 ± 0.96 while the calculated RS was 3.9 ± 1.04. Mean time from electronic booking to gastroscopy was 2.02 days -1.67 days for Group A and 2.04 days in Group B.
Of the 248 patients, 37 (15%) required endoscopic intervention. In Group A, 16 (15%) needed endoscopic intervention, compared to 11 (13%) in Group B. Ten (19%) of the 54 patients scored correctly required endoscopic therapy. Conclusion This study shows that it is important to document the RS at the time of first presentation of AUGIB rather than at the time when the patient is seen by the admitting doctor. This study however does not show a direct correlation between the RS and endoscopic intervention. It emphasises the fact that all doctors should be educated about the importance of documenting the correct RS as more than three quarter of the RS on request forms were found to be inaccurate. Introduction British Society of Gastroenterology (BSG) guidelines recommend cytological analysis of brushings taken from biliary strictures to support diagnosis of malignancy in suspected individuals. There is wide variation reported in literature on test sensitivities ranging between 16% and 65%. We set out to report a single centre experience on the diagnostic yield of cytology specimens of biliary strictures obtained from ERCP. Method Retrospective analysis on all biliary brushings taken between 2008 and 2011 was performed. All ERCPs were performed by two experienced endoscopists. Routinely, the cytology brush is sent to the lab immediately after the procedure, analysis undertaken by liquid-based methodology and results verified by a histopathologist with special interest in pancreato-biliary disorders. The brushing result was then correlated with the final diagnosis ('suspicious' and 'atypical cells' in cytology report were considered malignant for audit purposes). Results Cytological specimens taken from 89 patients supported diagnosis of benign or malignant pathology in 67 showing an overall sensitivity of 75.3%.

Disclosure of interest
Abstract PTU-026 Table 1 Cytology