Introduction - High-Throughput Image Reconstruction and Analysis - page 7

1.4 Part IV: Postprocessing
Once the microscope images have been acquired and analyzed, they need to be
stored in databases that facilitate their retrieval. The original images and the result
of processing also need to be visualized effectively, in order to provide the best
insight to the viewer.
1.4.1 Bisque: A Scalable Biological Image Database and Analysis Framework
In Chapter 13, Singh and Kvilekval describe how the availability of large image
databases and search technology has enabled a new level of hypothesis creation
and testing, thus transforming the field of biology. The emerging field of bioimage
informatics addresses a key concern, which is effective management and mining of
bioimage databases. The authors describe an image database solution they have
developed, called Bisque. This addresses several critical issues such as appropriate
data modeling, data integration from multiple sources, and scalability. The bulk of
existing search technology relies on textual keyword matching. Bisque goes beyond
this, and enables matching of images based on features derived automatically from
them. This solution is made scalable by using a services oriented architecture, where
services based on Web protocols link the user to the data.
1.4.2 High-Performance Computing Applications for Visualization of Large
Microscopy Images
In Chapter 14, Singh, Lin, Schulze, Peltier, Martone, and Ellisman examine the
problem of effective display, navigation, and manipulation of large microscopy
images. The creation of large multiresolution image datasets of a single biological
entity such as an organ opens up the possibility of visually inspecting the entire
dataset in a seamless manner. This allows scientists the ability to integrate informa-
tion across multiple scales, leading to better insight and understanding. However,
significant technological challenges need to be overcome before this vision can be
turned into reality. The large image sizes prevent them from being loaded onto
the memory of a single processor. Furthermore, for effective interactivity with the
data, computation needs to be done in near real-time. The visual display of the
large image sizes requires the use of multitiled displays. Due to all these consider-
ations, the use of high-performance computing becomes a necessity. The authors
describe a solution to this problem using a scalable cluster-based approach.
1.5 Conclusion
Though this topic covers a large number of relevant and interesting topics, the field
is vast, and there wil be many aspects missing. One of them is the description of a
single integrated system devoted to providing a comprehensive solution to image
processing tasks carried out on a parallel platform. We realize that to make this
happen, a concerted effort is required that combines the expertise of researchers
across academia, industry, and possibly governmental agencies. Though readers
may not find their specific problem addressed and solved in this topic, it is hoped