A Computational Approach to Edge Detection
IEEE Transactions on Pattern Analysis and Machine Intelligence
IEEE Transactions on Pattern Analysis and Machine Intelligence
| Hi-index | 0.00 |
Digital microscopy is a field that is becoming increasingly popular for tissue imaging. Most digital microscopes have a limited field-of-view, requiring the acquisition of multiple tiles across the tissue that are then stitched together in software. During the acquisition, however, the microscope may intermittently fail to focus correctly, which will result in out-of-focus images that no longer provide diagnostic value. While many approaches have been proposed to address this using absolute measures of focus quality, we here introduce a novel approach that instead operates on the overlap regions between acquired images. This provides a relative measure that is independent of tissue type and staining protocol. These automatic measures can then be used to identify failed images that need to be re-acquired. Our quantitative and qualitative results on large datasets demonstrate the accuracy and robustness of the approach.