WWDC 2007

Positron emission tomography provides images of metabolic function and is a powerful tool for both clinical management of patient and biomedical research. To support pre-clinical research to both understand basic disease processes and develop therapies, investigators are increasingly using metabolic imaging in small animals. Scanners with sufficient spatial resolution and sensitivity to support the needs of medical and biological researchers offer many significant challenges to the physicists, engineers, and mathematicians developing such systems. In our own group, a series of scanners are being developed that provide systems for both positron emission tomography (PET) and combined PET and Magnetic Resonance Imaging (MRI). In each case, we are using a data acquisition built around 1394a and 1394b Firewire (using in-house developed acquisition electronics that can function within a high magnetic field). A key component of that system is the Macintosh host application that communicates with each Firewire node and acquires the data in a series of list mode files. The ability to implement the full acquisition system in user space was critical in our design – minimizing maintenance issues and development time. Further, the ability to use several libraries of image reconstruction tools originally developed under other UNIX systems has allowed us to implement advanced image reconstruction algorithms on the host Macintosh computers with a minimum of effort. The application is written in Cocoa and C++. This presentation is an update on the Firewire system described in a poster at the 2006 WWDC and provides details on the design and implementation of the Macintosh host application (MiCES), including the steps taken to assure correct handling of the raw data on both PPC and Intel based Macintosh systems. The system has provided us with the needed flexibility and performance.

The Cocoa frameworks in Mac OS X empower individuals and small-teams to develop great software at a rapid pace. We have utilized Cocoa to build specialized custom tools for analyzing chemical reaction dynamics that have significantly boosted our research efforts over the course of the last few years. Developing custom tools for internal use is much less demanding than developing software for release to the wider community, and can result in tools better suited to the task than generic, off-the-shelf products. This poster discusses our approach to custom tools in general, and introduces two of the applications that have been developed: a 4D multitrack visualization application ('Dynasity'); and a tool for monitoring progress of long-running calculations running remotely on clusters and supercomputers  ('Remote Activity'). The scientific successes that we have realized with these tools are also summarized.

The Semantic Time Framework (STF) is a software library for Mac OS X for designing and constructing interactive systems that modify the timebase of multimedia. Current multimedia processing frameworks usually support one media type (e.g., Core Audio for audio or Core Video for video), but not both. Multimedia frameworks that support multiple media types, such as QuickTime, focus on media presentation, and it is difficult to add custom processing or otherwise manipulate the media data itself. STF supports a more general timing model based on the semantics of the media data. Just as we often refer to distance using "blocks on a street", such as when giving directions, it is usually more meaningful to refer to time using "beats of music" or "words of speech" rather than "number of audio samples" or "number of video frames". Media data in STF is treated as a continuous flow of data, rather than discretized packets or events. The architecture is based on a system of pipes and filters, with additional constructs for supporting time and synchronization. Each media type has its own pipe, and the rate at which media is played back can be interactively adjusted, much like a valve controls the rate of fluid flow through a pipe. Pipes can also be linked to provide synchronous playback, even if the media come from separate sources. Similarly, media pipes can be linked to external timing sources for interactive control over play position and rate. STF builds on a number of Mac OS X technologies, including Core Audio, Core Image, Core Video, and QuickTime. The library is written in Objective-C, which makes it simple to include in Cocoa-based applications. STF has been successfully deployed in a number of audio editing and interactive orchestral conducting applications. It is an open source project (http://styme.org).

EPICS is a popular, open source controls systems software infrastructure and is currently used at the Spallation Neutron Source (SNS) at Oak Ridge National Lab. While EPICS includes support for the Macintosh through Darwin, it lacks direct Cocoa support. An EPICS framework was developed which wraps the existing EPICS C API into an Objective-C API allowing developers object-oriented access to EPICS client functionality. This framework allows developers to rapidly develop Cocoa applications with EPICS client support. A student built an alarm annunciator for SNS using this framework and leveraged Apple technologies such as speech synthesis and Quartz Composer. Looking forward, we intend to use this framework to retrieve from EPICS live images of the beam on target and convert it to a QuickTime stream for users to view over the web.

"Embedded Compiler Research Using Mac OS X In the field of computer science, compiler and programming language research focuses on the efficient translation of human-readable directives into machine-digestible instructions. With today’s demand for small-scale computing devices, new challenges in compiler design are being presented. In the embedded environment, limits are imposed on memory and power resources while optimum performance continues to be demanded. Furthermore, the vast range of permutations in programming languages and target processors for system designers is a significant factor. RELIC (Retargetable Embedded Language Independent Compiler) is a compiler architecture designed to explore these emerging research topics. With distinct analytical compilation phases, RELIC utilizes mature optimization techniques to achieve size and performance goals in object code generation for microprocessors constrained by tight memory and power requirements. To bridge the chasm between the myriad of high-level computing languages and embedded microcontrollers available, RELIC utilizes a machine-neutral and language-neutral instruction set called RELAX (Relational Expression Language for Academic eXperimentation). As a vital part of the RELIC architecture, RELAX brings a number of added benefits to the compiler tool chain, including focused optimization, code portability and language unification. The design, implementation and testing of the RELIC architecture has been greatly facilitated by leveraging the Mac OS X operating system’s rich BSD heritage and robust Xcode development environment. As the project expands, Apple's Mac OS X continues to provide RELIC with a solid foundation for development, cross-compilation and testing of new languages and processor targets."

"The PROOF system is designed for interactive analysis of Tera Byte data sets in parallel on a cluster or multi-core machine. PROOF represents a high-performance alternative to a traditional batch-oriented distributed computing system. PROOF is being developed at CERN for Particle Physics data analysis but is now also being used in other areas where large amounts of data have to be mined, like in finance. At CERN we are running PROOF on Xserve/XRaid and Linux clusters. Running PROOF on an 8-core MacPro shows perfect linear speedup as function of the number of active cores. On clusters up to several hundred CPU's can be used in parallel with high efficiency (95%). The PROOF code is freely available under the LGPL license."

"For many years, multiprocessor PCs were uncommon, supported only by a few applications, and only purchased by a few users; servers could take advantage of multiple processors by running several different services at once, while graphics professionals might purchase multithreaded applications such as Photoshop to speed their computations as much as possible. As processor manufacturers continue to increase performance, there are diminishing returns to investing transistors into speeding up single-threaded performance. Instead, most processor manufacturers are now investing significant research and development into multithreaded execution using multiple processors, multiple cores, or a single multithreaded core. This project explored the real-world effects that this move to multithreading has. To measure performance, a comparison was done between different machines and configurations with a parallel sorting algorithm. The sorting algorithms implemented for this project are shear sort, parallel radix sort, partitioned radix sort, and single-threaded Quicksort. The program was characterized using staistics counters within the program and with Shark, giving information on time profiles, memory bandwidth, and scheduling behaviors. This data was used to find where the bottlenecks lie for the implementations of these algorithms, such as in the memory system, or the scaling of multiple threads."

"The main goal of this project was to design and build a visual language as a tool to define all kind of queries to a scientific database, which has stored microbiological information about CINVESTAV's Collection of microorganisms. This collection is part of the CONABIO Organization. MVLAB-LIDA as a system, includes other important functions about microbial cultives, such as statistical analysis, image processing and video visualization. With the system implemented, the wish for giving the remote research services to the microbiology area is more realistic now. This access would allow the researchers and institutes around the world to have immediate access to huge quantities of information about microorganisms in the Microbiology Laboratory of CINVESTAV. One of the main goals of this project was oriented by the strong need of the Microbiology Laboratory, with the objective to diffuse the information available in its Data Base using the Internet as platform. However, it is not enough to just give remote access to the information, but it is preferable to give services which make the information analysis easier. There is a set of functions that will be included in the system for performing several data analysis such as the statistical or qualitative analysis. Also the system will include features for performing operations of digital images or videos. Another of the requirements for the project is to defeat the geographical boundaries, because nowadays there is not a system in any country that produces such a low transportation cost for accessing the information. This is the reason why the project is considered such an adequate solution to the problem. Technically, the solution proposed, is the design and building of a remote information system which will have capabilities to work under web environments and is based on the Client-Server Model. In addition to this, a visual language for defining and executing queries is included. This allows the classifying of statistical data and processing of images and videos. The system allow to access a Data Base which is segmented in three parts: one of them corresponds to images, another to data and the last, corresponds to videos. The development was done using a xserve dual G5 as main server and iMac as clients. We use postgresql as our DBMS and Java, XML, XSLT, JavaScript, JDOM as our tools for this develpment."

"To achieve accessible computational power for our research goals, we developed, on the Macintosh platform, the tools to build easy-to-use numerically-intensive parallel computing clusters. We find that the usability and reliability of the Mac cluster technology is as important as its performance. Our approach is designed to allow the user, without expertise in the operating system, to most efficiently develop and run parallel code, enabling the most effective advancement of scientific research. By ""reinventing"" the cluster computer, we provide a unique solution designed to maximize accessibility for users. To support numerically-intensive and tightly-coupled problems that demand the computational power and networking capabilities of clusters of Macintoshes, our software technology supports five implementations of the Message-Passing Interface (MPI), today a dominant industry standard. MPI's status implies how its underlying computing paradigm has revealed itself to be the most efficient and economical way yet found to apply large numbers of processing ""cores"" effectively for general purposes. We present two new applications of our approach to clustering: 1. The new Supercomputing Engine for Mathematica enables Wolfram Research's Mathematica to be combined with the programming paradigm of today's supercomputers. In contrast to typical master-slave ""grid""s, this solution instead closely follows MPI, from inside the Mathematica environment, and has every kernel in the cluster communicate with each other directly and collectively, necessary to address the largest problems in scientific computing. 2. Modern compression/decompression (codec) algorithms for video processing increasingly have compression times that vastly exceed playback times. We address this emerging computational demand using MPI-based cluster computing. We implement load-balancing parallelization of QuickTime video compression, including frame-reordering H.264, and interface our cluster support software with mainstream desktop video-editing applications such as Final Cut Pro. "

"Papers is the latest program from the duo Mek & Tosj who previously created award winning scientific programs like EnzymeX and 4Peaks. Using the strength of the Mac OS X platform Papers allows scientists to create their own personal library of science. While thus far most programs focused primarily on the scientific writing and citation process, Papers provides researchers with a complete workflow for finding, retrieving, reading and organizing scientific articles. For this task it leverages and integrates a spectrum of different Mac OS X technologies, fused together through a user friendly and beautiful graphical user interface. We will discuss how different Mac OS X technologies helped us create this revolutionary new program, and we will outline future directions. In addition, we will show how Papers with the help of fellow Cocoa developers can be extended to address an even larger scientific audience. URL: http://www.mekentosj.com/papers"

"We present a modeling and simulation software tool - BeemmLink - that provides the dynamic connection and transparent two-way data exchange between geometric modeling/visualization and mathematical computations modules. BeemmLink uses Wolfram Mathematica computation engine as dynamics solver and Autodesk Maya as geometry modeler, visualizer, and simulation driver. User interface of BeemmLink is a set of Maya plug-ins, while Mathematica MathKernel runs on the background. Such implementation does not limit the native capabilities of the software tools it is linking, and allows the user to take full advantage of the unabridged development functionality existing within each of the linked software packages. BeemmLink is a set of Carbon frameworks and bundles. In addition to serving as Maya-Mathematica runtime connection, it provides Carbon environment for implementation in C/C++ of the custom algorithms that control the simulation parameters. presented example demonstrates the described above functionality: equations of motion are solved in Mathematica, geometry objects are set in Maya, and simulation is run from Maya workspace window. In order to verify the correct operation of Mathematica-BeemmLink assembly, a stand-alone Cocoa application is included in BeemmLink package."

"The ultimate goal is to understand complex biological processes, and a computational model must be built to study the behaviors of the biological systems, and therefore gaining a quantitative understanding of its functionality. However, to model a realistic biological system requires a large amount of agents to interact, resulting in large calculations regarding their movements, interactions with other agents, and their resulting state behavior that changes in response to their environment. As simulations grow in size and complexity, this will eventually exceed the processing power of a single computer, but will make an ideal candidate for using massively parallel computation. The parallel architecture consists of using Xgrid, which is a distributed computing technology built into Mac OS X that makes it easy to create and organize ad hoc group of Mac systems into a low-cost supercomputer. The three-tier architecture involves a client, controller and agents. The client is the computer that submits work to be done, the controller is a different computer that manages jobs, scheduling, data movement and agents. An agent is comprised of one or more dedicated computers that perform the task that the client requires. "

"Three-dimensional (3D) medical images, such as those acquired through Magnetic Resonance imaging (MRI) and Computed Tomography (CT) scans, Are important tools to aid the diagnosis and treatment of disease. Real-time 3D Visualization tools can greatly improve the speed and accuracy of diagnosis. Unfortunately, the majority of current tools are only available in dedicated 3D medical workstations. The prohibitive cost of these workstations has greatly limited the use of real-time visualization in day-to-day medical practice. The exponential growth of computing power in consumer workstations, especially in graphics processing units (GPUs), has opened the possibility of advanced 3D visualization tools on inexpensive consumer hardware. While recent GPU visualization tools have accomplished real-time frame rates, they often require major algorithmic changes to run on specialized GPU hardware, often resulting in inferior quality to offline solutions and custom visualization workstations. Our objective was to create an interactive tool for visualization of 3D medical images on current Mac OS X computers, without sacrificing the speed, quality, or functionality of state-of-the art dedicated 3D medical workstations. We use the advanced OpenGL 2.0 and GLSL shader language support in Mac OS X to develop state-of-the art ray-casting techniques that have previously been limited to offline CPU processors. By implementing and testing a number of ray-casting optimizations and improvements using Apple’s OpenGL Profiler and OpenGL Driver Monitor we have developed a 3D volume rendering engine that achieves interactive frame-rates, with increased quality when compared to offline CPU ray-casting algorithms. We hope that in providing more ubiquitous access to 3D medical visualization that they will be used more often in day-to-day practice and help to improve the speed and accuracy of diagnosis."

"Common Astronomy Software Applications are a suite of applications for the reduction and analysis of radio-astronomical data with a Python (IPython) interface. It's a big application for Linux and Macintosh scientific workstations. World-class radio-telescopes currently under construction with challenge the data-processing capabilities of high-end workstations. Multiple-core optimization will be required. Macintosh computers provide a well-integrated Unix environment in a high-performance desktop. Their laptops ain't too shabby, neither. Porting to the Mac was driven by user demand, and is supported by a small development team: three programmers part-time for Mac-specific support. CASA was initially developed for Solaris, and then Linux workstations. We ported this system to the Macintosh platform by using MacPorts packages to provide ""Linux-compatible"" user-space interfaces. Extensive modification of MacPorts enabled the Intel-based Macintoshes to surpass the performance of similarly-configured, Linux-based workstations. Most packages were also tuned for the Intel C++ and FORTRAN compilers, with good results. We believe these MacPorts modifications are of general, significant value to the Macintosh community, and we want to contribute them to the MacPorts project. Our astronomy-specific code is largely is C++, with a custom binding to the IPython/SciPy environment for the primary end-user interface. Some graphical user-interface tools are developed with TrollTech's Qt libraries. We have migrated some tools to ""native"" Macintosh, where the Qt implementation was unsatisfactory. CASA builds upon the libraries developed by the AIPS++ consortium, and is currently under development NRAO, the National Astronomical Observatory of Japan, the Dominion Radio Astrophysical Observatory of the National Reasarch Council Canada, and Univeristy of Calgary for the Atacama Large Millimeter Array (ALMA) and the Enhanced Very Large Array (EVLA) observatories."

"iBabel is an Applescript Studio application that provides a front-end for a variety of Cheminformatics tools, in particular OpenBabel. To date these include file conversion (between a vast range of chemical file formats), SMARTS-based substructure searching, similarity searching, list manipulation and superimposition of molecules all using OpenBabel. iBabel also uses several embedded java applets for molecule viewing, a 2D viewer using JChempaint, a 3D molecule viewer using Jmol, binaries for which are included in the iBabel application. As an alternative Marvin or ChemDraw can be used for both 2D and 3D display. Editing structures uses JChempaint or JME. In addition, iBabel provides a front-end for Spotlight searching for files for chemical information after the files have been indexed using ChemSpotlight. ChemSpotlight is a Spotlight metadata importer plugin for Mac OS X 10.4, which reads common chemical file formats(.mol, .sd, .sdf, .mol2, .pdb, .cml, and XYZ) using the Open Babel chemistry library, it then adds molecular formulas, molecular weight, and a variety of other information for Spotlight searches. iBabel demonstrates the integration of a variety of programming languages and tools including, applescript, PERL, shell scripts, Java, C++ together with in built Mac OS X utilities."

"ProteinPsi is a project that integrates a few of the old legacy UNIX C/C++ applications that our lab has collected over the years of protein folding analysis. One of our algorithms uses a 19th century mathematician's postulate about the geometric interface of a protein to characterize interfacing amino acids. Another UNIX program called Mustang was incorporated because it has a multiple structural alignment algorithm that uses a three dimensional distance matrix to compare proteins of similar sequence. Our program is also a Core Data based application allowing a simple and easy switch in our object model to incorporate new ideas as they arise. We are also using the Cocoa framework and the powerful OpenGL framework giving us the capability of our own Open GL display of multiple different views of a protein. Our goal is not to rewrite the many protein viewing tools available, but to try and restructure our current understanding of proteins by building our own application to view their dynamic interactions. These multiple vertex points live in a complex dynamic three dimensional matrix that is NP Hard to build, so the multitasking of NSTask job assignments help to distribute our intensive load nicely. We also used intensive Perl scripts to do a lot of our text manipulation outside the Core Data network. The transition will allow our laboratory to design better protein analysis algorithms without as many struggles. Our incorporation of the Mac OSX foundation makes it a breeze to create a solid application to store our protein files and resurrect our ancient legacy UNIX code from the depths of our code vault in order to really burn the new quad Xeon's."

"We present a reusable software package able to control a large and heterogeneous network of remote-controlled and robotic astronomical observatories. Special attention was given to the design of a robust, long-term observation scheduler that also allows the trading of observing time and resources within internal and external networks. The system is able to communicate with notification systems announcing X-ray/Gama-ray bursts events, and other astronomical events requiring immediate observation. The handling of the ""Phase I & II"" project-development process, the time-accounting between complex organizational structures, and usability issues for making the package accessible not only to professional astronomers, but also to amateurs and high-school students is discussed. Initially developed to run a network of two 1.2m telescopes, and a 50cm solar telescope, the system was developed using Cocoa and WebObjects, and - with the exception of a few low-level systems controlling various peripheral devices - is completely deployed on Apple hardware. "

"Since the 1859 publication of Darwin's Origin of species, the construction of a universal tree of life has been a major quest in biology. Thanks to DNA sequencing techniques, our understanding of the evolutionary history of life on earth has improved dramatically during the last two decades. Through the comparison of DNA segments, some branches of the Tree of Life are now, at least partly, resolved. The exponential increase of molecular data also triggered the development of new, sophisticated, and powerful analysis techniques, such as Bayesian inference of phylogeny and relaxed clock methods. New methods are adopted by the scientific community only when they are implemented in stable, easy-to-use applications. Today, plenty of high-qualtity applications are within reach of any evolutionary biologist and, unsurprisingly, almost all of them run on a Mac OS X platform. Here we present an evolutionary study in which Mac OS X applications and technologies play a major role. We analysed data of three genes to reconstruct the evolutionary history of Burmanniaceae, a small group of extremely rare tropical plants with very scattered populations. Sequence assembly, alignment, and phylogenetic analyses including complex divergence time estimations and ancestral area reconstructions were all done on a Mac with software that is available for free. Our results indicate that this plant family dates back to the late Cretaceous and originates from South America. During the Eocene a major radiation occurred resulting in the current pantropical distribution of Burmanniaceae. This study clearly shows that dating ancient lineages of organisms can be carried out by combining existing Mac applications, in a reasonable timeframe and without the need for supercomputers."

"The EyA system has been developed to digitally archive scientific lectures with the aim of widening their audience and of offering a high quality learning experience to the remote audience. It is a completely automated, non-intrusive system which allows the recording of any type of lecture (seminar, talk, presentation) without human intervention. Scientific presentations are more complex in form than PowerPoint or Keynote presentations: they can include the simultaneous use of a chalkboard, transparencies and overhead projector, the display of simulations via animations, the display of movies and photos from experiments, etc. The EyA system is completely based on Apple technologies. Video is recorded on a Mini Mac (so called “producer”) with an iSight and audio is recorded with a USB microphone. Photos are taken every 15 seconds with a digital camera controlled using AppleScript. Continuous hourly recordings are carried out during the day in slots of 1 hour. All photos, together with the QuickTime movie are transferred to a dedicated Mac server (the “master”). This is done immediately after every hour of recording, and may happen at the same time while the computer is recording the next hour. The “master” processes all files and creates a QuickTime synchronization track that, added to the movie file, provides the synchronization between the images and the movie. Images are also compared together to drop duplicates in order to decrease the space needed for the storage of the recordings. At the end of the process the images, movie file and synchronization track are combined together to build a web page. A javascript zoom facility allows the reading of tiny details in the pictures. The final result is published on web and also compressed into a zip file (with average dimension of 130MB per recorded hour) for CD/DVD production."

"Co-registration is an iterative procedure that maps a source image onto a target image. Automatic co-registration of 3D datasets is an important investigative and diagnostic tool in biomedical science. It is commonly used to align patients’ anatomy between imaging scans of similar or different modalities. However, enormous computational costs and difficulties in tuning and guidance have precluded widespread use of sophisticated 3D co-registration in many clinical and research workflows. We have developed a novel application that leverages recent advances in the power, programmability, and high memory bandwidth of commodity graphics processing units (GPUs) to greatly accelerate 3D rigid-body and deformable coregistration. We implement the computationally intensive tasks of image similarity metric evaluation, rigid/deformable transformation, and image resampling to execute on the massively parallel and optimized GPU structure. Our application performs 3D, deformable, multi-modal co-registration at rates up to 200 times the speed of other readily used applications. Our results match or surpass the accuracy of the other techniques. Our application is also modular and flexible, allowing ready adaptation to a variety of registration tasks. Furthermore, it provides visual feedback during the registration process. This allows inspection and interaction, and greatly simplifies the fine-tuning often required for successful co-registration. The core of our application relies heavily on the advanced OpenGL 2.0 technologies in Mac OS X. Our application is a universal binary that runs under Mac OS X and works well on widely available, standard Apple desktop and laptop computers. Together, these advantages may expand the utility and application of sophisticated co-registration of biomedical datasets."

"Simulation is a key component of many medical training programs, providing practical experience in safe, repeatable virtual environments which , in real life would be expensive and/or dangerous, and presenting rarely encountered but important clinical cases. Computer-based surgical simulation gives training in basic skills and surgical tasks for medical students, residents, and professionals. Simulations are often enhanced by realistic tool interfaces and force feedback (haptics). Benefits include more training opportunities, improved performance and learning through repetition, reduced cost, and less risk to animals and human patients. The SUMMIT group of Stanford’s University School of Medicine (http://summit.stanford.edu), in collaboration with the National Biocomputation Center (http://biocomp.stanford.edu), recently launched an OS X version of the SPRING surgical simulation platform. SUMMIT creates and researches innovative information medical training technologies. SPRING is a standards-based, networked, open source, multi-platform framework for 3-D surgery simulation. It combines deformable soft tissue modeling, OpenGL-based graphics, and force feedback in an accessible, extensible package. Interfaces devices are connected via TCP/IP sockets, and SPRING’s multi-threaded physics computation takes advantage of high performance, multi-CPU systems. Xcode 2.X greatly facilitated implementation and testing on OS X. Standards including OpenGL, GLUT, Posix threads, networking, and UNIX system calls smoothed the port of this C++ application, and a makefile-based build was quickly adapted. Easy integration with Objective-C modules maintains cross-platform compatibility while incrementally incorporating Cocoa and other OS X features. Shark and other developer tools instrument and diagnose performance-critical aspects of this resource-intensive application. This poster will discuss SPRING’s architecture and capabilities, and will highlight how OS X tools and standards simplify continued development of SPRING’s capabilities. SPRING is freely available for non-commercial use, and users are encouraged to use, extend, and improve the package. To learn more, see http://spring.stanford.edu/."

"It has been shown that the textural appearance of tissues on medical images can be indicative of certain types of pathology. Texture can be thought of as the local characteristic pattern of image intensity; regions of an image consisting of rapid fluctuations contain high spatial frequencies while slowly varying regions contain primarily low spatial frequencies. Multi-resolution analysis can provide a description of the spatial frequency content, or texture, of image regions. Continuous space-frequency techniques, such as the short-time Fourier transform or continuous wavelet transform, calculate the frequency content at every frequency and every point in the signal. In the two-dimensional case, these calculations become too complex for practical use. The discrete wavelet transform provides a fast and efficient framework for multi-resolution analysis of large images. However, optimal results require prior knowledge of the appropriate mother wavelet and conversion from dyadic scales to frequency. We present a two-dimensional extension of a recently published discrete, orthogonal S-transform. The discrete transform is a fast and efficient multi-resolution extension of the Fourier transform. This new transform allows for arbitrary sampling of space and frequency as well as inversion directly to the Fourier frequency domain. We used Python 2.4 on a PowerBook G4 running Mac OS 10.4.8 to calculate the discrete S-transform of a 256x256 image in 1/10th of a second. Texture images were displayed using the open-source matplotlib package. We plan to use this technique to create local texture maps of the brain, potentially allowing earlier identification of diffuse disease and improved monitoring of response to therapy. The speed and efficiency of the discrete S-transform will allow us for the first time to perform texture analysis on large imaging volumes. We plan to use the texture information derived from as an input to a machine learning algorithm to identify and segment pathological tissues in MR images."

"Microscopic image management and image analysis are extremely important aspects of biological research. While many high-quality Mac OS X solutions exist for medical images (e.g. Osirix), scientists have much fewer choices when it comes to image management and analysis of light and especially electron microscopic images. Most available applications are cross-platform and therefore not optimized for Mac OS X, both in terms of features and performance. To address this need, we are developing Macnification. The application will enable scientists to organise, annotate, manage, analyze and publish microscopic images. Highlights include automatic calibration of SEM images, an easy-to-use image browser, real-time, non-destructive filters and image enhancement, streamlined image capturing (supporting most FiWi and USB cameras), one-click creation of publication-ready image plates, ... We intend to make the application Leopard-only, enabling us to take advantage of new features, such as ImageKit, Core Image, Time Machine, Core Animation, QTKit and the iChat Theatre API, among others. The first release of the product is due later this year and will focus on light and electron microscopy. With this abstract and poster, we are soliciting feedback and suggestions from the scientific community. Macnification is co-developed in collaboration with the University of Leuven, Belgium."

"We have developed a real-time multi-spectral confocal imaging system (MSCIS) providing surgical teams instant optical biopsies for early cancer detection via a laparoscope or daughter endoscope. Optical biopsy delivers instant in-vivo cellular images, comparable to those provided by histology, through a minimally invasive procedure. For the surgeon, this means that more tissue can be explored and abnormalities can potentially be dealt with during the same procedure with less harm to the patient. Currently, surgeons must resect abnormal tissue and wait for frozen section processing and pathology analysis to determine if the tissue is abnormal. Interchangeable catheters provide both flexible endoscopic and rigid laparoscopic abilities for compatibility in a variety of common procedures. The confocal nature of the system optically sections the tissue and enables real-time display and collection of high-resolution grayscale video of tissue. The MSCIS’s ability to resolve at the cellular level can help diagnose the subtle morphological changes that take place during the early stages of cancer development. The MSCIS can also collect high-resolution multi-spectral images. These images have enhanced contrast and provide additional information about the tissue including targeted contrast agent concentrations, pH levels, and ion concentrations. Using apple hardware and Mac OS X technologies, we have developed a mobile system that is reliable and easy to use. Leveraging the unix foundation of OS X we have been able to integrate scientific libraries and existing code with Coca, OpenGL, Python, and PyObjC to control our hardware and develop a user interface suitable for the surgical suite. Initial applications of the MSCIS include diagnosing diseases of the gastro-intestinal tract and the female reproductive systems. For women at high risk of ovarian cancer, the MSCIS is currently in a small scale human trial to screen for the early onset of this disease."