As we move into the 21st century, informatics is progressing at an exponential rate?or is it? As we peruse through the vendors? booths at the Annual Meeting of the Healthcare Information and Management Systems Society (HIMSS) next month in New Orleans, we will see new features in both the hardware and software of PACS, RIS, speech-recognition software, decision support, and the list goes on. Back in the radiology department, more often than one would think, not much has changed in the past 5 years. DICOM and Health Level 7 (HL7) standards, as well as integration profiles put forth by Integrating the Healthcare Enterprise (IHE) initiative, have significantly expanded in the past couple years and offer software developers the framework for producing solutions with streamlined integration and interoperability. Most legacy PACS and RIS systems were developed when DICOM and HL7 were in their infancy and long before IHE was conceived.

In the New Year, as we get back to everyday life, let?s look into some of the things we would like to see available in PACS, from a radiologist?s perspective.

Better Worklists

Imagine logging into your PACS on a workday or while on call, and you are presented with a worklist populated with the studies that you are scheduled to read. All PACS vendors have various types of worklists?some more fully featured than others; some customizable, some not. Having a worklist or lists that can be coordinated with a given radiologist?s schedule to include sites covered, subspecialty, and the like, could greatly improve workflow and efficiency. If Dr Doe, the radiologist, is on a body CT rotation in the morning, does not read CT angiograms (CTAs) or high-resolution chest CTs, and is covering fetal ultrasounds from 11 am to noon, this should be easily configured into a worklist. There should be capability for in-progress and completed studies. A radiologist should be able to have one worklist that contains studies to be read, free from those that are waiting for outside comparisons, additional sequences, or a second opinion. The exam audit trail can often reveal a study that has been opened by nearly every radiologist in the group to find a note that so-and-so will dictate it, the patient will return for more sequences, and the list goes on. Worklists that contain studies that are not ready to be interpreted can be a significant impairment to workflow and patient care. It is also an impairment for a radiologist to have to interpret studies from multiple worklists. It would be optimal if a radiologist could select the first study from the worklist in the morning, and after that, the next study would come up automatically. Of course, this works only if all the studies on the list are ready to be interpreted. Some departments have radiologists, fellows, or residents assigned to scout the worklists to figure out why certain studies have not been interpreted. Frequently, only the system administrator can set up worklists. Implementation of a user-friendly graphical user interface (GUI) for setting up worklists could significantly improve workflow.

Hanging Protocols That Work

Imagine being able to pull up an MRI of the liver with the new study and the old study on the monitors of your preference; the right number of stacks displayed; the window-level setting correct; the sequences in the right order; the stacks ordered properly with respect to scrolling cranial to caudal, left to right, and anterior to posterior; and your choice of sets so you could display the next group of series according to your choice. Full implementation of DICOM hanging protocols (DICOM HPs) could provide this environment. DICOM HPs are a relatively new addition to the standard. HPs can facilitate consistent display of a study under preferences of the site, section, or user. The protocol can be specific for anatomic region, laterality, procedure code, and reason for procedure. They can be environment specific in terms of number of displays. DICOM HPs also can dictate sorting of a series in cases where a user chooses to view all chest CTs cephalad to caudal, for example. Various processing of image data can be specified. These HPs can be vendor neutral.

Most vendors have implemented some method of user-defined HPs. A number of issues hamper the development of rich?DICOM, if you will?HPs. To be able to develop an HP that would consistently display, for example, an MRI of the liver, the institution would have to have a set protocol for how that examination is performed. There can be significant variability among radiologists in a subspecialized section where different radiologists perform different MR sequences to image the same anatomy. A different HP could be written for each of these radiologists, but this solution would not take into account situations of missing and/or repeated sequences. DICOM image, series, and study attributes that are used in an HP by the image-display manager must be specific and consistent. Many institutions use the same name for examinations of multiple anatomic regions?for example, an MR of the ankle is called MR lower extremity, as is an MR of the knee or the tibia/fibula. Sites that have modalities from different vendors often have different naming conventions for the same series. Cooperation between modality, RIS, and PACS vendors is imperative to implementing DICOM HPs.

Full implementation of DICOM HPs would require a GUI for setting up and creating these DICOM objects that would be intuitive for both radiologists and engineers, would be efficient, and would not allow rule breaking (that is, would not create conflicting HPs). It would also require vendor implementation of a display manager that can match the correct HP object to the current study and display the study properly.

Timely Access to Old Studies

Prefetching algorithms have been implemented to some extent by most vendors and are not necessary in some cases in which all the previous exams are stored on rapidly accessible media. Many systems still require that the radiologist initiate fetching of significant prior studies. This can introduce inefficiency as the radiologist waits for the prior study to load, perhaps forgetting what he or she aimed to compare and having to revisit the current study again. This could potentially detract from patient care if a radiologist does not have the patience to wait for that CT from 5 years ago, for example. DICOM HPs may be the solution for determining which priors are relevant, as dictated by user preference. Network bandwidth and short-term storage capacity often limits a site?s ability to retrieve all prior studies for every current exam.

Postprocessing and Navigation Fixes

Navigation through studies and series within studies is becoming more challenging as they become larger and more complex. In years past, it was thought that more is better?so to speak, that the thinner the slices, the more information we can glean from them. Many sites routinely do sagittal and coronal reconstruction images of CT studies as well as process the data using a soft tissue and lung kernel. These additional stacks can be a mixed blessing. Future PACS should probably be able to do real-time multiplanar renditions of data sets on the radiologist viewing station so that any cutplane, maximum intensity projection (MIP), or volume rendering can be performed on the fly. For this process to be efficient, many of the steps would have to be automatic. Currently, postprocessing can take more than 30 minutes per case, which is a formidable amount of time to be spent by a radiologist in a busy practice. Technologists can produce beautiful volume-rendered images from extensive processing of CT and MRI data. But without the skill set to interpret these images at the same time, they can underprocess lesions and can publish less-than-optimal postprocessing to the PACS for permanent archival. Post-processing at the radiologist workstation currently can be performed through third-party software, some of which can be integrated into the PACS software. This is more of a rarity than the norm, and in the best of cases, it still can be very time-consuming. Improved algorithms for automatic postprocessing of data may improve efficiency while it can be more accurate and selective.

Getting Pertinent?or Any?History

Garbage in, garbage out. In any radiology department, this is as common as a dry day in the desert. ?Why do they want this exam?? ?What?s the clinical history?? Communication between ordering physicians and radiologists is suboptimal at best. Clinicians often get blamed for this with the assertion that they do not take the time to enter the history. Many clinicians enter lengthy histories into fields on paper forms, and only a portion gets inserted into the RIS. Handwriting can be a major impediment. Sometimes, technologists enter long histories into the RIS, and these histories do not populate any of the fields that the radiologist can see in the PACS. In other cases, clinicians enter orders in the HIS and the information is not fed properly into the RIS or PACS. This can seriously compromise patient care. Radiologists do not know what the patient?s symptoms are, so they give a differential diagnosis for any condition that could cause the findings. If the history of trauma is known, for example, a radiologist has an explanation for why a patient has a subdural hemorrhage. Without this history, it could be due to trauma, bleeding disorder, ruptured aneurysm, and so on. Some-times, this can be even more problematic. Imagine reading a CT scan of the abdomen with no history, which, by the way, is much more common than one would hope. In this case, the presence of pain, the location of the pain, its duration, the physical exam findings, and the laboratory findings would greatly help the radiologist make the right diagnosis and advise further analysis, in the case of a normal exam. The ordering physician often cannot be reached in a timely manner due to clinical or operating-room schedule, vacation, or off-hours.

Slow Deployment of IHE Integration

So, what?s the answer? IHE, of course. If every vendor tailored its software to be an IHE actor in an integration profile, health information and imaging would be more of a plug-and-play situation. Every HIS, RIS, PACS, and other health software vendor would have a defined piece or number of pieces of the puzzle. All PACS are not equal, though. In an IHE world, each actor has a strictly defined role. Each integration profile defines how information is to be shared between systems using HL7 and DICOM standards.

Why hasn?t every vendor conformed to an integration profile, and why isn?t the world of health imaging a fully integrated, efficient, user-friendly, intuitive place? Many radiologists are quick to point their finger at the vendors who, they say, are quick to sell them a system and are very slow to support or upgrade it. Vendors do have to turn a profit for their stockholders, assuming they are publicly traded, some of which are radiologists. As in many companies, there is disparagement between those in charge of improving the product through research and testing and those who are responsible for showing acceptable earnings each quarter. It is very challenging for a PACS vendor to devise the perfect system, because each site is different. Without IHE compliance among all hardware and software vendors, PACS vendors have to come up with individualized solutions for integrating every modality and piece of software a site uses to make their systems functional. Sites are sometimes not aware of what they need and what to buy from a PACS vendor, and the vendors can only work with the information they are provided.

Lack of Radiologist Participation

Some point the finger at the DICOM or HL7 committees saying they come up with ambiguous standards that have so many options that a vendor cannot possibly come out with a standard solution. Who composes the standards committees? How many radiologists are there on these committees? DICOM committees are predominantly composed of representatives from different vendors. Vendors must be familiar with the standards to develop software that is compliant, so they are inherently more familiar with them. Radiologists, on the other hand, spend a tremendous number of hours and amount of energy learning clinical and image-interpretative skills. This limits their ability to become familiar with the standards and participate in committees that develop standards. The standards committees, contrary to popular belief, welcome input from radiologists. Members of many of the standards committees dedicate a lot of unpaid hours, travel, and vacation time to making improvements. As radiologists, we are not the only ones who want better systems. Everyone has a family member who was once ill and could have benefited from better care.

Who do we blame for the current PACS inadequacies? Everyone and no one. We are all making progress; it is just not enough to satisfy most. Radiologists need to become more familiar with the standards that affect us. We need to be proactive in participating, not only in DICOM, HL7, and IHE committees, but within the societal infrastructure at large that affects us as radiologists, physicians, and human beings. We cannot wait for someone else to make things better. We all need to participate.

Benjamin Johnson, MD, is an imaging informatics fellow in the department of radiology, Northwestern University, Chicago. For more information, contact .