Efficient semi-automatic 3D segmentation for neuron tracing in electron microscopy images

Abstract

0.1. Background In the area of connectomics, there is a significant gap between the time required for data acquisition and dense reconstruction of the neural processes contained in the same dataset. Automatic methods are able to eliminate this timing gap, but the state-of-the-art accuracy so far is insufficient for use without user corrections. If completed naively, this process of correction can be tedious and time consuming. 0.2. New Method We present a new semi-automatic method that can be used to perform 3D segmentation of neurites in EM image stacks. It utilizes an automatic method that creates a hierarchical structure for recommended merges of superpixels. The user is then guided through each predicted region to quickly identify errors and establish correct links. 0.3. Results We tested our method on three datasets with both novice and expert users. Accuracy and timing were compared with published automatic, semi-automatic, and manual results. 0.4. Comparison with Existing Methods Post-automatic correction methods have also been used in [1] and [2]. These methods do not provide navigation or suggestions in the manner we present. Other semi-automatic methods require user input prior to the automatic segmentation such as [3] and [4] and are inherently different than our method. 0.5. Conclusion Using this method on the three datasets, novice users achieved accuracy exceeding state-of-the-art automatic results, and expert users achieved accuracy on par with full manual labeling but with a 70% time improvement when compared with other examples in publication.

Efficient Classifier Training to Minimize False Merges in Electron Microscopy Segmentation

“…A comprehensive search for segmentation errors is not practical in connectomics [30]. The presence of no or minimal undersegmentation is a prerequisite for applying intelligent proofreading methods that avoids 100% screening of the volume [17]. Efficiency in creating a groundtruth and training will also be valuable for computing different predictors for different neuropils for large scale reconstruction.…”

Microscopy Image Browser: A Platform for Segmentation and Analysis of Multidimensional Datasets

“…Next, by using various quantitative features (e.g., intensities, texture, and morphology) of the labelled area, the classifier can extend the local segmentation to cover the whole dataset. In some cases, the results can be improved when classification is combined with other methods, for example, the use of supervoxels with watershed [ 28 ] or graph-cut-based algorithms [ 29 ]. Usually, the use of supervoxels facilitates processing without significant degradation of segmentation results.…”

Automatic segmentation of high pressure frozen and freeze-substituted mouse retina nuclei from FIB-SEM tomograms

“…For boundary segmentation, membrane signals are usually enhanced by filtering and then traced by thresholding (Martinez-Sanchez et al, 2014). For region segmentation, a superpixel-based method is often used to over-segment the image and then a split-and-merge strategy is used for region grouping (Jones et al, 2015). A classifier trained with human annotated data can also be used for direct pixel classification (Sommer et al, 2011).…”