Microscopy Image Analysis

Open Workshop on Microscopy Image Analysis in ITK and VTK

The goal of this workshop is to foster the growth of a community of scientists interested in microscopy image analysis for biology using ITK and VTK

Background

Optical microscopy is by far the most common form of imaging in biomedical research due to its high spatial resolution (subcellular), high specificity (molecular in the case of fluorescence), and suitability for use in living specimens. A Google Scholar search for "fluorescence microscopy", only one of several types of optical microscopy, returns 1.7 million articles compared with < 1 million for "MRI". Traditionally, the vast majority of these users of microscopy have performed qualitative analysis on a small number of images, but this is quickly changing. There is increasingly a need to perform quantitative analysis on microscopy images and to perform this analysis on large image sets (>100,000 images). In addition to higher throughput, recent advances in microscopy have made higher dimensional imaging commonplace. Researchers now routinely capture microscopy images over the dimensions of space (x,y,z), time (t), and multiple channels of color (lambda). Due to the large datasets, high dimensions, and complexity of analysis, current approaches to microscopy image analysis relying on Java, Matlab, and “home brew” applications are reaching their limits. We believe that a community based effort centered on developing microscopy-specific algorithms and applications built on the C++ class libraries of VTK and ITK represents the best path forward.

Focus

The focus of this workshop will be on segmentation and tracking of cells in optical microscopy images. Segmentation and tracking of cells represents a very common problem in microscopy image analysis. Although there is a common pipeline for many users (e.g. image preprocessing to remove noise, detection of seeds, detection of cells at single timepoints, tracking movements over time, data analysis) the algorithm parameters and algorithms themselves are often dependent on the specifics of the experimental setup. There is thus a strong need to develop a framework to allow users to choose algorithms and tune parameters to most importantly achieve robust segmentation and secondarily minimize computational cost.

Format

The format for this meeting will be as a “track” within the NAMIC Project Week 2010 meeting at MIT in Boston, MA on June 21-25. Participants in this workshop should all have specific coding projects relating to cell segmentation and tracking that they wish to complete within the week. Ideally these projects should be collaborative. At the beginning of the meeting on Monday, workshop participants will present a 1 slide summary of the goals of their project as part of the overall meeting. For the rest of the week, workshop participants will sit in a common area and code on their projects. These project weeks tend to be quite productive because of the concentration of available expertise at the meeting. During the week we will also break from the coding to have a more formal discussion of our current individual efforts, the needs of the microscopy community, the technical issues of combining and exchanging code, and how we should move forward.

Participants

Please add your name to the list if you are interested in participating in this workshop

1. Raghu Machiraju, Ohio State University
2. Kannappan Palaniappan, University of Missouri
3. Badri Roysam, Rensselaer Polytechnic Institute
4. Arnaud Gelas, Harvard Medical School
5. Kishore Mosaliganti, Harvard Medical School
6. Nicolas Rannou, Harvard Medical School
7. Antonin Perrot-Audet, Harvard Medical School
8. Lydie Souhait, Harvard Medical School
9. Sean Megason, Harvard Medical School
10. Luis Ibanez, Kitware