Home

Cover Story

Table of Contents

E-Newsletter

Article Archive

Editorial Calendar

Datebook

Writers' Guidelines

Orgs/Links

Opinion Polls

Reprints

Forum

July 16, 2007

Looking at Things Differently
By David Yeager
Radiology Today
Vol. 8 No. 14 P. 18

The ever-growing number of images that radiologists must review calls for them to change how they look at images. Experts share their views on how to maximize reading efficiency, including how to harness computer technology for the task.

The numbers are daunting. While advancement in medical imaging technology produces more detailed studies and, presumably, more accurate diagnoses, it has also spawned greater numbers of images. And with image volumes constantly growing, clinicians are looking for ways to improve accuracy and speed by better utilizing technology and the human eye.

“I’d say, on the low end, if you’re doing 1,000-image CTs and MRs, you might have to look at 30,000 to 50,000 images in a day, and on the high end, you might get up to 150,000 to 200,000 images,” says Murray Reicher, MD, a diagnostic radiologist and chairman of DR Systems, Inc.

For radiologists at the high end of that range, Reicher points out that this leaves approximately 0.1 seconds for viewing an image per four hours of reading. If radiologists and radiology departments are going to keep up with their reporting in this instant gratification world, they’re going to have to rethink how they read and process images.

As J. Neal Rutledge, MD, an interventional neuroradiologist with the Austin (Tex.) Radiology Association explains it, “The explosion of images requires that we rethink our review paradigm. While it’s currently possible to review every image, it quickly becomes impossible when there are 10,000 plus images per case. I believe we will need to mimic the way pathologists review cases, [using] gross overviews then selected, detailed review.”

To manage this onslaught of images, forward-thinking radiologists are continually seeking to streamline the reading experience. To achieve that objective, they’re looking to optimize efficiency in the human eye, the reading room environment, technology, and workflow.

The Eyes Have It
The human eye is the key mediator between the clinical image and a radiologist’s clinical opinion. But image presentation can have a drastic effect on how the eye processes light, which can subsequently have a dramatic effect on how the brain perceives images. An old adage cautions that people shouldn’t believe anything they hear and only half of what they see. Based on what is known about vision, an individual’s perception of an image depends largely on how the image is viewed.

“If you see a cluster of calcifications on a single 2-D view of the breast, a common inference of perception is that they are grouped together,” says Rutledge. “Our visual system uses assumptions when processing information to infer what is present in the world.”

Two major visual pathways operate at the same time in human brains. One processes the “where” information in the parietal lobe of the brain; the other processes the “what” information in the temporal lobe. The brain, however, only sees one image. “So what’s very interesting is we’re having multiple parts of our brain assess what’s going on in the image, but we’re only conscious of one of them at a time,” says Rutledge. Perception and lesion detection can be improved by recognizing this limitation of consciousness.

“Additional things can be picked up. If you’re looking at an image and you have a feeling that something’s wrong, you should pay attention because it may be the alternative pathway that’s trying to clue you in and trying to tell your consciousness, ‘Hey, look at this,’” Rutledge adds. “And so it’s basically, ‘trust your feelings,’ which a lot of people don’t. Do an analytical, systematic approach to your image review, but if you feel something’s funny, take more time to assess.”

Rutledge says two fundamental visual biases relate particularly to interpreting images: change blindness and inattention blindness. Change blindness frequently occurs when looking at similar images, for example, comparisons. The images may have obvious differences, but absence of focused attention on the area of change makes them hard to detect. Inattention blindness can occur if a clinician is looking too narrowly for a particular finding. The search for what’s expected may cause an unexpected finding to be missed. When the fatigue factor that accompanies viewing a large number of images is added, more misses can occur. A proper viewing environment becomes paramount.

Reading Environment
One of the difficulties with many reading rooms is the presence of confounding variables that can alter image perception. “The confounding variables that are most important are reflections because they trigger motion sensitivity. Other examples include moving the mouse on the screen or multiple cells scrolling at the same time,” says Rutledge. “People who are putting buttons all over the screens, that’s the wrong way to go. They’re distracters that decrease your accuracy rate.”

Another mistake people frequently make, according to Rutledge, is keeping the reading room too dark. “That’s another thing that hits the iris. When you change gaze from a black background to a bright monitor, your iris is constricting. Frequent changes add to a fatigue factor, so optimally your background, the wall behind the monitor, should be about the same brightness as the monitor.”

As far as the monitors are concerned, Rutledge states that the fovea of the eye dictates the optimum specifications. “Your sweet spot of vision, [the] high resolution foveal part of your retina, covers only a small part of any image, about the size of your thumbnail at arm’s length; its best resolution is about 300 pixels per inch at 300 candelas per meter squared or above, so buy appropriately. I use big monitors but try to keep image size down to about six inches, as long as I’m not losing resolution, to minimize eye movement,” he says. “Blowing a CT up to the size of a wall wastes time and energy as the eyes scan the image and, over time, results in quicker fatigue and increased errors.” One low-tech remedy that he advocates is taking a short break—five minutes or so—every hour to give your visual system a chance to rest.

Along with an optimized physical environment, one of the best ways to limit mistakes is to limit interruptions. One way Reicher does this is by getting his reports out early. “If a referring physician is calling me, he’s not calling to ask where my report is; he’s calling to discuss a case. So I’m not constantly being interrupted while I’m trying to read.”

Utilizing Technology
Even with a well-designed reading room, radiologists are still faced with an ever-growing caseload. Computer software such as computer assisted detection (CAD) can help, Rutledge says, because it consistently performs the same operations, reducing fatigue. While this technology has its benefits, it doesn’t fundamentally address the way people perceive images. New technological approaches may help to minimize the effects of visual biases.

“The one that’s been ignored largely is, ‘We’re going to use computers to present information and images in such a way that the human computer is adapted to take greater advantage of [them],’” says Reicher. “We’re going to take advantage of human perceptive physiology, the whole visual cognitive pathway, and the way it’s designed in our brain to present things in a way where we’re natively able to make the analysis better than when it’s presented another way. And that’s kind of an exciting thing.”

One technology that appears to make good use of human perceptive strengths is blinking, sometimes referred to as shuffling or flip-flop. “It refers to the ability to take images that are either a series of images from a new and an old case or a series of images done previous to contrast and post-contrast injection or with dynamic injection,” Reicher explains, “and to rearrange those images so that instead of having to turn your head and look at the comparative images side-by-side, you can commingle two, three, four, or more stacks of images and be able to, for example, watch contrast flow into and out of a lesion or watch not just contrast but watch the passage of time as the same anatomical area is imaged.”

More advantageous presentation methods may help clinicians identify pathology more accurately, says Reicher, but that’s only part of the process. “It’s not a one-step thing. One is the optical presentation but being able to present the precise history, the old history, the referral documents,” in a timely and efficient manner are important aspects of pathology perception, he adds.

“It’s called ‘just in time’; it’s a concept that’s been pushed in the military. Just-in-time information, heads-up displays,” says Rutledge. “In war, there’s a thing called information fog, or information overload, so what you want to do is limit input to just the information needed, just at the time it’s needed.” Rutledge’s reading routine is designed to combat information overload. “I review starting with the images with the highest conspicuity, and then those with optimized characteristics,” he explains, “then I go through my checklist, get the history, and go back, and I do focal examination based on the history.”

Reicher advocates using a template or itemized checklist report “that pops up, and before you’ve even looked at the CT abdomen and pelvis, the report is populated with the patient information, the appropriate history, the appropriate demographic, [and] the technique that was performed,” in order to help the clinician find potential problems faster and improve accuracy.

“The analogy that I use is: Suppose a radiologist takes care of 120 patients in a day. Today, in most practices, he does that without a checklist. The average radiologist says, ‘Well, I don’t need a checklist, I’m a smart guy.’” Reicher says. “But a pilot wouldn’t even dream of getting on an airplane and flying in the plane without a checklist even though he’s a smart guy.”

Reicher also notes that better workflow tracking can improve efficiency, a process he calls report output action tracking. “When I click to approve my radiology report, within seconds, I have a heads-up display that indicates not only that the report was supposed to be archived, but that it did get archived,” he says. “Whatever the automated output action—faxing, archiving, sending to a third party billing system, distributing on the Web—action tracking provides immediate feedback as to whether the output action was attempted, successful, or failed. By receiving this feedback immediately, users can intervene appropriately to assure prompt, automated communication.”

Workload Distribution
With digital technology allowing greater data manipulation, is there a danger that more technologist duties may be task-shifted to radiologists, increasing fatigue and reducing efficiency? “I think that’s a myth,” says Reicher. “That has nothing to do with digital technology. That has to do with where you decided to distribute the workload.”

The goal, Reicher says, is to implement systems that balance the workload. “There are some systems that are implemented with the concept that ‘We’re going to bypass the technologist completely in a way that dumps the work on the radiologist or potentially vice versa,’” he says. “I think we’re now seeing more intelligently designed systems that systematically automate any rote activity so that technologists, radiologists, and other professionals all have more time to focus on their uniquely human abilities.”

By appropriately organizing workflow and utilizing digital technology to perform repetitive tasks, radiology departments can help their staffs be more productive. “Any activity that requires the unique experience and analytical ability of a human being, the human being should do,” says Reicher. “Newer systems contain master files that store information about the patient, exam, history, referring physicians, allergies, site policies, exam techniques, billing codes, supplies, reports, forms, critical results, insurance, user preferences, and more. As a result, workflow becomes automated, and the greater the automation, the more time the doctors, technologists, IT staff, management, and other support can spend focusing on their clinical and financial success.” By planning for and adapting to the changing nature of medical imaging, radiology departments can ensure that the light they see at the end of the tunnel isn’t a runaway train of studies.

— David Yeager is assistant editor of Radiology Today.

Practical Tips for Optimizing Interpretation
When you’re dealing with economies of scale, little things add up fast. The growing volume of imaging studies is one example of this phenomenon. A few extra seconds per image can drastically reduce the number of studies read in a day, which can cause backlogs and increase radiologist fatigue, a factor that’s known to adversely affect reading accuracy. There are a few steps that radiologists and radiology departments can take to maximize efficiency.

Murray Reicher, MD, a diagnostic radiologist and chairman of DR Systems, Inc., explains that several things happen before and after an image is read that can have an impact on efficiency. He notes that good histories, scanned documents, and precise communication between the referring physician and the radiologist are critical for reducing reading time and improving accuracy.

“The next step is that the images have to be acquired in a way that optimizes detection, which has to do with things like technologist protocols, technologists being able to look at the old study in order to be able to replicate techniques, and then it moves to the reading environment where image presentation methods are important,” says Reicher.

Once the images have been acquired, radiologists can help themselves by controlling some of the variables in their work environment. J. Neal Rutledge, MD, an interventional neuroradiologist with the Austin (Tex.) Radiological Association, offers the following tips for optimizing a reading room:

• Use large, flat-screen monitors with resolutions at or above 300 pixels per inch and illumination of 300 candelas per meter squared or brighter.

• Do not keep the room too dark; it forces the eye to work harder when shifting from a dark background to a bright monitor.

• Work in a quiet place and try to limit distractions.

• Try to minimize reflections and buttons on the screen and use a small field of view to look at images.

• Follow a set reading pattern, review a checklist and clinical histories, and take regular breaks to reduce fatigue.

— DY

Do you look at volume rendering first and then individual slices? Share you thoughts here.