In this research, we suggest a novel Fully/Convolutional/Neural/Network/(F-C-N) engineering meaning to help the identification of variations from the norm, for example, polyps, ulcers, also blood, in gastrointestinal (G/I) endoscopy pictures. The projected engineering, termed Look-Behind/MFCN/(LB-MFCN), is fit for removing multi-scale picture includes through utilizing squares of similar convolutional coatings with various channel sizes. These squares are associated through Look-Behind (LB) associations, so the highlights they produce remain joined through highlights removed since behind layers, accordingly protecting the particular data. Besides, it has fewer open, limitations than regular Convolutional/Neural/Network-(C/N/N) structures, which creates it reasonable on behalf of preparing through littler datasets. This is especially valuable in restorative picture examination subsequently information accessibility is generally restricted payable to ethicolegal limitations. The presentation of LB-MFCN is assessed on together adaptable also remote case endoscopy datasets, arriving at 99.82% as well as 95.50%, as far as Area Beneath accepting working Characteristic individually.