Whole slide images stitching for osteosarcoma detection
Whole slide imaging (WSI), the conversion of conventional glass-slide histopathology to a digital image, is at the heart of telepathology, which will allow disease diagnosis and the practice of pathology from a distance. Scanning systems that generate WSIs at 20x and 40x resolution allow practically equal precision to that available by light microscopy critical for pathologists to interpret specimens. WSI image stitching allows the combination of multiple WSIs into a single representative specimen of the tissue in question. This process suffers from problems of high resolution, disorientation, and the presence of white spaces. In this paper we discuss an image stitching technique based on pairwise image correlation and using gross image information. We are given a bone I which is a global image and a subdivision of I into m cells. The m cells' resolution is higher than the gross image. We calculate the pairwise correlation of the individual images which are numbered in certain order and perform seamless image stitching. We use transformation and rotation in case the image is misaligned.
