Interview


The way forward

In an exclusive interview with Callan Emery, Professor Osman Ratib, the president of the EuroPACS scientific society and one of the key initiators of the innovative and globally successful OsiriX open source PACS project, discusses PACS technology, the future of PACS and the OsiriX initiative.


Callan Emery: The digital technology behind PACS is now well advanced, with these systems being able to link up large networks of hospitals. What is the current state of PACS?

Osman Ratib: PACS has matured and almost become a commodity with different commercial systems being offered by small and large vendors on the market. With the rapid growth in imaging modalities and the overwhelming volume of digital data being generated, there is an urgent need for rapid deployment of PACS in all hospitals and medical institutions. While the majority of vendors have adopted the DICOM standard and provide solutions that can accommodate heterogeneous implementations of imaging devices from a variety of vendors, there is still quite a diversity in the type of systems and features that each PACS vendor provides.

Some efforts are emerging from the Integrated Health Enterprise (IHE) initiative in an attempt to harmonise the different technical solutions and unify the workflow models and system utilisation scenarios, but only a limited number of vendors have adopted it to date. Integration of PACS with hospital information systems and electronic patient systems are still challenging due to the lack of standardisation of these systems. It is often not a technical issue, but rather a design issue and implementation of specific clinical scenarios suitable in different clinical practices.

Most PACS systems are often ahead of the other systems such as electronic patient records and other hospital IT systems. Integration of such systems is a key for wider adoption and acceptance by clinical users and physicians outside radiology and imaging departments.

CE: Has the technology supporting PACS reached its peak? Is there still room for development?

OR: While PACS technology has matured, it is being challenged by constant changes in requirements. In particular, there has been a massive increase in the volume of images, particularly with multidetector CTs and high definition MR studies. It is not uncommon today to have standard clinical protocols generate over 2,000 images per study. Most common workstation software has some difficulties loading and managing such large data sets in an efficient and practical way. Another change is the emergence of multimodality and 3D imaging techniques. Combinations of PET scanners with CT or MRI imaging modalities as well as a combination of parametric functional and molecular imaging techniques require advanced and complex visualisation software programs capable of handling and combining very large data sets simultaneously. Furthermore, the need for advanced and high performance image management and analysis tools has expanded beyond the radiology department. An increasing number of therapeutic procedures and interventional techniques require adequate pre-surgical planning and image processing. Such tools that are traditionally needed in radiology are now becoming an essential part of the departments of surgery, cardiology, gastro-enterology and so on.

These new requirements force industry and systems designers to constantly adapt and modify the tools and architecture of PACS and image management systems.

CE: What other innovations are waiting in the wings? Do you foresee any major technological breakthroughs down the line?

OR: New “disruptive” technologies have emerged in the world of information technology – that is the unexpected emergence of technology that has changed the way people and systems communicate – starting with the wide and rapid adoption of web technologies, followed by a whole new concept of data streaming and multimedia navigation and more recently new communication patterns and ways of exchanging data through “social network” applications such as Facebook. These new technologies are slowly making their way into professional applications such as PACS and medical informatics.

CE: Do you foresee PACS having a broader application beyond radiological imaging?

OR: Definitely. It is absolutely certain that access to imaging will become part of everyday clinical practice in any ward or clinic. More importantly, there is an increasing need of special tools for specific clinical taskand treatment-planning activities. There is no doubt that surgeons will be relying more and more on multidimensional images to prepare and simulate complex interventions especially with a trend into minimally invasive procedures that require adequate pre-visualisation of the patients anatomy and structural anomalies.

Besides, an increasing number of medical decisions and therapeutic follow-ups are relying on imaging data. All cancer treatments now rely on complex image protocols for follow up of therapeutic response and treatment efficacy. Repeated PET-CT imaging is becoming mandatory in many cancer treatment protocols. These hybrid images need to be presented and discussed at tumour boards and multidisciplinary conferences. Traditional image viewers and displays are just not adequate for such presentation. Conference rooms and clinical wards are now being equipped with advanced image display and manipulation workstations allowing clinicians to manipulate complex 3D imaging data from PET, CT and MRI and superimposing the results in different projections to evaluate the changes of disease progression over time. Plus most clinical decisions rely on some quantitative measurements of size, intensity and distribution of complex parameters.

For all these reasons, advanced imaging workstations and image distribution infrastructure are needed in large medical institutions, given that clinicians cannot rely anymore solely on the radiologists reports and some printed images.

CE: What new research is taking place around PACS?

OR: The research in PACS has shifted toward what is referred to as “Imaging Informatics”. That is the development of new tools for integration and communication of imaging data in large and complex networks of users. It includes research in new technologies such as gridcomputing, wide-area storage, mobile and portable devices, and most importantly integration with new concepts of electronic patient records. It has largely extended beyond radiology into all other clinical domains and, in particular, into areas of surgery and imageguided interventions, such as robotics and endoscopic minimally-invasive surgery. This has led some authors to restate PACS as “Picture Archiving and Communication in Surgery”. 3D imaging and virtual reality techniques are also being adopted in medicine with numerous new applications in “augmented reality”.

CE: What key features should a hospital administrator of a single medium-sized hospital look for in a PACS system?

OR: The challenge of selecting a PACS is essentially in its specifications regarding the size and requirements of the institution using it. It would be very different in a small radiology practice versus a large academic institution. The most important feature is the diversity of usage and workflow of images not only in radiology and image provider services, but among all referring physicians and clinical wards. Large storage capacity and high performance network bandwidth are, of course, essential.

CE: These large PACS systems – often custom built – can be expensive? Does the open source OsiriX offer a good alternative?

OR: (Quoting from a paper published in the International Journal of Computer Assisted Radiology and Surgery 2008 [IJCAR]) Recent reports show that adoption of computerised medical records and medical informatics in medicine have significantly lagged behind expectations due to three major barriers: excessive cost, the transience of vendors, and the lack of common standards and adequate models. Many authors suggested however that open source software reduces these barriers by reducing ownership and development costs and facilitating the adaptation of customised tools for clinical practice. Open source offers freedom from software licensing costs with reduced-cost software upgrades and no license expiration.

The medical informatics community has welcomed the open source model, which fits naturally its scientific model of shared, peer-reviewed knowledge. Distributors of open source applications can share development costs with the user community and focus on implementation and support services. The resulting paradigm shift is that open source medical system vendors can become professional service providers, competing on service quality rather than on the basis of software secrets.

The impact of open source is even greater in specialised areas of medicine, such as medical imaging. These vertical markets have always been a challenge for vendors and manufacturers due to the small size of specialised users and high expectations in terms of complexity and performance of the tools required. This has naturally driven the market to high-end and high-cost developments and marketing strategies that also try to cope with very rapid evolution of computer technologies and software developments that make most products obsolete in a very short time interval, which does not allow the manufacturers to generate sustainable return on investment. In very specialised niches, such as in medical imaging, it is almost impossible to maintain profitable business models that are self-sustained. Most manufacturers will cover the cost through revenues from other business avenues, such as sales of imaging modalities, scanners and imaging devices, or by providing expensive global solutions and charging high costs for implementation and support and maintenance contracts of complex integrated information systems.

CE: Can you tell me about the OsiriX project and explain how this system works and how it is different to regular PACS? Why was it developed?

OR: (Quoting from a paper published in IJCARS in 2008) The OsiriX project started in 2004 as a collaborative effort between the University of California in Los Angeles (UCLA) and the University Hospital of Geneva (HUGE) with the goal to develop a multi-modality image viewing and navigation software on Apple Macintosh computers. The objective was to develop a completely new software platform that would allow users to efficiently and conveniently navigate through large sets of multidimensional data without the need of high-end expensive hardware or software. We – that is Antoine Rosset and Joris Heuberger, the developers of the software, and I – also elected to develop our system on existing open source software libraries allowing the system to be easily adaptable to a variety of hardware platforms. This project benefits from our past experience of developing one of the first free DICOM viewers several years ago called OSIRIS running on Macintosh and Windows personal computers. The new version was completely redesigned and developed from scratch to be more suitable for navigating through large multidimensional data sets while providing a convenient and easy-to-use environment for physicians that are usually not familiar with image processing tools.

The OsiriX software was specifically designed to take advantage of the significant progress in performance and flexibility of 3D animation on personal computers mostly driven by the computer graphics and game industry. Most video games today are developed on OpenGL graphic libraries that benefit from hardware acceleration and processing capabilities of today’s ultra-fast video cards. Also, OpenGL being an industry standard, it will automatically adapt to any hardware configuration and take advantage of any hardware accelerator that is provided for video display of two and three-dimensional data. In the scientific community, a new set of open source libraries has emerged for the visualisation of multidimensional data. Therefore we elected to build all our visualisation tools on Open GL standard. We also adopted open source libraries for image rendering and image manipulation.

The Visualization Toolkit or VTK is a well recognised and widely adopted software library that runs on multiple platforms and has been used for numerous scientific and medical applications so far. The recent adjunction of the Insight Toolkit or ITK, mostly funded by the US National Library of Medicine as part of the Visible Human Project, adds a wealth of additional rendering and image processing tools for medical applications. These open-source software toolkits offer powerful functions to do complex image manipulations and great performance for realtime 3D image visualisation. Our objective was to incorporate these powerful tools in a new graphic user interface (GUI) that is more suitable for clinical applications and image interpretation of large multidimensional datasets. We made special efforts to design a software platform that allows the easy and quick development of new generation multi-dimensional viewers that could replace many of the existing functions that are available only on highend expensive 3D workstations.

We also elected to explore, evaluate and integrate new emerging technologies from the consumer market industry to expand capabilities and convenience of use of PACS workstations beyond their current limitations. We selected a set of new products and services recently released by Apple Computer for their general computer products and we adapted them for medical imaging applications:

1) the iPod, a popular portable music player, was integrated to serve as a high capacity portable DICOM storage with a high-speed transfer rate.

2) iChat AV instant messaging and videoconferencing software was adapted to allow real-time radiology videoconferencing tool for remote image viewing and screen sharing.

3) iDisk, an Internet service provided by Apple for secure data storage on a virtual hard disk was adopted as a DICOM data storage and communication alternative.

By integrating such tools, OsiriX workstation can offer more features for communication between physicians. The software includes, for example, a feature that allows users to instantly send an email with an image (or image sets) from an open window in OsiriX. One can also initiate a live conference with any other OsiriX user on the network (or on the Internet) using the iChat application, allowing users to communicate live and share content of a window of OsiriX. This adaptation of a iChat provides a simple and convenient mechanism for teleradiology and remote consultations. Based on all the features that are implemented in OsiriX for data exchange and communication as well as the full set of DICOM services, it is easy to configure an OsiriX workstation as a DICOM compliant storage and communication node. That has lead many institutions who adopted OsiriX to install the software on a computer with large storage capacity and use it as a mini-PACS for storage of large image data sets. Our benchmarks showed that on a Macintosh server equipped with several Terabytes of Raid disks, OsiriX software can handle several millions of images accessible to tens of OsiriX workstations on a conventional local area network. The performance of retrieving and displaying images remotely from such a storage unit exceeds by far the performance of traditional commercial PACS architectures. Such a local archive node provides, for example, a convenient solution for temporary storage of large datasets that are generated on multi-detector CT scanners (several thousand images per acquisition) and cannot be stored on enterprise PACS due to their large seize.

CE: What type, size of medical facility is best suited to use this system?

OR: The most amazing observation is that OsiriX was adopted by thousands of institutions around the world ranging from small clinical offices to large academic hospitals. It is slowly becoming the most widely adopted imaging platform across all specialties.

CE: Is this system entirely free, besides the hardware required?

OR: The OsiriX software program and its source code are available under Open Source licensing agreement and can be downloaded free of charge at: (www.osirix-viewer.com). A special version that supports 64- bit memory addressing is available for US$149. Special companion OsiriX software that runs on iPhone is available at the App Store (the online store for applications for iPhone) for $19.99. Commercial (FDA certified) versions of OsiriX are also available from third party vendors that provide integration and support services. In December last year we created a foundation (http://foundation.osirixfoundation. com) that will continue to support open ource developments in medicine. One of the first projects of the foundation is to sponsor the certification of an FDA-certified version of OsiriX that will be distributed at much lower cost than existing versions. A free open source version will always be available as well.

CE: What are the advantages and drawbacks of this system?

OR: There are no drawbacks. One of the most attractive features of OsiriX remains its ability to manipulate and visualise large sets of image data using advanced volume rendering and three dimensional navigation tools. OsiriX user interface was designed to allow physicians to rapidly become familiar with the manipulation of 3D objects and navigating through large sets of images. All the traditional visualisation tools such as multiplanar reformatting, volume and surface rendering, transparency and intensity projects are available at a click of a mouse. Most importantly the software allows the physician to use those advanced navigation tools in real time on standard off-the-shelf hardware platforms. It also supports dynamic time/varying data sets that can be manipulated in 3D, while being displayed in cine motion essentially showing a 4D rendered image (where time is referred to as the fourth dimension). With the advent of multimodality imaging and molecular imaging devices such as hybrid PET/CT scanners, it is possible now to generate functional images that represent metabolic and biological dimensions superimposed over morphological and anatomical data. OsiriX was designed to conveniently handle the fusion of metabolic images and anatomical images in, essentially, a 5D image rendering mode (where the anatomical information is referred to as the fifth dimension). OsiriX is probably among the very few software platforms available today allowing users to display and manipulate 5D image sets in a convenient and user-friendly navigation environment.

Advanced image processing and analysis tools are being added to the program everyday. Developers from all around the world have contributed to the extension of OsiriX by adding innovative and specialised image processing features. Furthermore, the OsiriX software architecture allows for separate processing modules to be added to the program as plug-ins. Such plug-ins will be imbedded in the program when it is launched, but they don’t have to be integrated in the core of the main program. These external plug-ins could also be components that are not shared as Open Source software, but could be protected as binary modules or even sold as commercial extensions to the OsiriX platform.

CE: I imagine one of the main reasons for resistance against installing such a system is that there is the perception that there is no onsite training how to use it and no tech support. Is this the case with OsiriX?

OR: The concept of free open source software being used in professional environments such as medical facilities may be somewhat disturbing and some users (and especially IT services and management of these institutions) may feel that they need support, training and service. This is why several third party companies have emerged and made a business in providing service “packages” around the free software. While the majority of users are able to install the software and use it without any help, some institutions are willing to pay for professional support.

We also provide regular courses at major international conferences as well as special courses in different locations of the world for training users on advanced features of the software. We also published a book – OSIRIX the pocket guide – the second edition of which should be off press in April-May and is available in our online store.

CE: How successful has OsiriX been around the world?

OR: Currently, according to our latest surveys we have estimated that there are more than 40,000 active users that correspond with us on a regular basis around the world. This number does not account for users that have simply downloaded the software and are using it on their own workstation with no interaction with the rest of the user community.

Industry has also started to adopt OsiriX as a base for new business models where they provide the support and integration services as well as training and customisation of the generic platforms. Several certified versions for Europe and FDA-certified for the US have already appeared on the market.

And finally, and probably most importantly, the academic community has started to regroup its efforts to support and promote open source initiatives in medical imaging and medical informatics.

We have been lucky to surf on the amazing wave of success of Apple products in the recent years. The new Apple platforms offered us the best performance for advanced graphic processing. Apple has also given us a significant boost by featuring OsiriX on their home page several times and by publishing numerous success stories and video clips on how OsiriX is being used in real hospitals and clinics. Our iPhone version of OsiriX also received a lot of attention. We are very proud that it was selected by Apple for a TV ad at the 2009 Superbowl in the US in February.

Websites
● OsiriX viewer free download: www.osirix-viewer.com
● OsiriX Online Store: www.osirix-viewer.com/Store.html
● OsiriX third party vendors: www.osirix-viewer.com/Partners.html
● OsiriX and Apple: www.apple.com/science/medicine
Visualization Toolkit or VTK: http://public.kitware.com/VTK
Insight Toolkit or ITK: http://itk.org
 

Patient photos spur radiologist empathy and eye for detail

Including a patient's photo with imaging exam results may enable a more meticulous reading from the radiologist interpreting the images, as well as a more personal and empathetic approach, according to researchers in the USA.

Many radiologists have limited contact with patients. A referring physician will order imaging exams, such as MRI or computed tomography (CT), and the radiologist interprets the results, never having met the patient.

“Our study emphasises approaching the patient as a human being and not as an anonymous case study,” said lead author Yehonatan N. Turner, MD, radiology resident at Shaare Zedek Medical Center in Jerusalem, Israel.

Technological advances have further distanced the radiologist from interaction with the patient. With the advent of teleradiology, radiologists are now able to view images from remote locations via the Internet or satellite.

“We feel it is important to counteract the anonymity that is common in radiologic exams, especially with the growth of teleradiology,” Dr Turner said. The researchers set out to determine if the addition of a patient’s photograph to the file would affect how radiologists interpreted the results.

For the study, 318 patients referred for CT agreed to be photographed prior to the exam. The images of the patients were added to their files in the hospital's picture archiving and communication system (PACS), a network for storage and retrieval of medical images. The photograph appeared automatically when a patient’s file was opened.

After interpreting the results of the exams, 15 radiologists were given questionnaires to gather data about their experience. All 15 radiologists admitted feeling more empathy towards the patients after viewing their photos. In addition, the photographs revealed medical information such as suffering or physical signs of disease.

More importantly, the results showed that radiologists provided a more meticulous reading of medical image results when a photo of the patient accompanied the file. Incidental findings are unexpected abnormalities found on an image that may have health implications beyond the scope of the original exam. In order to assess the effect of the photographs on interpretation, 81 examinations with incidental findings were shown in a blinded fashion to the same radiologists three months later but without the photos. Approximately 80% of the radiologic incidental findings reported originally were not reported when the photograph was omitted from the file.

The radiologists involved in the study commented that while the addition of the photo did not lengthen the time spent reading, it was a factor in how meticulously they interpreted the images. All 15 radiologists agreed that the inclusion of a photograph in a patient’s file should be adopted into routine practice. The photos can also be included in long-distance teleradiology practices.

“The photos were very helpful both in terms of improving diagnosis and the physicians’ own feelings as caregivers,” Dr Turner said. “Down the road, we would like to see photos added to all radiology case files.”

 Date of upload: 16th May 2009