View larger version (85K): [in this window] [in a new window]   Figure 2. Hard-tissue chart in Patterson Eaglesoft (Patterson Dental Supply, St. Paul, Minn.). Reprinted with permission of Patterson Dental Supply.

In recent years, we have seen the rapid emergence of a variety of clinical software applications that incorporate 3-D imaging. Those applications have been used for diagnosis and treatment planning in orthodontics,^5–9 computer-assisted planning of oral implant surgery,¹⁰ the design and production of fixed restorations,¹¹ dental education^12,13 and patient education.^14,15 Dentrix Dental Systems (American Fork, Utah) implemented a basic 3-D model in the most recent release of its practice management system, Dentrix G2.¹⁶ Recent technological advances have made it possible to develop such applications in dentistry.^17–20 (Owing to space reasons, we do not elaborate on the details of existing 3-D systems and underlying imaging modalities. Interested readers can find more information on those topics in the literature we have referenced.)

Before attempting to study the larger research question of how a 3-D model would affect general dental care, we decided to investigate the feasibility of a 3-D patient record system. In this study, we constructed a basic prototype for such a system and conducted an early evaluation with representative end users. We were interested primarily in how a 3-D model of the patient’s dentition could be integrated with other information in a patient’s record, how clinicians could interact with the model and what novel functions could be associated with the 3-D model. Clearly, this study is only a first step in determining whether 3-D patient records in general dentistry can transcend some of the limitations of today’s approach—that is, the paper chart—and thus provide value for clinicians.

	MATERIALS, METHODS AND PARTICIPANTS

TOP ABSTRACT MATERIALS, METHODS AND... RESULTS DISCUSSION CONCLUSION REFERENCES

 
We used a user-centered design approach (sidebar, "What Is User-Centered Design?", page 1081) to develop the system described in this article. An interdisciplinary team of five undergraduate students in human-computer interaction (HCI), one faculty member and one postdoctoral associate in dental informatics, and one computer science faculty member collaborated on a proof of concept for a 3-D dental record. The students’ background included computer science, psychology, design, decision science and information science. We completed the project in 14 weeks as part of an HCI undergraduate project course at Carnegie Mellon University (CMU) in Pittsburgh.

The project extended over five phases: background research, low-fidelity (lo-fi) prototyping and evaluation, high-fidelity (hi-fi) prototyping, user testing and finalization. In answer to the research goals, the team conducted background research with the following aims: understand how dentists and dental auxiliaries use dental records in the clinical context; identify dentists’ informational needs for performing clinical tasks; gain insights into the modalities of data entry (who, what, when, how); gain a general understanding of the advantages, disadvantages and potential areas for improvement of paper records and current software packages; and synthesize the findings into a set of design implications for the next phase.

During the background research phase, the CMU students reviewed research papers and the results of recently completed and ongoing evaluation studies of the Center for Dental Informatics at the University of Pittsburgh School of Dental Medicine; examined the functionality and design of several major dental practice management programs and two 3-D applications (Google Earth [Google, Mountain View, Calif.], a mapping application, and SolidWorks [SolidWorks, Concord, Mass.], a 3-D engineering and modeling tool); and conducted observations in four dental offices in the Pittsburgh metropolitan area. (Owing to time constraints and lack of access to programs, we could not include other 3-D applications, such as CEREC [Sirona, Bersheim, Germany], OrthoCAD [Cadent, Carlstadt, N.J.] and Invisalign [Align Technology, Santa Clara, Calif.] in our background research.) The team observed clinicians at work not only to understand the clinical work flow, but also to find "breakdowns"—that is, instances in which technology or other factors created barriers to or inefficiencies in completing clinical tasks. We intended the new design to address these breakdowns, if possible.

After the background research phase, we constrained the scope of the project to increase its feasibility. First, we decided not to implement any data entry functions and instead focused only on information and how it should be displayed. Second, we designed the system to display selected patient information at a limited level of detail. Third, we decided that the prototype would support a limited number of clinical tasks and exclude any administrative functions, such as billing and scheduling. Last, we planned to create a prototype that was refined enough for limited user testing and that would allow us to identify the strengths and weaknesses in our design.

Based on a conceptual design and usage scenarios derived from the background research, we proceeded to develop a series of three lo-fi prototypes on paper. The project team’s dental informatics experts (T.K.L.S. and T.P.T.) evaluated and critiqued the first prototype; the second and third prototype were subjected to formal user testing with two and three dentists, respectively. On the basis of the results of the user tests, the research team modified and implemented the third paper prototype as a first version of a hi-fi prototype in Java, Java3-D and JFlashPlayer (all Version 1.5, Sun Microsystems, Santa Clara, Calif.) and Macromedia Flash 8 (Adobe Systems, San Jose, Calif.) on a tablet personal computer (PC) running Windows XP Professional (Microsoft, Redmond, Wash.). The first version of the hi-fi prototype included a stylized, hand-drawn 3-D model of a patient’s dentition; intra-oral images and radiographs; and selected other clinical information, such as medical alerts, recent progress notes and planned procedures. In the final prototype, we replaced the stylized 3-D model with a high-resolution scan of the maxillo-facial portion of a skull.

We recruited two faculty members and two students at the University of Pittsburgh School of Dental Medicine to evaluate the hi-fi prototype. The participants practiced dentistry three days a week or more and had experience with dental computer systems. Because the objective of the user tests was to evaluate the intuitiveness and understandability of the prototype, we gave the participants no training. We asked them to complete three clinical tasks:

– retrieve clinical information—such as the periodontal status, a hard-tissue–only view of the dentition and radiographs—about the patient;

– identify clinical findings from the data provided;

– diagnose and plan treatment for the condition of tooth no. 12 using the 3-D view.

We conducted the evaluation using a think-aloud protocol, which is part of the standard usability testing methodology.²¹ Two observers (M.M. and P.P.) recorded each experiment; in addition, we captured a video of the computer screen and the corresponding audio track using Camtasia Studio Version 2.1.0 (Techsmith, Okemos, Mich.).

The primary objective of the user test was to gather qualitative data about the hi-fi prototype. Since we tested only a proof of concept, we determined that it would be inappropriate to measure more quantitative aspects, such as task completion time and error rate. Rather, qualitative data provided us with a more useful and richer data set, because they showed more clearly the degree to which the users understood the system in the way that the designers intended, as well as which aspects of the system would require additional design work.

The development team finalized the hi-fi prototype by implementing additional changes that were based on the results of the user tests. We did not seek outside evaluation of the final prototype again because the design changes were too limited to warrant the expense and effort of an additional user test. In the Results section, we describe the major findings from the background research and the resulting design goals. Because a full description of all the prototype versions is beyond the scope of this article, we describe only the final prototype to help readers understand its design and function. We highlight differences from previous versions where appropriate, as well as findings from the user tests that led to the final design.

This study was approved by the University of Pittsburgh Institutional Review Board.

	RESULTS

TOP ABSTRACT MATERIALS, METHODS AND... RESULTS DISCUSSION CONCLUSION REFERENCES

 
In developing the prototype, we were guided by several design goals formulated as a result of our background research (Table 1). Our intent was to take a fresh, unbiased look at the design of computer-based dental records that integrate a 3-D model of the patient’s dentition; therefore, the principles are general and address only high-level issues, such as efficiency of computer use and infection control requirements. One key intent of our prototype was to develop a system that was centered on the requirements and needs of dentists and dental hygienists and, thus, was more likely to be used by them routinely than is currently the case.²²

Integration of the 3-D model with existing patient information. Figure 3 shows the main screen of our final prototype (featuring a case described in Stefanac and Nesbit²³). It uses a three-pane design that provides access to the patient information summary, data from clinical examinations and treatment plans. (The Treatment Plan pane, intended for viewing and entering treatment plans, was outside the scope of implementation of the prototype and is not discussed further in this article.) On startup, the application displays the Patient Information Summary. The goal of this view is to provide the clinician with an overview of the patient’s status. The left portion of the screen displays the patient’s picture and general information, such as current medications and medical alerts. Personal notes at the bottom can be hidden from the patient. The center of the screen shows additional clinical information, such as the chief complaint, progress notes, planned procedures and medical conditions.

View larger version (54K): [in this window] [in a new window]   Figure 3. Startup view of the three-dimensional (3-D) dental record prototype. The screen is divided into three panes: patient information summary, clinical examinations and treatment plans. Patient images reprinted with permission of Elsevier from Stefanac and Nesbit.23 Image of 3-D model published with permission of Brown and Herbranson Imaging, Portola Valley, Calif.

View larger version (65K): [in this window] [in a new window]   Figure 4. Three-dimensional (3-D) dental record prototype with the Exam pane maximized. The Exam pane shows an actual 3-D model of the patient’s dentition and associated radiographs and intraoral photographs. Patient images reprinted with permission of Elsevier from Stefanac and Nesbit.23 Image of 3-D model published with permission of Brown and Herbranson Imaging, Portola Valley, Calif.

The users interacted with the prototype, including the 3-D model, by means of a stylus. The stylus allows the screen design to be more detailed and dense, since it has a greater precision in pointing than does the finger one typically uses with a touch screen. In addition, a stylus resembles many of the hand instruments used during clinical care and, thus, does not require the clinician to use different psychomotor functions when using the computer.

Clinically relevant functions associated with the 3-D model. The Exam Pane implements four navigation and display functions, several of which are driven by interacting directly with the 3-D model.

Automatic image/radiograph retrieval. Most practice management systems require an explicit user action to retrieve clinical images or radiographs associated with an area of interest. For instance, the user typically selects the image of a particular tooth to display one or more corresponding radiographs. Our prototype eliminates this inefficiency by automatically displaying the corresponding radiographs and clinical images when the user rotates the 3-D model. For instance, when the user looks at the frontal view of the model, the software displays radiographs that show all or a portion of the anterior sextant. Similarly, rotating the model to the right displays the images and radiographs for the right sextants.

Flexible teeth, soft tissue and bone view. The user can display the patient’s teeth, soft tissue and bone individually or in any combination (Figure 5 shows sample views). For instance, endodontists can examine a tooth’s root morphology in detail, and periodontists can do the same with alveolar bone. In addition, the user can display completed and planned procedures.

View larger version (36K): [in this window] [in a new window]   Figure 5. Selected views of the dentition in the three-dimensional (3-D) model. A. Bone, soft tissue and teeth. B. Soft tissue and teeth. C. Teeth. Image of 3-D model published with permission of Brown and Herbranson Imaging, Portola Valley, Calif.

Focus on a single tooth. By right-clicking on a tooth in the 3-D model, the user can bring the particular tooth into focus. In our prototype, the model is repositioned to display the selected tooth in the center of the screen, and the other structures become transparent. Additionally, the tooth becomes the center point for the model’s rotation. An enhanced prototype we are planning will display data pertaining to the particular tooth.

The prototype implements additional functions, such as maximizing clinical images and displaying a 2-D intraoral chart. However, owing to space constraints, we will not describe these features in this article.

Selected results from user tests. Table 2 shows the results of the user tests with the almost-final prototype, which resembled the system described above very closely. (We describe relevant differences in the text below.) Although all users retrieved at least one selected patient information item in task 1, their success in doing so varied. No test participant was successful in retrieving all information items. All test participants successfully completed task 2, obtaining additional clinical findings. None of the participants completed the diagnosis and treatment plan called for in task 3.

During the tests, several users commented on the value of selected features of the program, such as the automatic display of corresponding clinical photographs and radiographs when the 3-D model was manipulated, the ability to view the patient’s dentition from any angle and the easy access to historical clinical data. Users also said they appreciated the exclusively clinical focus of our application.

	DISCUSSION

TOP ABSTRACT MATERIALS, METHODS AND... RESULTS DISCUSSION CONCLUSION REFERENCES

 
In this research project, we developed a prototype of a 3-D computer-based dental record for general dentistry and conducted an initial evaluation of it with dentists and dental students. While the prototype included only limited functionality, our evaluation demonstrated that the test participants conceptually understood and welcomed the program’s addition of a 3-D model of the patient’s dentition and the associated features. The user tests showed that participants were able to complete several information retrieval tasks relatively easily, but the test results also highlighted the challenges of designing a powerful and easy-to-use 3-D dental record. While our approach to designing the interaction with the 3-D model seemed appropriate for the users, we identified several usability issues that we addressed in the final prototype. The major problems that users found centered on aspects of interacting with the 3-D model, the insufficient realism of the earlier 3-D model of the dentition and the lack of intuitiveness of novel features, such as selecting a tooth on the 3-D model.

The development and evaluation of the prototype were subject to several limitations. Since the design of computer-based dental records is a large and complex undertaking, we limited the scope of this project significantly to produce a workable prototype within our resource constraints. We did not design a detailed interface for many important clinical information categories, such as the medical history, progress notes and treatment plans. As evidenced by current practice management systems, the multitude of data types and program functions, as well as the volume of data, create difficult and challenging design problems that are not overcome easily.²⁴ The lack of refinement of the 3-D model clearly affected the sense of realism that test participants experienced and made it impossible for them to complete higher-order clinical tasks. An evaluation with a larger set of representative end users most likely would have resulted in a richer understanding of strengths and weaknesses of the user interface and produced more design suggestions. While Nielsen and Landauer²⁵ considered between three and five users to be typically sufficient to identify more than 60 percent of the usability problems in a system, Shneiderman and Plaisant²⁶ pointed out that this low range is controversial. Since the primary purpose of this study was to establish the feasibility of a 3-D record system and not strictly to validate our particular design, we considered the number of test participants to be adequate. A relatively small team of core developers (five people) and a constrained project time frame (14 weeks) limited our ability to refine this project further.

Our prototype opens a wide perspective for further research. First, additional development should make the feature set more robust. For instance, while our prototype implemented one approach to interacting with a 3-D model (several standard views combined with the ability to rotate and zoom freely), other alternatives are possible. In addition, further studies should determine which degree of realism of the 3-D model (for instance, with respect to resolution, coloring and shading) is necessary to support diagnostic and therapeutic decision making. Finally, a more mature system should be compared with existing practice management software on a number of performance parameters, such as learnability, ease of use and error rate.

	CONCLUSION

TOP ABSTRACT MATERIALS, METHODS AND... RESULTS DISCUSSION CONCLUSION REFERENCES

 
Users who tested our prototype of a 3-D charting interface for a general dental record accepted the prototype’s basic design, but they also identified several challenges in developing intuitive, easy-to-use navigation methods and hi-fi representations in a 3-D record. Significant further research must be conducted before the concept can be applied and evaluated in clinical practice.