Improved access to DICOM studies to both physicians and patients is changing the ways medical imaging studies are
visualized and interpreted beyond the confines of radiologists' PACS workstations. While radiologists are trained for
viewing and image interpretation, a non-radiologist physician relies on the radiologists' reports. Consequently, patients
historically have been typically informed about their imaging findings via oral communication with their physicians,
even though clinical studies have shown that patients respond to physician's advice significantly better when the
individual patients are shown their own actual data. Our previous work on automated semantic annotation of DICOM
Computed Tomography (CT) images allows us to further link radiology report with the corresponding images, enabling
us to bridge the gap between image data with the human interpreted textual description of the corresponding imaging
studies. The mapping of radiology text is facilitated by natural language processing (NLP) based search application.
When combined with our automated semantic annotation of images, it enables navigation in large DICOM studies by
clicking hyperlinked text in the radiology reports. An added advantage of using semantic annotation is the ability to
render the organs to their default window level setting thus eliminating another barrier to image sharing and distribution.
We believe such approaches would potentially enable the consumer to have access to their imaging data and navigate
them in an informed manner.
Existing literature describes a variety of techniques for semantic annotation of DICOM CT images, i.e. the
automatic detection and localization of anatomical structures. Semantic annotation facilitates enhanced image
navigation, linkage of DICOM image content and non-image clinical data, content-based image retrieval, and
image registration. A key challenge for semantic annotation algorithms is inter-patient variability. However,
while the algorithms described in published literature have been shown to cope adequately with the variability in
test sets comprising adult CT scans, the problem presented by the even greater variability in pediatric anatomy
has received very little attention. Most existing semantic annotation algorithms can only be extended to work
on scans of both adult and pediatric patients by adapting parameters heuristically in light of patient size. In
contrast, our approach, which uses random regression forests ('RRF'), learns an implicit model of scale variation
automatically using training data. In consequence, anatomical structures can be localized accurately in both
adult and pediatric CT studies without the need for parameter adaptation or additional information about
patient scale. We show how the RRF algorithm is able to learn scale invariance from a combined training set
containing a mixture of pediatric and adult scans. Resulting localization accuracy for both adult and pediatric
data remains comparable with that obtained using RRFs trained and tested using only adult data.
Global linear registration is a necessary first step for many different tasks in medical image
analysis. Comparing longitudinal studies1, cross-modality fusion2, and many other
applications depend heavily on the success of the automatic registration. The robustness and
efficiency of this step is crucial as it affects all subsequent operations. Most common
techniques cast the linear registration problem as the minimization of a global energy
function based on the image intensities. Although these algorithms have proved useful, their
robustness in fully automated scenarios is still an open question. In fact, the optimization step
often gets caught in local minima yielding unsatisfactory results. Recent algorithms constrain
the space of registration parameters by exploiting implicit or explicit organ segmentations,
thus increasing robustness4,5. In this work we propose a novel robust algorithm for automatic
global linear image registration. Our method uses random regression forests to estimate
posterior probability distributions for the locations of anatomical structures - represented as
axis aligned bounding boxes6. These posterior distributions are later integrated in a global
linear registration algorithm. The biggest advantage of our algorithm is that it does not
require pre-defined segmentations or regions. Yet it yields robust registration results. We
compare the robustness of our algorithm with that of the state of the art Elastix toolbox7.
Validation is performed via 1464 pair-wise registrations in a database of very diverse 3D CT
images. We show that our method decreases the "failure" rate of the global linear registration
from 12.5% (Elastix) to only 1.9%.
In the current health-care environment, the time available for physicians to browse patients' scans is shrinking due
to the rapid increase in the sheer number of images. This is further aggravated by mounting pressure to become
more productive in the face of decreasing reimbursement. Hence, there is an urgent need to deliver technology
which enables faster and effortless navigation through sub-volume image visualizations. Annotating image regions
with semantic labels such as those derived from the RADLEX ontology can vastly enhance image navigation
and sub-volume visualization. This paper uses random regression forests for efficient, automatic detection and
localization of anatomical structures within DICOM 3D CT scans. A regression forest is a collection of decision
trees which are trained to achieve direct mapping from voxels to organ location and size in a single pass. This
paper focuses on comparing automated labeling with expert-annotated ground-truth results on a database of 50
highly variable CT scans. Initial investigations show that regression forest derived localization errors are smaller
and more robust than those achieved by state-of-the-art global registration approaches. The simplicity of the
algorithm's context-rich visual features yield typical runtimes of less than 10 seconds for a 5123 voxel DICOM
CT series on a single-threaded, single-core machine running multiple trees; each tree taking less than a second.
Furthermore, qualitative evaluation demonstrates that using the detected organs' locations as index into the
image volume improves the efficiency of the navigational workflow in all the CT studies.
Recently, there is an increasing need to share medical images for research purpose. In order to respect and preserve
patient privacy, most of the medical images are de-identified with protected health information (PHI) before
research sharing. Since manual de-identification is time-consuming and tedious, so an automatic de-identification
system is necessary and helpful for the doctors to remove text from medical images. A lot of papers have been
written about algorithms of text detection and extraction, however, little has been applied to de-identification of
medical images. Since the de-identification system is designed for end-users, it should be effective, accurate and
fast. This paper proposes an automatic system to detect and extract text from medical images for de-identification
purposes, while keeping the anatomic structures intact. First, considering the text have a remarkable contrast with
the background, a region variance based algorithm is used to detect the text regions. In post processing, geometric
constraints are applied to the detected text regions to eliminate over-segmentation, e.g., lines and anatomic
structures. After that, a region based level set method is used to extract text from the detected text regions. A GUI
for the prototype application of the text detection and extraction system is implemented, which shows that our method can detect most of the text in the images. Experimental results validate that our method can detect and extract text in medical images with a 99% recall rate. Future research of this system includes algorithm improvement, performance evaluation, and computation optimization.
Knee-related injuries including meniscal tears are common in both young athletes and the aging population and require
accurate diagnosis and surgical intervention when appropriate. With proper techniques and radiologists' experienced
skills, confidence in detection of meniscal tears can be quite high. However, for radiologists without musculoskeletal
training, diagnosis of meniscal tears can be challenging. This paper develops a novel computer-aided detection (CAD)
diagnostic system for automatic detection of meniscal tears in the knee. Evaluation of this CAD system using an
archived database of images from 40 individuals with suspected knee injuries indicates that the sensitivity and
specificity of the proposed CAD system are 83.87% and 75.19%, respectively, compared to the mean sensitivity and
specificity of 77.41% and 81.39%, respectively obtained by experienced radiologists in routine diagnosis without using
the CAD. The experimental results suggest that the developed CAD system has great potential and promise in automatic
detection of both simple and complex meniscal tears of knees.
Osteoarthritis (OA) is the most common form of arthritis and a major cause of morbidity affecting millions of adults in
the US and world wide. In the knee, OA begins with the degeneration of joint articular cartilage, eventually resulting in
the femur and tibia coming in contact, and leading to severe pain and stiffness. There has been extensive research
examining 3D MR imaging sequences and automatic/semi-automatic techniques for 2D/3D articular cartilage
extraction. However, in routine clinical practice the most popular technique still remain radiographic examination and
qualitative assessment of the joint space. This may be in large part because of a lack of tools that can provide clinically
relevant diagnosis in adjunct (in near real time fashion) with the radiologist and which can serve the needs of the
radiologists and reduce inter-observer variation. Our work aims to fill this void by developing a CAD application that
can generate clinically relevant diagnosis of the articular cartilage damage in near real time fashion. The algorithm
features a 2D Active Shape Model (ASM) for modeling the bone-cartilage interface on all the slices of a Double Echo
Steady State (DESS) MR sequence, followed by measurement of the cartilage thickness from the surface of the bone,
and finally by the identification of regions of abnormal thinness and focal/degenerative lesions. A preliminary
evaluation of CAD tool was carried out on 10 cases taken from the Osteoarthritis Initiative (OAI) database. When
compared with 2 board-certified musculoskeletal radiologists, the automatic CAD application was able to get
segmentation/thickness maps in little over 60 seconds for all of the cases. This observation poses interesting
possibilities for increasing radiologist productivity and confidence, improving patient outcomes, and applying more
sophisticated CAD algorithms to routine orthopedic imaging tasks.
Knee-related injuries involving the meniscal or articular cartilage are common and require accurate diagnosis and
surgical intervention when appropriate. With proper techniques and experience, confidence in detection of meniscal
tears and articular cartilage abnormalities can be quite high. However, for radiologists without musculoskeletal training,
diagnosis of such abnormalities can be challenging. In this paper, the potential of improving diagnosis through
integration of computer-aided detection (CAD) algorithms for automatic detection of meniscal tears and articular
cartilage injuries of the knees is studied. An integrated approach in which the results of algorithms evaluating either
meniscal tears or articular cartilage injuries provide feedback to each other is believed to improve the diagnostic
accuracy of the individual CAD algorithms due to the known association between abnormalities in these distinct
anatomic structures. The correlation between meniscal tears and articular cartilage injuries is exploited to improve the
final diagnostic results of the individual algorithms. Preliminary results from the integrated application are encouraging
and more comprehensive tests are being planned.
Knee-related injuries, including meniscal tears, are common in young athletes and require accurate diagnosis and
appropriate surgical intervention. Although with proper technique and skill, confidence in the detection of meniscal
tears should be high, this task continues to be a challenge for many inexperienced radiologists. The purpose of our study
was to automate detection of meniscal tears of the knee using a computer-aided detection (CAD) algorithm. Automated
segmentation of the sagittal T1-weighted MR imaging sequences of the knee in 28 patients with diagnoses of meniscal
tears was performed using morphologic image processing in a 3-step process including cropping, thresholding, and
application of morphological constraints. After meniscal segmentation, abnormal linear meniscal signal was extracted
through a second thresholding process. The results of this process were validated by comparison with the interpretations
of 2 board-certified musculoskeletal radiologists. The automated meniscal extraction algorithm process was able to
successfully perform region of interest selection, thresholding, and object shape constraint tasks to produce a convex
image isolating the menisci in more than 69% of the 28 cases. A high correlation was also noted between the CAD
algorithm and human observer results in identification of complex meniscal tears. Our initial investigation indicates
considerable promise for automatic detection of simple and complex meniscal tears of the knee using the CAD
algorithm. This observation poses interesting possibilities for increasing radiologist productivity and confidence,
improving patient outcomes, and applying more sophisticated CAD algorithms to orthopedic imaging tasks.
KEYWORDS: Digital photography, Photography, 3D image processing, 3D image reconstruction, Computed tomography, Diagnostics, Software, Statistical analysis, Radiology, Facial recognition systems
3D and multi-planar reconstruction of CT images have become indispensable in the routine practice of diagnostic
imaging. These tools cannot only enhance our ability to diagnose diseases, but can also assist in therapeutic planning as
well. The technology utilized to create these can also render surface reconstructions, which may have the undesired
potential of providing sufficient detail to allow recognition of facial features and consequently patient identity, leading
to violation of patient privacy rights as described in the HIPAA (Health Insurance Portability and Accountability Act)
legislation. The purpose of this study is to evaluate whether 3D reconstructed images of a patient's facial features can
indeed be used to reliably or confidently identify that specific patient. Surface reconstructed images of the study
participants were created used as candidates for matching with digital photographs of participants. Data analysis was
performed to determine the ability of observers to successfully match 3D surface reconstructed images of the face with
facial photographs. The amount of time required to perform the match was recorded as well. We also plan to
investigate the ability of digital masks or physical drapes to conceal patient identity. The recently expressed concerns
over the inability to truly "anonymize" CT (and MRI) studies of the head/face/brain are yet to be tested in a prospective
study. We believe that it is important to establish whether these reconstructed images are a "threat" to patient
privacy/security and if so, whether minimal interventions from a clinical perspective can substantially reduce this
possibility.
Over the past decade, several computerized tools have been developed for detection of lung nodules and for providing
volumetric analysis. Incidentally detected lung nodules have traditionally been followed over time by measurements of
their axial dimensions on CT scans to ensure stability or document progression. A recently published article by the
Fleischner Society offers guidelines on the management of incidentally detected nodules based on size criteria. For this
reason, differences in measurements obtained by automated tools from various vendors may have significant
implications on management, yet the degree of variability in these measurements is not well understood. The goal of this
study is to quantify the differences in nodule maximum diameter and volume among different automated analysis
software. Using a dataset of lung scans obtained with both "ultra-low" and conventional doses, we identified a subset of
nodules in each of five size-based categories. Using automated analysis tools provided by three different vendors, we
obtained size and volumetric measurements on these nodules, and compared these data using descriptive as well as
ANOVA and t-test analysis. Results showed significant differences in nodule maximum diameter measurements among
the various automated lung nodule analysis tools but no significant differences in nodule volume measurements. These
data suggest that when using automated commercial software, volume measurements may be a more reliable marker of
tumor progression than maximum diameter. The data also suggest that volumetric nodule measurements may be
relatively reproducible among various commercial workstations, in contrast to the variability documented when
performing human mark-ups, as is seen in the LIDC (lung imaging database consortium) study.
During the last decade, x-ray computed tomography (CT) has been applied to screen large asymptomatic smoking and nonsmoking populations for early lung cancer detection. Because a larger population will be involved in such screening exams, more and more attention has been paid to studying low-dose, even ultra-low-dose x-ray CT. However, reducing CT radiation exposure will increase noise level in the sinogram, thereby degrading the quality of reconstructed CT images as well as causing more streak artifacts near the apices of the lung. Thus, how to reduce the noise levels and streak artifacts in the low-dose CT images is becoming a meaningful topic.
Since multi-slice helical CT has replaced conventional stop-and-shoot CT in many clinical applications, this research mainly focused on the noise reduction issue in multi-slice helical CT. The experiment data were provided by Siemens SOMATOM Sensation 16-Slice helical CT. It included both conventional CT data acquired under 120 kvp voltage and 119 mA current and ultra-low-dose CT data acquired under 120 kvp and 10 mA protocols. All other settings are the same as that of conventional CT. In this paper, a nonparametric smoothing method with thin plate smoothing splines and the roughness penalty was proposed to restore the ultra-low-dose CT raw data. Each projection frame was firstly divided into blocks, and then the 2D data in each block was fitted to a thin-plate smoothing splines' surface via minimizing a roughness-penalized least squares objective function. By doing so, the noise in each ultra-low-dose CT projection was reduced by leveraging the information contained not only within each individual projection profile, but also among nearby profiles. Finally the restored ultra-low-dose projection data were fed into standard filtered back projection (FBP) algorithm to reconstruct CT images. The rebuilt results as well as the comparison between proposed approach and traditional method were given in the results and discussions section, and showed effectiveness of proposed thin-plate based nonparametric regression method.
Monochrome monitors typically display 8 bits of data (256 shades of gray) at one time. This study determined if monitors that can display a wider range of grayscale information (11-bit) can improve observer performance and decrease the use of window/level in detecting pulmonary nodules. Three sites participated using 8 and 11-bit displays from three manufacturers. At each site, six radiologists reviewed 100 DR chest images on both displays. There was no significant difference in ROC Az (F = 0.0374, p = 0.8491) as a function of 8 vs 11 bit-depth. Average Az across all observers with 8-bits was 0.8284 and with 11-bits was 0.8253. There was a significant difference in overall viewing time (F = 10.209, p = 0.0014) favoring the 11-bit displays. Window/level use did not differ significantly for the two types of displays. Eye position recording on a subset of images at one site showed that cumulative dwell times for each decision category were lower with the 11-bit than with the 8-bit display. T-tests for paired observations showed that the TP (t = 1.452, p = 0.1507), FN (t = 0.050, p = 0.9609) and FP (t = 0.042, p = 0.9676) were not statistically significant. The difference for the TN decisions was statistically significant (t = 1.926, p = 0.05). 8-bit displays will not impact negatively diagnostic accuracy, but using 11-bit displays may improve workflow efficiency.
We propose a computationally efficient and effective analysis technique to classify X-Ray Computed Tomography (CT) images into four anatomic regions: neck, chest, abdomen, and pelvis. The proposed technique divides a single scan (performed with a single bolus of contrast) into multiple anatomic regions that can be stored in separate electronic folders for each region. Our CT analysis technique extracts relevant features from the image slices and classifies the images into the four anatomic regions using a multilayer perceptron network. The technique is tested on a number of CT images and shown to result in an acceptable level of classification performance.
The widespread use of multi-detector CT scanners has been associated with a remarkable increase in the number of CT slices as well as a substantial decrease in the average thickness of individual slices. This increased number of thinner slices has created a marked increase in archival and network bandwidth requirements associated with storage and transmission of these studies. We demonstrate that although compression can be used to decrease the size of these image files, thinner CT slices are less compressible than thicker slices when measured by either a visual discrimination model (VDM) or the more traditional peak signal to noise ratio. The former technique (VDM) suggests that the discrepancy in compressibility between thin and thick slices becomes greater at greater compression levels while the latter technique (PSNR), suggests that this is not the case. Previous studies that we and others have performed suggest that the VDM model probably corresponds more closely with human observers than does the PSNR model. Additionally we demonstrated that the poor relative compressibility of thin sections can be substantially negated by the use of JPEG 2000 3D compression which yields superior image quality at a given level of compression in comparison with 2D compression. Additionally, thin and thick sections are approximately equally compressible for 3D compression with little change with increasing levels of compression.
The relatively low (20%-25%) sensitivity of conventional radiography for lung nodules is an impetus for investigations into computer-assisted diagnostic (CAD) algorithms and into alternative acquisition techniques (such as dual-energy subtraction [DES]), both of which have been shown to increase diagnostic sensitivity for lung nodule detection. This pilot study combined these synergistic techniques in the diagnosis of digital clinical chest radiographs in 26 individuals. A total of 59 marks were identified by the CAD algorithm as suspicious for a nodule using a "conventional" chest direct radiography posterior/anterior image (an average of 2.3 marks per radiograph). Only 39 marks were identified on the soft tissue image of the corresponding DES radiographs (an average of 1.5 marks per radiograph). The sensitivity for nodules considered subtle but "actionable" in the 10-15-mm range was 0% (correctly identifying 0 of 4 nodules), whereas the sensitivity for the same radiographs with DES was 75% (correctly identifying 3 of 4 nodules). These pilot data suggest that the algorithms for at least one commercial CAD system may not be fully able to differentiate overlying bones and other calcifications from pulmonary lesions (which is also a difficult task for radiologists) and that the combination of CAD and DES acquisition may result in a substantial improvement in both sensitivity and specificity in the detection of relatively subtle lung nodules. This study has been expanded to evaluate a much larger set of images to further investigate the potential for the routine use of CAD with DES.
KEYWORDS: Received signal strength, Imaging informatics, Databases, Radio propagation, Medical imaging, Document management, Picture Archiving and Communication System, Internet, Image processing, Information science
There are over 40 open source projects in the field of radiology informatics. Because these are organized and written by volunteers, the development speed varies greatly from one project to the next. To keep track of updates, users must constantly check in on each project's Web page. Many projects remain dormant for years, and ad hoc checking becomes both an inefficient and unreliable means of determining when new versions are available. The result is that most end users track only a few projects and are unaware when others that may be more germane to their interests leapfrog in development. RSS feeds provide a machine readable XML format to track software project updates. Currently only 8 of the 40 projects provide RSS feeds for automatic propagation of news updates. We have a built a news aggregation engine around open source projects in radiology informatics.
The JPEG2000 compression standard is increasingly a preferred industry method for 2D image compression. Some vendors, however, continue to use proprietary discrete cosine transform (DCT) JPEG encoding. This study compares image quality in terms of just-noticeable differences (JNDs) and peak signal-to-noise ratios (PSNR) between DCT JPEG encoding and JPEG2000 encoding. Four computed tomography and 6 computed radiography studies were compressed using a proprietary DCT JPEG encoder and JPEG2000 standard compression. Image quality was measured in JNDs and PSNRs. The JNDmetrix computational visual discrimination model simulates known physiological mechanisms in the human visual system, including the luminance and contrast sensitivity of the eye and spatial frequency and orientation responses of the visual cortex. Higher JND values indicate that a human observer would be more likely to notice a significant difference between compared images. DCT JPEG compression showed consistently lower image distortions at lower compression ratios, whereas JPEG2000 compression showed benefit at higher compression ratios (>50:1). The crossover occurred at ratios that varied among the images. The magnitude of any advantage of DCT compression at low ratios was often small. Interestingly, this advantage of DCT JPEG compression at lower ratios was generally not observed when image quality was measured in PSNRs. These results suggest that DCT JPEG may outperform JPEG2000 for compression ratios generally used in medical imaging and that the differences between DCT and JPEG2000 could be visible to observers and thus clinically significant.
The authors identify a fundamental disconnect between the ways in which industry and radiologists assess and even discuss product performance. What is needed is a quantitative methodology that can assess both subjective image quality and observer task performance. In this study, we propose and evaluate the use of a visual discrimination model (VDM) that assesses just-noticeable differences (JNDs) to serve this purpose. The study compares radiologists' subjective perceptions of image quality of computer tomography (CT) and computed radiography (CR) images with quantitative measures of peak signal-to-noise ratio (PSNR) and JNDs as measured by a VDM. The study included 4 CT and 6 CR studies with compression ratios ranging from lossless to 90:1 (total of 80 sets of images were generated [n = 1,200]). Eleven radiologists reviewed the images and rated them in terms of overall quality and readability and identified images not acceptable for interpretation. Normalized reader scores were correlated with compression, objective PSNR, and mean JND values. Results indicated a significantly higher correlation between observer performance and JND values than with PSNR methods. These results support the use of the VDM as a metric not only for the threshold discriminations for which it was calibrated, but also as a general image quality metric. This VDM is a highly promising, reproducible, and reliable adjunct or even alternative to human observer studies for research or to establish clinical guidelines for image compression, dose reductions, and evaluation of various display technologies.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.