Adenoma Dysplasia Grading of Colorectal Polyps Using Fast Fourier Convolutional ResNet (FFC-ResNet)

Colorectal polyps are precursor lesions of colorectal cancer; hence, early detection and dysplasia grading essential for determining cancer risk, the possibility developing subsequent polyps, follow-up recommendations. The significant contribution this study is development an enhanced deep-learning model called Fast Fourier Convolutional ResNet (FFC-ResNet) to classify grades polyps. It based on ResNet-50 architecture uses cross-feature fusion, which combines local features extracted using conventional spatial convolution with global convolution. Because compensatory effect between features, learnability performance FFC-ResNet have increased. proposed was developed tested UniToPatho, a dataset containing 7000 ?m 800 hematoxylin-and-eosin (H&E)-stained images. And favourable sensitivity 0.95, specificity 0.93, balance accuracy 0.94, precision F1 score 0.95 AUC 0.99 obtained polyp patches.