http://www.neurolens.org
...from:
http://developer.apple.com/business/macmarket/neurolens.html
NeuroLens is a faster, more powerful, user-friendlier brain imaging
application that owes its existence both to the insight of Rick Hoge,
its developer, and to the unique features of the Macintosh platform. A
Mac is a powerful and versatile tool in the life sciences—from the UNIX
core of Mac OS X and the power of the G5 processor, to the Xcode Tools
and the Mac’s native graphics capabilities. It’s a platform that lets
researchers focus on their work instead of learning programming and
waiting for results.
Many of the previous brain imaging applications used by Rick and his
collaborators were written for an earlier generation of computer
workstations with performance and memory limitations that are now
archaic. Rick and his team saw the opportunity with Mac OS X to develop
an imaging application with modern computer performance parameters in
mind. “The Apple platform is the largest installed base of any UNIX
variant,” Rick explains. “In addition, it is extremely prevalent in the
life sciences, and increasing in popularity and in adoptions in
neuroimaging labs. It was perfect for what we needed to accomplish.”
The result is NeuroLens, an integrated visualization and analysis
package for quantitative physiological neuroimaging, now in public
Beta. It is a research imaging tool that is easy to use, intuitive, and
has the capabilities to analyze and combine data from many different
sources in forms that are extremely useful for researchers. NeuroLens
was developed by Hoge at the A. A. Martinos Center for Biomedical
Imaging in Charlestown, Mass. Rick is a researcher at Massachusetts
General Hospital, a faculty member at the Harvard Medical School,
Department of Radiology, whose interests focus on cerebrovascular
physiology and the physics of how this affects the signal screen seen
in MRI scans. The application is targeted specifically at the research
community, who works with large data sets—neurologists, biologists, and
neuropsychologists, all doing basic research on brain function. The
work is funded by the Office of National Drug Control Policy as part of
a project to understand the genetic bases of addiction and depression.
Discovering a Familiar UNIX Environment on Mac OS X
“We were using UNIX software systems, and when Mac OS X arrived with
its UNIX core, we realized that this was an ideal development
opportunity,” Hoge explains.
“Apple has a strong commitment to consumer-oriented systems that are
easy to use, and we needed a program that eliminated the need for users
to learn advanced UNIX concepts. In addition, many life sciences folks
have Macs already, and UNIX was familiar to the developers.”
First, Rick ported their current UNIX code to Mac OS X, which worked
easily. The entire imaging team was very happy to have their UNIX
applications on the same machine as their word processing and email
programs. He adds, “But we saw immediately how outdated our application
interfaces were after running these X11 applications side-by-side with
Cocoa applications on Mac OS X. While the Mac became our new UNIX
workstation, our legacy Unix applications were missing out on the huge
performance boost offered by the AltiVec processor, and the advanced
user interface refinements available through Cocoa.”
The next logical step then was to design an application that could
take advantage of this new platform.
Advantages of Mac OS X Development
It was an easy choice for Rick to chose Mac OS X as the development
platform for NeuroLens, a well-engineered, high performance
application, using the variety of Mac OS X development tools and
performance capabilities, not just its UNIX APIs. Some of the many
advantages of developing NeuroLens on the Mac included:
▪ A very refined user interface—NeuroLens was created using the
Xcode Tools, especially Interface Builder, with very little work on the
part of the developers.
▪ The highly optimized image and signal processing routines that
Apple has developed for its successful consumer multimedia products are
perfect for the requirements they have in medical imaging performance.
▪ Familiar, trusted UNIX APIs available under the hood, so that
previous functionality was not wasted.
▪ Powerful tools like Xcode and Cocoa that made it possible to
develop full-powered applications quickly, with a wide range of
functionality, without writing a lot of code.
Cocoa Makes Development Easy
“We were extremely impressed with the Mac OS X development tools,”
Rick says. “We ran through Cocoa tutorials initially and found them
remarkable—in relatively short order I went from having nothing to
having a complicated application, with very little effort on my part. I
realized then that I could write a complicated imaging application with
little work and taking advantage of the extremely fast performance of
the OS, as well as its graphics and multimedia capabilities.”
“The appeal of Cocoa for science users is that Objective-C is a simple
superset of standard C, and it is easy to drop in or integrate existing
C code or algorithms (unlike using Java, which would otherwise be
another excellent choice for an imaging application), while also
producing a great user interface,” Rick adds.
“Interface Builder made it easy to build polished user interfaces, and
features like Cocoa bindings made it easy to link the interfaces to
program logic.” The integration of old code into the existing Mac OS X
development framework is supported by Cocoa, rather than keeping it
separated, which results in code that is easier to maintain, and that
has a better user interface. “ It’s really all about ease of use,” he
summarizes. “We made sure that we used the Apple Human Interface
Guidelines when designing the NeuroLens interface, and that was
extremely helpful.”
The NeuroLens application is designed such that all processing
operations (actions that take one or more datasets read from disk as
input to generate a new dataset as output) are implemented as plugins.
Plugin bundles for some standard processing operations are included
within the application bundle for NeuroLens, and researchers can write
their own plugins to extend the functionality of NeuroLens. To make
this easy, Hoge wrote a template which, when installed, allows a
developer to create a new Xcode project containing a skeletal but fully
functional plugin complete with a basic user interface.
Development using Xcode also gives users the ability to get data easily
from the program into other documents. Brain images from NeuroLens can
be dragged and dropped into Microsoft Word documents, emails, or
presentations. Graphs of brain signals can be easily copied and pasted.
“Cocoa makes it very easy to go from image display in the visualization
environment to export of high quality graphics for presentation,” Rick
says. “Because of the deep integration of PDF into the operating system
and Cocoa frameworks, it’s very easy to drag a graph from the
visualizer into a Keynote presentation as a high-quality vector PDF
representation with nice anti-aliased fonts. This was very difficult in
older UNIX/X11 apps in which typically we were limited to making a
bitmap screen grab of a plot. And all this can be achieved with
relatively little developer effort using the standard
drag-and-drop/cut-and-paste interface familiar to Mac users.”
By redesigning data structures from scratch based on the memory
capacity of modern computers, it was possible to alleviate much of the
disk I/O inherent in older applications and also keep more data in
contiguous memory—greatly improving cache performance and memory
efficiency. This would accelerate performance on any platform, but the
additional performance gains afforded by adapting code for linear
algebra and DSP functions to use the AltiVec processor—the 128-bit
vector processing unit available on G4 and G5 chips that permits math
operations to be done in parallel—resulted in an unprecedented level of
performance.
NeuroLens Solves Command Line Woes
Much of the imaging software in common use now was first developed in
the mid-1990s, by physicists using difficult-to-use UNIX command-line
code. The engineering behind that software is now dated, focusing on
things like a small memory footprint—performance bottlenecks were
completely different than they are with current systems. Today, brain
imaging software is being developed primarily by people in the life
sciences and psychology, rather than physicists, and the training costs
for them to come up to speed on this older software (as they frequently
do not have UNIX or programming backgrounds) is quite high.
NeuroLens takes care of these issues. “It can be challenging for a
developer or researcher to deal with all the housekeeping associated
with file formats and accurate visualization,” Rick says. There is
often need to write custom code for data analysis, and the custom Xcode
template allows them to do that without having to worry about detailed
programming or how the data is arranged on disk or how to display it.
With the Xcode template, a grad student who wants to perform a custom
processing operation on a dataset can just create a new project in
Xcode and immediately gain access to simple data structures. This
allows anyone with a reasonable background in programming C to
implement their algorithm. Hoge explains, “Such customization provides
the developer/researcher with ‘free’ access to multiple complex file
formats like DICOM, MINC, as well as Analyze and AFNI, and provides an
automatic display of results and the ability to save them to disk.
“We plan to release NeuroLens under an open source license and
distribute an Xcode template with it so that users can create custom
plug-in models with very little work, specific to their own research
needs,” says Hoge.
Better Interoperability with Multiple File Formats
Most older programs were also written for specific labs that used only
a few file formats and so tend to be very restrictive about what
formats they will support. Rick designed NeuroLens much like [Adobe]
Photoshop—it can read many data storage formats and get it into the
program. Cocoa provides the architecture to support this—achieving that
kind of broad functionality makes NeuroLens very useful in conjunction
with other packages, allowing researchers to use NeuroLens’ remarkable
imaging capabilities with data sets generated by other applications
without having to convert file formats on the command line. “We focused
on interoperability,” Rick explains. “In some instances, NeuroLens will
be used for its front end to read three dimensional surface images of
the brain with another program. In this way, it doesn’t replace other
tools, but can be used in conjunction with them. The user interface we
developed through Xcode makes the NeuroLens viewing environment much
easier to use than our previous generation of software.”
Faster Imaging with NeuroLens
Brain imaging from MRI studies result in a series of three-dimensional
images. Image analysis generally consists of image processing steps for
data quality improvement, followed by statistical analysis to identify
regions of brain activation during a task or stimulus that was applied
when the subject was scanned.
“For example,” Rick says, “a series of 3D images must be aligned to
reduce the effect of subject motion and are often spatially smoothed in
3D to improve signal-to-noise ratio. Using the improved data
structures, AltiVec-based processing routines, and the interactive
workflow of Mac OS X, tasks that might have taken half an hour in the
past can now be done in seconds.
“Because of increases in speed, we can use interfaces that get away
from the batch job approach that was required in the past when many
operations took overnight.
“This makes the processing more transparent to users—it’s like a
Photoshop session where you might do a series of steps on a photo and
save the results. At the same time, we use Cocoa’s strong XML support
to embed detailed history about acquisition and processing in the
output data files.”
The end result is that NeuroLens bridges a large workflow obstacle by
allowing a process that was often interrupted for minutes or sometimes
hours to be done continuously. This creates a more logical process for
researchers, and the performance means that users achieve a new level
of interactivity with the work, resulting in more detailed quality
assurance than was previously possible with such interrupted workflows.
With the new level of interactivity, there is greater ability to spot
problems with data that may occur at the time of acquisition (such as
scanner spiking or excessive subject movement), and this immediacy
allows users to understand the impact of various processing settings
much more intuitively than in the past.
“This has been a great teaching tool for us,” Rick explains, “as it
allows students to get a grasp of the purpose of processing steps and
how various parameters affect the end result. It is also very useful
not to have to spend several hours or more getting students up to speed
on how to use the UNIX command line. Mac OS X allowed us to create an
extremely user-friendly and intuitive interface.”
In addition, MRIs can produce both detailed structural and functional
information, by showing images of the brain as well as its activity
over time. NeuroLens allows researchers to overlay function on top of
structure in one image, as well as putting the data into graphs,
showing brain response over time. NeuroLens allows researchers to
manipulate the basic data set in various ways, using Mac OS X
functionality: extracting brain scan images for papers, producing
graphs from data analysis of response signals.
For more information on NeuroLens, see the NeuroLens Website at
http://www.neurolens.org.
...from:
http://cit.duke.edu/about/ipod_project.dohttp://cit.duke.edu/about/ipod_faculty_projects.dohttp://www.duke.edu/ipod/help/faq.html
In Fall 2004, in collaboration with Apple Computer, Inc., Duke is
distributing 20GB Apple iPod device to each of its first year students.
Duke hopes to stimulate creative uses of digital content by providing
the iPod devices as a mobile computing device. We hope that students
and faculty will use the device creatively and effectively to enhance
their intellectual life at Duke. This is a collaborative project that
includes Duke's Office of Information Technology, Division of Student
Affairs, Office of the Provost, and the Office of the Executive Vice
President.
[...]
Evaluation of Academic iPod Projects
One goal of the Duke First-Year Experience is to assess the potential
of the iPod and digital audio content in the teaching and learning
context. CIT is developing a framework for assessing how this
technology-intensive computing initiative has affected Duke faculty and
students. CIT will work with faculty to gather data to identify:
• successes and barriers in the integration of the iPod and digital
audio content in courses
• improvements and innovations in course design and delivery
• measurable changes in student learning and student outcomes
CIT supported projects will be assessed using a variety of evaluation
strategies. Each project will have an individual evaluation plan to
support assessment appropriate to the specific goals of the course and
project.
[...]
=======================
Courses using iPods as part of course work are:
Music, Sound, and Style
University Writing Program
Theory and Practice of Tonal Music I
Economic Principles
Social Minds: Memory as Collective Practice
Berlin in the 20th Century
Fundamentals of Digital Signal Processing
Perspectives on Information Science and Information Studies
Living Downstream: Ethics, Communities and Water Conservation
Intensive Elementary Spanish
Computational Methods in Engineering
