REAL-TIME DETECTION OF IMAGING ERRORS IN THE KNIFE-EDGE SCANNING MICROSCOPE THROUGH CHANGE DETECTION

Advances in high-resolution, high-throughput 3D microscopy techniques are enabling subcellular investigation of whole small animal organs such as the mouse brain. Knife-Edge Scanning Microscopy (KESM) is one such technique based on physical (or serial) sectioning to overcome diffraction limited imaging in optical sectioning approaches. However, due to the physical sectioning process depending on a mechanical process, vibration (chatter) and obstruction of the light path by floating debris can cause a varying degrees of image error. In this paper, we present a change-detection-based error detection method to minimize such errors. Change detection is done in three steps: preprocessing (subsampling and illumination equalization), change detection, and postprocessing (morphology-based operation). Based on the change detection results, a finite state machine is used to alert a human operator or stop the machine. The method has been tested on three KESM data sets, demonstrating its effectiveness. The approach is expected to be widely applicable to other physical sectioning microscopy techniques.