No description available.

No description available.

Fluorescence microscopy is an essential tool for imaging subcellular structures in tissue. Two-photon microscopy enables imaging deeper into tissue using near-infrared light. The use of image analysis and computer vision tools to detect and extract information from the images is still challenging due to the degraded microscopy volumes by blurring and noise during the image acquisition and the complexity of subcellular structures presented in the volumes. In this thesis we describe methods for segmentation and detection of fluorescence microscopy images in 3D. We segment tubule boundaries by distinguishing them from other structures using three dimensional steerable filters. These filters can capture strong directional tendencies of the voxels on a tubule boundary. We also describe multiple three dimensional convolutional neural networks (CNNs) to segment nuclei. Training the CNNs usually require a large set of labeled images which is extremely difficult to obtain in biomedical images. We describe methods to generate synthetic microscopy volumes and to train our 3D CNNs using these synthetic volumes without using any real ground truth volumes. The locations and sizes of the nuclei are detected using of our CNNs, known as the Sphere Estimation Network. Our methods are evaluated using real ground truth volumes and are shown to outperform other techniques.

Obtaining an accurate estimate of a land-cover classifier's performance over a wide geographic area is a challenging problem due to the need to generate the ground truth that covers the entire area that may be thousands of square kilometers in size. The current best approach constructs a testing dataset by drawing samples randomly from the entire area --- with a human supplying the true label for each such sample --- with the hope that the selections thus made statistically capture all of the data diversity in the area. A major shortcoming of this approach is that it is difficult for a human to ensure that the information provided by the next data element chosen by the random sampler is non-redundant with respect to the data already collected. In order to reduce the annotation burden, it makes sense to remove any redundancies from the entire dataset before presenting its samples to a human for annotation. This dissertation presents a framework that uses a combination of clustering and compression to create a concise-set representation of the land-cover data for a large geographic area. Whereas clustering is achieved by applying Locality Sensitive Hashing (LSH) to the data elements, compression is achieved through choosing a single data element to represent a given cluster. This framework reduces the annotation burden on the human and makes it more likely that the human would persevere during the annotation stage. We validate our framework experimentally by comparing it with the traditional random sampling approach using WorldView2 satellite imagery.