The purpose of this article is to describe a system we developed for importing images on compact discs (CDs) from external imaging departments into our clinical image viewing system, and to report on key metrics regarding veracity of information seen on the CDs. We recommend careful attention to the process of CD importation because of the error rate we have seen. We developed a system and process for importing images on CD into our EMR. The importation system scans the CD for digital imaging and communications in medicine (DICOM) images, and collects all patient information seen. That information is presented to the patient for verification. Once validated, the image data is copied into our clinical viewing system. The importation system includes facilities for collecting instances of incorrect data. About 90% of images are now exchanged between our healthcare enterprise and other entities via CD. Data for the wrong patient (e.g., the wrong CD) is seen in about 0.1% of cases, and a similar number of CDs have data for more than one patient on the CD(s) the patient bring to our facility. Most data are now exchanged via DICOM files. DICOM images burned onto CD media are now commonly used for image exchange. However, applications to import DICOM images are not enough. One must implement a process to assure high confidence that the data imported belongs to the patient you are importing.

Background

Plasma exchange (PLEX) is a beneficial rescue therapy for acute, steroid-refractory central nervous system inflammatory demyelinating disease (CNS-IDD). Despite the ~45% PLEX response rate reported among CNS-IDD patients, determinants of interindividual differences in PLEX response are not well characterized.

Objective

To perform an exploratory analysis of clinical, radiographic, and serological features associated with beneficial PLEX response.

Design

Historical cohort study

Setting

Neurology practice, Mayo Clinic College of Medicine, Rochester, Minnesota.

Patients

All Mayo Clinic patients treated with PLEX between1990–2007 for a steroid-refractory CNS-IDD attack.

Main Outcome Measure

The primary outcome was PLEX response in attack-related, targeted neurological deficit(s) (TND) assessed within the 6 month period following PLEX.

Results

We identified 153 patients treated with PLEX for a steroid-refractory CNS-IDD, of whom 90 (59%) exhibited moderate to marked functional neurological improvement within 6 months following treatment. Pre-PLEX clinical features associated with a beneficial PLEX response were shorter disease duration (p=0.02) and preserved deep tendon reflexes (p=0.001); post-PLEX variables included a diagnosis of relapsing-remitting MS (RRMS) (p=0.008) and a lower EDSS (p<0.001) at last follow-up. PLEX was less effective for MS patients who subsequently developed a progressive disease course (p=0.046). Radiographic features associated with a beneficial PLEX response were presence of ring-enhancing lesions (RELs; OR=4.0, p=0.029) and/or mass effect (OR=3.0, p=0.024). No association was found between NMO-IgG serostatus and PLEX response.

Conclusions

We have identified clinical and radiographic features which may aid in identifying those patients with fulminant, steroid-refractory CNS-IDD attacks who are more likely to respond to PLEX.

Intracranial aneurysms represent a significant cause of morbidity and mortality. While the risk factors for aneurysm formation are known, the detection of aneurysms remains challenging. Magnetic resonance angiography (MRA) has recently emerged as a useful non-invasive method for aneurysm detection. However, even for experienced neuroradiologists, the sensitivity to small (<5 mm) aneurysms in MRA images is poor, on the order of 30~60% in recent, large series. We describe a fully automated computer-aided detection (CAD) scheme for detecting aneurysms on 3D time-of-flight (TOF) MRA images. The scheme locates points of interest (POIs) on individual MRA datasets by combining two complementary techniques. The first technique segments the intracranial arteries automatically and finds POIs from the segmented vessels. The second technique identifies POIs directly from the raw, unsegmented image dataset. This latter technique is useful in cases of incomplete segmentation. Following a series of feature calculations, a small fraction of POIs are retained as candidate aneurysms from the collected POIs according to predetermined rules. The CAD scheme was evaluated on 287 datasets containing 147 aneurysms that were verified with digital subtraction angiography, the accepted standard of reference for aneurysm detection. For two different operating points, the CAD scheme achieved a sensitivity of 80% (71% for aneurysms less than 5 mm) with three mean false positives per case, and 95% (91% for aneurysms less than 5 mm) with nine mean false positives per case. In conclusion, the CAD scheme showed good accuracy and may have application in improving the sensitivity of aneurysm detection on MR images.

Distinction between acute disseminated encephalomyelitis and acute multiple sclerosis is often clinically difficult. Perivenous demyelination is the pathological hallmark of acute disseminated encephalomyelitis, whereas confluent demyelination is the hallmark of acute multiple sclerosis. We investigated whether perivenous demyelination versus confluent demyelination distinguishes acute disseminated encephalomyelitis from multiple sclerosis. Patients with perivenous demyelination (n = 13; median age 43 years, range 5–67) on brain biopsy and/or autopsy, ascertained retrospectively, were compared with a cohort with confluent demyelination only (n = 91; 84% multiple sclerosis, 16% isolated syndrome at follow-up; median age 39 years, range 10–69). Clinical presentation, course and the International Paediatric Multiple Sclerosis Study Group clinical criteria for acute disseminated encephalomyelitis were assessed in both cohorts. Among the perivenous demyelination cohort, 10 patients had only perivenous demyelination and three also had confluent demyelination. All but one patient with perivenous demyelination only had a monophasic course, whereas two of three with both types had a relapsing course. The perivenous demyelination cohort was more likely than the confluent demyelination cohort to present with encephalopathy (P < 0.001), depressed level of consciousness (P < 0.001), headache (P < 0.001), meningismus (P = 0.04), cerebrospinal fluid pleocytosis (P = 0.04) or multifocal enhancing magnetic resonance imaging lesions (P < 0.001). A distinct pattern of cortical microglial activation and aggregation without associated cortical demyelination was found among six perivenous demyelination patients, all of whom had encephalopathy and four of whom had depressed level of consciousness. This pattern of cortical pathology was not observed in the confluent demyelination cohort, even in one patient with depressed level of consciousness. Clinical criteria were 80% sensitive and 91% specific for pathologically defined acute disseminated encephalomyelitis (perivenous demyelination), but misdiagnosed acute disseminated encephalomyelitis among 9% of patients with confluent demyelination and multiple sclerosis diagnosis at last follow-up. Perivenous demyelination is associated with meningoencephalopathic presentations and a monophasic course. Depressed level of consciousness is a more specific clinical criterion for pathologically confirmed acute disseminated encephalomyelitis than encephalopathy, which over-diagnosed acute disseminated encephalomyelitis among multiple sclerosis patients. A distinct pattern of cortical microglial activation without cortical demyelination may be the pathological correlate of depressed level of consciousness in acute disseminated encephalomyelitis. Although pathological evidence of perivenous demyelination may be superior to clinical criteria for diagnosing acute disseminated encephalomyelitis, the co-occurrence of perivenous and confluent demyelination in some individuals suggests pathogenic overlap between acute disseminated encephalomyelitis and multiple sclerosis and misclassification even with biopsy.

The purpose of this study was to develop a method to measure brain and white matter hyperintensity (leukoaraiosis) volume that is based on the segmentation of the intensity histogram of fluid attenuated inversion recovery (FLAIR) images, and to assess the accuracy and reproducibility of the method. Whole head synthetic image phantoms with manually introduced leukoaraiosis lesions of varying severity were constructed. These synthetic image phantom sets incorporated image contrast and anatomic features which mimicked leukoaraiosis found in real life. One set of synthetic image phantoms was used to develop the segmentation algorithm (FLAIR-histoseg). A second set was used to measure its accuracy. Test re-test reproducibility was assessed in 10 elderly volunteers who were imaged twice. The mean absolute error of the FLAIR-histoseg method for measurement of leukoaraiosis volume was 6.6% and for brain volume 1.4%. The mean test re-test coefficient of variation for leukoaraiosis volume was 1.4% and for brain volume was 0.3%. We conclude that the FLAIR-histoseg method is an accurate and reproducible method for measuring leukoaraiosis and whole brain volume in elderly subjects.

The goal of this study was to determine the compression level at which 3D JPEG2000 compression of thin-slice CTs of the chest and abdomen–pelvis becomes visually perceptible. A secondary goal was to determine if residents in training and non-physicians are substantially different from experienced radiologists in their perception of compression-related changes. This study used multidetector computed tomography 3D datasets with 0.625–1-mm thickness slices of standard chest, abdomen, or pelvis, clipped to 12 bits. The Kakadu v5.2 JPEG2000 compression algorithm was used to compress and decompress the 80 examinations creating four sets of images: lossless, 1.5 bpp (8:1), 1 bpp (12:1), and 0.75 bpp (16:1). Two randomly selected slices from each examination were shown to observers using a flicker mode paradigm in which observers rapidly toggled between two images, the original and a compressed version, with the task of deciding whether differences between them could be detected. Six staff radiologists, four residents, and six PhDs experienced in medical imaging (from three institutions) served as observers. Overall, 77.46% of observers detected differences at 8:1, 94.75% at 12:1, and 98.59% at 16:1 compression levels. Across all compression levels, the staff radiologists noted differences 64.70% of the time, the resident’s detected differences 71.91% of the time, and the PhDs detected differences 69.95% of the time. Even mild compression is perceptible with current technology. The ability to detect differences does not equate to diagnostic differences, although perception of compression artifacts could affect diagnostic decision making and diagnostic workflow.

The goal of this study was to determine the compression level at which 3D JPEG2000 compression of thin-slice CTs of the chest and abdomen–pelvis becomes visually perceptible. A secondary goal was to determine if residents in training and non-physicians are substantially different from experienced radiologists in their perception of compression-related changes. This study used multidetector computed tomography 3D datasets with 0.625–1-mm thickness slices of standard chest, abdomen, or pelvis, clipped to 12 bits. The Kakadu v5.2 JPEG2000 compression algorithm was used to compress and decompress the 80 examinations creating four sets of images: lossless, 1.5 bpp (8:1), 1 bpp (12:1), and 0.75 bpp (16:1). Two randomly selected slices from each examination were shown to observers using a flicker mode paradigm in which observers rapidly toggled between two images, the original and a compressed version, with the task of deciding whether differences between them could be detected. Six staff radiologists, four residents, and six PhDs experienced in medical imaging (from three institutions) served as observers. Overall, 77.46% of observers detected differences at 8:1, 94.75% at 12:1, and 98.59% at 16:1 compression levels. Across all compression levels, the staff radiologists noted differences 64.70% of the time, the resident’s detected differences 71.91% of the time, and the PhDs detected differences 69.95% of the time. Even mild compression is perceptible with current technology. The ability to detect differences does not equate to diagnostic differences, although perception of compression artifacts could affect diagnostic decision making and diagnostic workflow.

Information in electronic medical records is often in an unstructured free-text format. This format presents challenges for expedient data retrieval and may fail to convey important findings. Natural language processing (NLP) is an emerging technique for rapid and efficient clinical data retrieval. While proven in disease detection, the utility of NLP in discerning disease progression from free-text reports is untested. We aimed to (1) assess whether unstructured radiology reports contained sufficient information for tumor status classification; (2) develop an NLP-based data extraction tool to determine tumor status from unstructured reports; and (3) compare NLP and human tumor status classification outcomes. Consecutive follow-up brain tumor magnetic resonance imaging reports (2000–­2007) from a tertiary center were manually annotated using consensus guidelines on tumor status. Reports were randomized to NLP training (70%) or testing (30%) groups. The NLP tool utilized a support vector machines model with statistical and rule-based outcomes. Most reports had sufficient information for tumor status classification, although 0.8% did not describe status despite reference to prior examinations. Tumor size was unreported in 68.7% of documents, while 50.3% lacked data on change magnitude when there was detectable progression or regression. Using retrospective human classification as the gold standard, NLP achieved 80.6% sensitivity and 91.6% specificity for tumor status determination (mean positive predictive value, 82.4%; negative predictive value, 92.0%). In conclusion, most reports contained sufficient information for tumor status determination, though variable features were used to describe status. NLP demonstrated good accuracy for tumor status classification and may have novel application for automated disease status classification from electronic databases.

Information in electronic medical records is often in an unstructured free-text format. This format presents challenges for expedient data retrieval and may fail to convey important findings. Natural language processing (NLP) is an emerging technique for rapid and efficient clinical data retrieval. While proven in disease detection, the utility of NLP in discerning disease progression from free-text reports is untested. We aimed to (1) assess whether unstructured radiology reports contained sufficient information for tumor status classification; (2) develop an NLP-based data extraction tool to determine tumor status from unstructured reports; and (3) compare NLP and human tumor status classification outcomes. Consecutive follow-up brain tumor magnetic resonance imaging reports (2000–­2007) from a tertiary center were manually annotated using consensus guidelines on tumor status. Reports were randomized to NLP training (70%) or testing (30%) groups. The NLP tool utilized a support vector machines model with statistical and rule-based outcomes. Most reports had sufficient information for tumor status classification, although 0.8% did not describe status despite reference to prior examinations. Tumor size was unreported in 68.7% of documents, while 50.3% lacked data on change magnitude when there was detectable progression or regression. Using retrospective human classification as the gold standard, NLP achieved 80.6% sensitivity and 91.6% specificity for tumor status determination (mean positive predictive value, 82.4%; negative predictive value, 92.0%). In conclusion, most reports contained sufficient information for tumor status determination, though variable features were used to describe status. NLP demonstrated good accuracy for tumor status classification and may have novel application for automated disease status classification from electronic databases.

In this study, we present preliminary data on the effect of automated 3D image alignment on the time to arrive at a decision about an imaging finding, the agreement of multiple of multiple observers, the prevalence of comparison examinations, and technical success rates for the image alignment algorithm. We found that automated image alignment reduced the average time to make a decision by 25% for cases where the structures are rigid, and when the scanning protocol is similar. For cases where these are not true, there is little or no benefit. In our practice, 54% of cases had prior examinations that could be automatically aligned. The overall benefit seen in our department for highly similar exams might be 20% for neuro and 10% for body; the benefit seen in other practices is likely to vary based on scanning practices and prevalence of prior examinations.

The Cancer Bioinformatics Grid (caBIG™) program was created by the National Cancer Institute to facilitate sharing of IT infrastructure, data, and applications among the National Cancer Institute-sponsored cancer research centers. The program was launched in February 2004 and now links more than 50 cancer centers. In April 2005, the In Vivo Imaging Workspace was added to promote the use of imaging in cancer clinical trials. At the inaugural meeting, four special interest groups (SIGs) were established. The Software SIG was charged with identifying projects that focus on open-source software for image visualization and analysis. To date, two projects have been defined by the Software SIG. The eXtensible Imaging Platform project has produced a rapid application development environment that researchers may use to create targeted workflows customized for specific research projects. The Algorithm Validation Tools project will provide a set of tools and data structures that will be used to capture measurement information and associated needed to allow a gold standard to be defined for the given database against which change analysis algorithms can be tested. Through these and future efforts, the caBIG™ In Vivo Imaging Workspace Software SIG endeavors to advance imaging informatics and provide new open-source software tools to advance cancer research.

The Cancer Bioinformatics Grid (caBIG™) program was created by the National Cancer Institute to facilitate sharing of IT infrastructure, data, and applications among the National Cancer Institute-sponsored cancer research centers. The program was launched in February 2004 and now links more than 50 cancer centers. In April 2005, the In Vivo Imaging Workspace was added to promote the use of imaging in cancer clinical trials. At the inaugural meeting, four special interest groups (SIGs) were established. The Software SIG was charged with identifying projects that focus on open-source software for image visualization and analysis. To date, two projects have been defined by the Software SIG. The eXtensible Imaging Platform project has produced a rapid application development environment that researchers may use to create targeted workflows customized for specific research projects. The Algorithm Validation Tools project will provide a set of tools and data structures that will be used to capture measurement information and associated needed to allow a gold standard to be defined for the given database against which change analysis algorithms can be tested. Through these and future efforts, the caBIG™ In Vivo Imaging Workspace Software SIG endeavors to advance imaging informatics and provide new open-source software tools to advance cancer research.

Modern imaging systems are able to produce a rich and diverse array of information, regarding various facets of anatomy and function. The quantity of information produced by these systems is so bountiful, however, as to have the potential to become a hindrance to clinical assessment. In the context of serial image evaluation, computer-based change detection and characterization is one important mechanism to process the information produced by imaging systems, so as to reduce the quantity of data, direct the attention of the physician to regions of the data which are the most informative for their purposes, and present the data in the form in which it will be the most useful. Change detection and characterization algorithms may serve as a basis for the creation of an objective definition of progression, which will reduce inter and intra-observer variability, and facilitate earlier detection of disease and recurrence, which in turn may lead to improved outcomes. Decreased observer variability combined with increased acuity should make it easier to discover promising therapies. Quantitative measures of the response to these therapies should provide a means to compare the effectiveness of treatments under investigation. Change detection may be applicable to a broad range of cancers, in essentially all anatomical regions. The source of information upon which change detection comparisons may be based is likewise broad. Validation of algorithms for the longitudinal assessment of cancer patients is expected to be challenging, though not insurmountable, as the many facets of the problem mean that validation will likely need to be approached from a variety of vantage points. Change detection and characterization is quickly becoming a very active field of investigation, and it is expected that this burgeoning field will help to facilitate cancer care both in the clinic and research.

Conclusion

The requirements for a clinical viewing station have some similarities to requirements of an interpretation workstation, but have some important differences. The user base is a heterogeneous group, and the tasks the clinicians perform vary substantially, increasing the variation in the requirements. However, it is possible to determine a list of requirements that satisfy most clinicians under most circumstances. In addition to the standard image manipulation tools, we have found that integration with the other clinical tools is essential to maintaining or increasing their efficiency. Providing for varying levels of functionality/complexity is also useful, because of the variation in the user base.

This report describes several image archival problems facing the authors' department and the results of their attempt to define the requirements for an enterprise digital image archive. The problems identified include the costs of supporting multiple distinct archives, the increased complexity of supporting multiple archive interfaces, the differences in data handling policies and resulting variations in data integrity, and variability in support for nonimage data. The authors also describe the data collected including image volumes and trends and imaging device trends. Finally, the resulting specification for an enterprise digital image archive, including storage and retrieval performance and interface requirements are presented.

A clinical viewing system was integrated with the Mayo Clinic Scottsdale picture archiving and communication system (PACS) for providing images and the report as part of the electronic medical record (EMR). Key attributes of the viewer include a single user log-on, an integrated patient centric EMR image access for all ordered examinations, prefetching of the most recent prior examination of the same modality, and the ability to provide comparison of current and past exams at the same time on the display. Other functions included preset windows, measurement tools, and multiformat display. Images for the prior 12 months are stored on the clinical server and are viewable in less than a second. Images available on the desktop include all computed radiography (CR), chest, magnetic resonance images (MRI), computed tomography (CT), ultrasound (U/S), nuclear, angiographic, gastrointestinal (GI) digital spots, and portable C-arm digital spots. Ad hoc queries of examinations from PACS are possible for those patients whose image may not be on the clinical server, but whose images reside on the PACS archive (10TB). Clinician satisfaction was reported to be high, especially for those staff heavily dependent on timely access to images, as well as those having heavy film usage. The desktop viewer is used for resident access to images. It is also useful for teaching conferences with large-screen projection without film. We report on the measurements of functionality, reliability, and speed of image display with this application.

The purpose of the study was to evaluate the effects of lossy compression on grayscale ultrasound images to determine how much compression can be applied while still maintaining images that are acceptable for diagnostic purposes. The study considered how the acquisition technique (video frame-grabber versus directly acquired in digital form) influences how much compression can be applied. For directly acquired digital images, the study considered how text (that is burned into the image) affects the compressibility of the image. The lossy compression techniques that were considered include JPEG and a Wavelet algorithm using set partitioning in hierarchical trees (SPIHT).

Although it is intuitively valuable that more expedient delivery of radiographic images and reports to clinicians would improve patient care, it is important to document these outcomes to validate further advances in these areas. We evaluated the care of 215 patients seen at a walk-in clinic to determine what benefit digital imaging is to the patient. Cohorts consisted of all patients for whom specified radiology examinations were ordered during a 7-day period. The first cohort was recruited when analog films were used. The second cohort received examinations performed with computed radiography (CR) acquisition and computer display, which had been in use for 2 years. Patients were categorized as to the type of study they received, as well as whether a staff radiologist was immediately available to read the study. Clinical behavior was characterized by outcome measures of time to final diagnosis, time to final treatment, and need for follow-up. Our analysis demonstrated a reduction in time to final diagnosis that was better appreciated during the times when a staff radiologist was not immediately available. It also suggested that greater time reductions were seen for patients who received extremity examinations than those who received chest, sinus, or rib films. These data suggest that digital imaging is a useful tool to improve clinical outcome of patients seen in the acute care setting.

We prospectively compared image and report delivery times in our Urgent Care Center (UCC) during a film-based practice (1995) and after complete implementation of an electronic imaging practice in 1997. Before switching to a totally electronic and filmless practice, multiple time periods were consistently measured during a 1-week period in May 1995 and then again in a similar week in May 1997 after implementation of electronic imaging. All practice patterns were the same except for a film-based practice in 1995 versus a filmless practice in 1997. The following times were measured: (1) waiting room time, (2) technologist’s time of examination, (3) time to quality control, (4) radiology interpretation times, (5) radiology image and report delivery time, (6) total radiology turn-around time, (7) time to room the patient back in the UCC, and (8) time until the ordering physician views the film. Waiting room time was longer in 1997 (average time, 26∶47) versus 1995 (average time, 15∶54). The technologist’s examination completion time was approximately the same (1995 average time, 06∶12; 1997 average time, 05∶41). There was also a slight increase in the time of the technologist’s electronic verification or quality control in 1997 (average time, 7∶17) versus the film-based practice in 1995 (average time, 2∶35). However, radiology interpretation times dramatically improved (average time, 49∶38 in 1995 versus average time 13∶50 in 1997). There was also a decrease in image delivery times to the clinicians in 1997 (median, 53 minutes) versus the film based practice of 1995 (1 hour and 40 minutes). Reports were available with the images immediately upon completion by the radiologist in 1997, compared with a median time of 27 minutes in 1995. Importantly, patients were roomed back into the UCC examination rooms faster after the radiologic procedure in 1997 (average time, 13∶36) than they were in 1995 (29∶38). Finally, the ordering physicians viewed the diagnostic images and reports in dramatically less time in 1997 (median, 26 minutes) versus 1995 (median, 1 hour and 5 minutes). In conclusion, a filmless electronic imaging practice within our UCC greatly improved radiology image and report delivery times, as well as improved clinical efficiency.

An efficient environment for picture archiving and communications systems (PACS) in the radiology department and throughout a medical practice requires good hardware, good software, and integration of the information sources that exist in a radiology department and institution. This tutorial will describe some of the considerations in evaluating a PACS workstation, with a view to the hardware requirements, user interface designs and integration with the information systems.