Gastric cancer frequently has a poor prognosis due to late presentation of patients. There are over 500 patients diagnosed every year in Ireland (Irish Cancer Society). Unlike colorectal cancer where there is a screening programme in place (Bowelscreen) there is no screening test available for gastric cancer. However, there has been an increase in bariatric surgeries carried out in Ireland over the last decade with the development of infrastructure in that area in University Hospital Galway. The close examination of material from these cases for risk factors for gastric cancer including Helicocbacter pylori infection (25-30% of the Irish population) and intestinal metaplasia. The downstream effect of analysis of this material would be an increasing workload on histopathology resources including immunohistochemtry and special staining tests to confirm the presence of these risk factors.
We aim to develop a machine learning algorithm and pathway to identify both the presence of Helicobacters and intestinal metaplasia on standard histopathological slides using routine H&E stains without the need for ancillary tests and also reducing consultant time in screening these cases.
The study will involve scanning slides already taken for routine histopathological analysis from consented patients undergoing routine bariatric surgery or gastroscopy. These whole slide images will be analysed using an open source image analysis software package called QuPATH which is well established a leader in the field of image analysis software.
The results of this study will streamline the identification of important background risk factors in patients undergoing routine gastric examination. The study will be a proof of principle with all slides examined routinely by the gold standard of a consultant Histopathologist compared with the algorithm developed.