Traditionally, radiology is a support department that provides imaging services to other hospital departments. In this conventional framework, the primary concerns of a radiology department were how to enhance the productivity of imaging workflows. Most efforts have been made principally to remove unnecessary communications and thereby reduce report turnaround time. The introduction of information systems such as PACS (Picture Archiving and Communication System) and RIS (Radiology Information System) are typical examples of such efforts.
Over the past decades, imaging technologies have advanced remarkably, and have led to the proliferation of digital imaging services. Many imaging solution providers are offering various off-the-shelf software programs at more affordable prices. Those programs are equipped with sophisticated imaging functions, and can easily manipulate the large amounts of image data generated from high-performance imaging modalities. As a result, the number of imaging centers providing diagnostic imaging services has grown considerably, and competition between them has intensified.
In this evolving environment, enhanced productivity of imaging workflow is not sufficient to guarantee a competitive and successful imaging business. Rather, more diversified perspectives of customer satisfaction must be considered, and technological advancements must be leveraged for the quality and the competitiveness as well as the productivity of imaging services.
In this article, we envision digital imaging services in radiology, with emphasis on the recent advancements in digital imaging technology and its future direction. Specifically, we focus on the four major issues prevailing in current imaging business practices: specialization, flexibility, reliability, and usability. We investigate the kinds of technologies pertaining to each issue, as well as the ways in which such technologies have enabled the invention of innovative services in diagnostic imaging practice.
Specialized Service: Protocol-Based Imaging
Traditionally, imaging software programs have been sold as expensive components of a new modality. The principal role of such programs has been to perform volumetric reconstruction of slice images, and thereby to provide multi-planar reformation (MPR) images, which are cross-sectional images of the volumetric data. These generic functions are usually not intended for specific diagnoses, and therefore radiologists need to be skillful in using those functions to obtain the desired imaging results.
Now, imaging software programs are becoming more specialized and thus more diversified. Most recent imaging programs can facilitate a protocol-based procedure that is optimized for a specific diagnostic purpose. Radiologists are guided in a step-by-step manner through the entire procedure until they complete their job. They have only to select a study type, and then various tools automatically manipulate 3D images to generate result images based on the selected study type. In this way, many labor-intensive tasks such as segmentation, contrast enhancement, and navigation can be automated, so that the user’s unnecessary involvement is significantly reduced.
Virtual CT-Colonoscopy (CTC) is a good example. It has been proposed as an alternative to the current gold standard, optical colonoscopy. In conventional CTC, radiologists have to navigate manually through the entire inner colon to find polyps, a very time-consuming task. Tools have recently been devised to automate many steps of the CTC procedure, such as colon segmentation, navigation-path detection, and automatic navigation through the path. Those tools could even replace the interpretations of radiologists with faster, more accurate auto-detection of polyps, as currently published methods have reported encouraging results.6
Besides CTC, various other diagnostic protocols can be facilitated by imaging software. With vessel analysis tools, a radiologist can quickly perform flexible and accurate quantitative analysis of vascular anatomy, and can easily generate, based on the analytical results, various types of context-sensitive reports. Calcium scoring tools for diagnosing heart diseases facilitate fast generation of a diagnosis report through automated calculation of calcification in the coronary arteries. Users can choose a preferred calculation method from among the many previously designed methods.
Enhancing Flexibility: Customizable Report Management
Even though supporting a protocol-specific working procedure is an important issue with regard to achieving the desired performance improvement, too much focus on specialization is likely to lead to somewhat rigid and inflexible working environments. In fact, when considering the provision of diagnostic services to local clinicians and/or outpatients, imaging centers have to consider the diversity of their customers’ requirements as well as more responsive means of satisfying them.
One of the most typical problems regarding such flexibility in diagnostic imaging is related to the issue of supporting various hanging protocols. Hanging protocols (HPs) were originally developed to control the initial display of images in order to more efficiently view films. They determine what types of images to be included in a report and in what order they are displayed. In general, hanging protocols vary according to the diagnostic protocols. For example, types of images and their layout for the orthodontics protocol, which is a typical diagnostic protocol in dental imaging, are different from those for the implant protocol or for the standard MPR protocol. In addition, because imaging centers and referring clinics usually have their own preferred hanging protocols, many different kinds of reports need to be produced to meet their needs.
Although the hanging protocol has a significant role in enhancing imaging workflows, even recent implementations of PACS fail to describe scrolling interaction, the underlying layout algorithm, and complex screen layouts that cannot be represented in rows and columns8. XML technologies are expected to provide solutions to such problems. In fact, some imaging workstations are equipped with a dedicated report design tool that can customize the layout of images at the user’s discretion. The report structures are designed and stored in the form of XML files. By using such tools, users can modify the layout of elements in their reports, or change the styles (font, size, colors, and so on) of the elements, easily. Also, a variety of templates can be prepared and made readily available for creating diagnostic reports.3
Enhancing Reliability: Computer-Aided Diagnosis
In general, the reliability of diagnostic results tends to reflect the radiologists’ experience or expertise rather than the information systems support. Now, radiologists are acquiring increasingly complex sets of high-resolution image data, including large anatomic datasets generated by multi-detector CT scanners, which can result in studies exceeding 1,000 images, large combinations of anatomic and functional datasets such as complex MR sequences, complex functional datasets such as MR spectroscopy or functional MR imaging, and fusion datasets such as PET/CT.10 However, the available computing power and sophisticated image processing algorithms, together with the plentiful amount of information obtained from high-performance modalities, are radically changing radiologists’ ability to analyze and interpret the information contained in medical images.
The most noticeable trend in medical imaging is the simultaneous employment of multiple exams acquired at different times, in different patient postures, and even from different modalities. For example, dual CTC, which concurrently examines dual studies acquired from two different positions (the supine position and the prone position), is a widespread protocol to screen colon cancer. It can improve the accuracy and reliability of diagnosis compared with single CTC. In many angiographic analyses, it is also becoming increasingly popular to use two CT exams one scanned with contrast material injected and the other not to obtain more precise vessel segmentation. While performing angiography, radiologists are supported by various tools that can facilitate the exploration of vessel to find potential aneurysms.
Image fusion also can be thought of as a form of diagnosis that can increase the accuracy of radiologists’ interpretation, by taking advantage of heterogeneous images acquired from various modalities. Image fusions provide radiologists with collective information obtained by distinctive modalities. Three-dimensional images acquired by CT, MRI, PET, and other imaging devices are computed and merged into a single volume, thus combining the information of all of the modalities.5 Moreover, as a result of the great amount of attention that image fusion has received among radiologists, imaging devices that can perform dual scans at one time, for example, PET-CT or PET-MRI, have emerged. They can create a fused image by automatically registering and overlaying two distinctive types of images acquired from different modalities.
Computer-Aided Diagnosis (CAD) is also of great promise in digital imaging. Using novel computing methodologies developed in the field of Artificial Intelligence (AI), CAD tools support radiologists by presenting significant findings with regard to image interpretation. In this way, they can considerably reduce the number of human errors and mistakes in interpreting images, and thus can enhance the reliability of diagnostic results. For example, in detecting lung nodules from lung CT images, different radiologists might formulate different understandings of the same finding. In such cases, AI techniques can supplement radiologists’ diagnostic decisions by providing statistical or probabilistic backup data, which can be computed by using a prediction model associating the current finding with those of similar past cases found in diagnosis archives. Mammography is also a good example of the application of CAD techniques. In mammography, CAD tools can help find important information that is difficult to detect by eye.
Enhancing Usability: On-Demand Imaging Service
Digital imaging services usually involve enormous data processing and computation. As such, they require high computing power, which consequently hampers the usability of those services. The effective way to increase the usability of imaging services is to implement a powerful server that can provide those services to many remote clients having low-performance computing devices. However, the problem of limited performance and network bandwidth make the client-server architecture difficult to implement in digital imaging.
Advances in recent networking technologies and data compression techniques are making it possible to provide 3D imaging services through the Internet. This means that many radiologists can be simultaneously connected to a single powerful imaging server. An imaging server can provide a 3D imaging service over a network to several simultaneously connected thin-client PCs. Furthermore, some advanced features such as video conferencing, whereby physicians connected via thin client can observe 3D, Maximum Intensity Projection (MIP), and MPR manipulations, can be provided in real time. For example, vessel analysis on thin-client PCs is available using advanced Curved Planar Reformatting (CPR) functions that support automatic calculation of vessel stenosis. Many client-server-based imaging systems are commercially available.11 Additionally, the thin-client architecture in digital imaging is best suited to soliciting external radiologists for consultation. The external radiologists can access related images and reports, and, in that case, the benefits of using such architecture are obvious. Web-based access is manifestly the most preferable mode of external consultation.
Moreover, this on-demand imaging architecture also can change the traditional methods of working with diagnostic images. In effect, in this architecture, radiologists can have much more accessibility to original images for diagnosis, and thus they can be empowered to directly edit, modify, and even generate new images. This means that the traditional relationships between radiological technicians and radiologists can be rethought so that unnecessary conflicts between them, and the additional time and effort required to solve those conflicts, can be eliminated. Server-based imaging workstations might even encompass the traditional RIS functions. In practice, many workstations have workflow enhancements including scheduled auto-routing based on time of day or day of week, as well as the ability to generate reports using a selected word processor.
New Opportunities for Digital Imaging
It has been a primary role of radiology to provide exact diagnosis of disease by examining patient images. However, recent trends in digital imaging show that its role is being broadened to support a more extensive range of healthcare services, as illustrated in Figure 1. In various sectors of healthcare service, many innovative attempts are being made to take advantage of the high availability of digital images. Typical examples include image-based treatment planning, replica service, and image-guided surgical navigation.
A good example of image-based treatment planning can be found in dental implant surgery. In general, implant surgery is known to be a very delicate procedure requiring careful and cautious operative procedures, because the critically important inferior alveolar nerve (IAN) in the mandible area can be damaged while inserting the implant through the gums. Accordingly, the biggest concern of surgeons is to determine the accurate positioning, angle and depth of implantation so as to avoid harming the nerve.
Implant planning software, as shown in Figure 2, supports many useful functions such as implant simulation, inferior alveolar nerve identification, and evaluation of bone density, which can be performed preoperatively using CT images. Those functions help surgeons to prepare an accurate surgical plan in advance, and thereby enhance their ability to operate on patients with confidence. Furthermore, some such applications produce a surgical stent that, for dental surgeons, can serve as a physical guide for the positioning of an implant.
Sometimes, a volumetric image of a human body part is transformed into a physical model that can be used for surgical planning. Rapid Prototyping (RP) machines, which were originally introduced for rapid pre-manufacturing of mechanical parts, can produce a physical replica of a human body part with various materials based on a 3D model exported from imaging workstations. With a replica, surgeons can plan or simulate future treatment in a more intuitive way. For example, 3D replicas of various human bones can be used for surgical simulation of malformation surgery, maxillofacial surgery, and orthopedic surgery, among others.
In orthodontic treatment, a company has succeeded in devising a totally new treatment process by effectively leveraging the advanced imaging technology. Instead of using braces, the normal means of straightening teeth, they developed a totally different treatment procedure. Initially, bite impressions of a patient are taken, and are sent to the company for processing. Then, the company uses advanced 3D computer imaging technology to transform the bite impressions into a custom-made series of clear and removable aligners. The course of subsequent treatment involves changing the aligners approximately every two weeks, moving teeth into a straighter position step by step. And, unlike braces, these clear aligners can be removed while a patient eats or brushes their teeth.4
Similar cases of other companies that produce surgical devices can be found. One such company has revolutionized traditional surgical techniques by introducing into the market a new type of device, surgical navigators. Surgical navigators can provide surgeons with a means of navigating through the human body using advanced 3D images as their guide. As shown in the Figure 3, they can facilitate the surgical procedure by real-time correlation of the operative field to 3D patient images on the computer screen. This new way of performing an operation, called Image-guided surgery (IGS), allows ‘frameless stereotaxy’, or the precise guidance of a surgical instrument without need of a stationary device. By registering the pre-surgery image coordinates into the patient’s real anatomy coordinates, the position of surgical devices inside the patient’s body can be overlaid on 3D images in real time. Surgical instruments are affixed with light-emitting balls, so that their physical position can be tracked by an optical tracker in the operating room. This enables surgeons to see the exact location of the selected surgical instrument in the surrounding structures deep inside of a human body2.
In this article, we have examined the state-of-the-art of imaging technology and its future direction. Technologies provide us with the means of envisioning totally different forms of business. Digital imaging technologies will also continue to evolve, and their applications will be much more diversified than at present. Areas of application will not be limited to diagnostic imaging, but will expand to play a central role in treating patients. It is expected that digital imaging technology will overcome various limitations of human surgeons and even, by providing still more advanced services, replace them.
As we have discussed in this paper, information technology has the potential to support new ways of working and provide new services. But such innovations cannot be achieved without significant consideration on organizational issues in the healthcare domain. Work process of the healthcare domain is highly complex, distributed, dynamic, regulated, knowledge-intensive and often time-critical.7 Various medical staffs constantly need to cooperate with each other and extensive coordination are required between the medical actors involved.1,7,9
In these complex work settings, the successful implementation of information technology ultimately heavily depends upon the receptivity and preparedness of the healthcare organization. Therefore, in order to realize the actual benefits from technological advancement, substantial efforts have to be devoted to organizational change. This is because innovation is not just a matter of technology. Rather, it takes place through the relations between humans and technologies.7 This inter play between technology and organization is an ongoing challenging issue that has to be further explored.