Automating Perforator Flap MRA and CTA Reporting.

Surgical breast reconstruction after mastectomy requires precise perforator coordinates/dimensions, perforator course, and fat volume in a radiology report. Automatic perforator reporting software was implemented as an OsiriX Digital Imaging and Communications in Medicine (DICOM) viewer plugin. For perforator analysis, the user identifies a reference point (e.g., umbilicus) and marks each perforating artery/vein bundle with multiple region of interest (ROI) points along its course beginning at the muscle-fat interface. Computations using these points and analysis of image data produce content for the report. Post-processing times were compared against conventional/manual methods using de-identified images of 26 patients with surgically confirmed accuracy of perforator locations and caliber. The time from loading source images to completion of report was measured. Significance of differences in mean processing times for this automated approach versus the conventional/manual approach was assessed using a paired t test. The mean conventional reporting time for our radiologists was 76 ± 27 min (median 65 min) compared with 25 ± 6 min (median 25 min) using our OsiriX plugin (p < 0.01). The conventional approach had three reports with transcription errors compared to none with the OsiriX plugin. Otherwise, the reports were similar. In conclusion, automated reporting of perforator magnetic resonance angiography (MRA) studies is faster compared with the standard, manual approach, and transcription errors which are eliminated.

A fast Edge-preserving Bayesian reconstruction method for Parallel Imaging applications in cardiac MRI.

Among recent parallel MR imaging reconstruction advances, a Bayesian method called Edge-preserving Parallel Imaging reconstructions with GRAph cuts Minimization (EPIGRAM) has been demonstrated to significantly improve signal-to-noise ratio when compared with conventional regularized sensitivity encoding method. However, EPIGRAM requires a large number of iterations in proportion to the number of intensity labels in the image, making it computationally expensive for high dynamic range images. The objective of this study is to develop a Fast EPIGRAM reconstruction based on the efficient binary jump move algorithm that provides a logarithmic reduction in reconstruction time while maintaining image quality. Preliminary in vivo validation of the proposed algorithm is presented for two-dimensional cardiac cine MR imaging and three-dimensional coronary MR angiography at acceleration factors of 2-4. Fast EPIGRAM was found to provide similar image quality to EPIGRAM and maintain the previously reported signal-to-noise ratio improvement over regularized sensitivity encoding method, while reducing EPIGRAM reconstruction time by 25-50 times.

Automated framework for digital radiation dose index reporting from CT dose reports.

OBJECTIVE: Radiation exposure from CT studies has increased over the past 30 years in the United States and now constitutes approximately 50% of the radiation dose index administered in the health care setting. Tracking CT dose index (CTDI) is cumbersome because it relies on a manufacturer-generated screen capture, which contains the estimated dose index exposure for the patient. The radiation dose index information is not digital but, rather, is "burned" into the image (i.e., not in numeric form, not as part of the image header or elsewhere associated with the study), making it difficult to automatically share these data with other information systems. The purpose of the dose index reporting application (DIRA) we developed for CT is to extract the radiation dose index information from the CTDI reports to eventually perform automated quality control, promote radiation safety awareness, and provide a longitudinal record of patient-specific health care-related radiation exposure. MATERIALS AND METHODS: A random selection of 518 CTDI reports were processed by the DIRA and the dose index information was extracted. CTDI reports using a standard DICOM C-STORE to the DIRA allow an automated process to compile radiation dose index and patient information in a Web-based framework using a structured query language (SQL) database. RESULTS: Our initial tests showed that the DIRA accurately extracted dose index information from 518 of 518 CTDI reports (100%). Because the extracted CTDI descriptor-dose-length product-is based on standard CTDI measurements obtained using fixed-size cylindric polymethylmethacrylate phantoms, preliminary studies have been performed to correct for patient size by applying correction factors derived from CTDI measurements using a range of phantom sizes from 6 to 32 cm in diameter. Our system provides a way to automatically track CTDI on existing CT scanners and does not rely on the DICOM SR Dose Index Report standard, which is available on only the newest CT scanners. CONCLUSION: A modular and vendor-independent DIRA system can be integrated with any existing CT scanner. This system greatly facilitates digital dose index reporting and makes it possible to provide a longitudinal record of the health care radiation exposure estimate in an individual patient's health record.

A comparative study of energy minimization methods for Markov random fields with smoothness-based priors.

Among the most exciting advances in early vision has been the development of efficient energy minimization algorithms for pixel-labeling tasks such as depth or texture computation. It has been known for decades that such problems can be elegantly expressed as Markov random fields, yet the resulting energy minimization problems have been widely viewed as intractable. Recently, algorithms such as graph cuts and loopy belief propagation (LBP) have proven to be very powerful: for example, such methods form the basis for almost all the top-performing stereo methods. However, the tradeoffs among different energy minimization algorithms are still not well understood. In this paper we describe a set of energy minimization benchmarks and use them to compare the solution quality and running time of several common energy minimization algorithms. We investigate three promising recent methods graph cuts, LBP, and tree-reweighted message passing in addition to the well-known older iterated conditional modes (ICM) algorithm. Our benchmark problems are drawn from published energy functions used for stereo, image stitching, interactive segmentation, and denoising. We also provide a general-purpose software interface that allows vision researchers to easily switch between optimization methods. Benchmarks, code, images, and results are available at http://vision.middlebury.edu/MRF/.

A maximum likelihood approach to parallel imaging with coil sensitivity noise.

Parallel imaging is a powerful technique to speed up magnetic resonance (MR) image acquisition via multiple coils. Both the received signal of each coil and its sensitivity map, which describes its spatial response, are needed during reconstruction. Widely used schemes such as SENSE assume that sensitivity maps of the coils are noiseless while the only errors are in coil outputs. In practice, however, sensitivity maps are subject to a wide variety of errors. At first glance, sensitivity noise appears to result in an errors-in-variables problem of the kind that is typically solved using total least squares (TLSs). However, existing TLS algorithms are in general inappropriate for the specific type of block structure that arises in parallel imaging. In this paper, we take a maximum likelihood approach to the problem of parallel imaging in the presence of independent Gaussian sensitivity noise. This results in a quasi-quadratic objective function, which can be efficiently minimized. Experimental evidence suggests substantial gains over conventional SENSE, especially in nonideal imaging conditions like low signal-to-noise ratio (SNR), high g-factors and large acceleration, using sensitivity maps suffering from misalignment, ringing, and random noise.

Conversion-to-open in laparoscopic appendectomy: A cohort analysis of risk factors and outcomes.

BACKGROUND: Identifying risk factors for conversion from laparoscopic to open appendectomy could select patients who may benefit from primary open appendectomy. We aimed to develop a predictive scoring model for conversion from laparoscopic to open based on pre-operative patient characteristics. METHODS: A retrospective review of the State Inpatient Database (2007-2011) was performed using derivation (N = 71,617) and validation (N = 143,235) cohorts of adults ≥ 18 years with acute appendicitis treated by laparoscopic-only (LA), conversion from laparoscopic to open (CA), or primary open (OA) appendectomy. Pre-operative variables independently associated with CA were identified and reported as odds ratios (OR) with 95% confidence intervals (CI). A weighted integer-based scoring model to predict CA was designed based on pre-operative variable ORs, and complications between operative subgroups were compared. RESULTS: Independent predictors of CA in the derivation cohort were age ≥40 (OR 1.67; CI 1.55-1.80), male sex (OR 1.25; CI 1.17-1.34), black race (OR 1.46; CI 1.28-1.66), diabetes (OR 1.47; CI 1.31-1.65), obesity (OR 1.56; CI 1.40-1.74), and acute appendicitis with abscess or peritonitis (OR 7.00; CI 6.51-7.53). In the validation cohort, the CA predictive scoring model had an optimal cutoff score of 4 (range 0-9). The risk of conversion-to-open was ≤5% for a score <4, compared to 10-25% for a score ≥4. On composite outcomes analysis controlling for all pre-operative variables, CA had a higher likelihood of infectious/inflammatory (OR 1.44; CI 1.31-1.58), hematologic (OR 1.31; CI 1.17-1.46), and renal (OR 1.22; CI 1.06-1.39) complications compared to OA. Additionally, CA had a higher likelihood of infectious/inflammatory, respiratory, cardiovascular, hematologic, and renal complications compared to LA. CONCLUSIONS: CA patients have an unfavorable complication profile compared to OA. The predictors identified in this scoring model could help select for patients who may benefit from primary open appendectomy.

Incidence and implications of postoperative supraventricular tachycardia after pulmonary lobectomy.

OBJECTIVE: We sought to determine the rate of postoperative supraventricular tachycardia (POSVT) in patients undergoing pulmonary lobectomy, and its association with adverse outcomes. METHODS: Using the State Inpatient Database, from the Healthcare Cost and Utilization Project, we reviewed lobectomies performed (2009-2011) in California, Florida, and New York, to determine POSVT incidence. Patients were grouped by presence or absence of POSVT, with or without other complications. Stroke rates were analyzed independently from other complications. Multivariable regression analysis was used to determine factors associated with POSVT. RESULTS: Among 20,695 lobectomies performed, 2449 (11.8%) patients had POSVT, including 1116 (5.4%) with isolated POSVT and 1333 (6.4%) with POSVT with other complications. Clinical predictors of POSVT included age ≥75 years, male gender, white race, chronic obstructive pulmonary disease, congestive heart failure, thoracotomy surgical approach, and pulmonary complications. POSVT was associated with an increase of: stroke (odds ratio [OR] 1.74; 95% confidence interval [CI] 1.03-2.94); in-hospital death (OR 1.85; 95% CI 1.45-2.35); LOS (OR 1.33; 95% CI 1.29-1.37); and readmission (OR 1.29; 95% CI 1.04-1.60). The stroke rate was <1% in patients who had isolated POSVT, and 1.5% in patients with POSVT with other complications. Patients with isolated POSVT had increased readmission and LOS, and a marginal increase in stroke rate, compared with patients with an uncomplicated course. CONCLUSIONS: POSVT is common in patients undergoing pulmonary lobectomy and is associated with adverse outcomes. Comparative studies are needed to determine whether strict adherence to recently published guidelines will decrease the rate of stroke, readmission, and death after POSVT in thoracic surgical patients.

Variability in length of stay after uncomplicated pulmonary lobectomy: is length of stay a quality metric or a patient metric?†.

OBJECTIVES: Previous studies have identified predictors of prolonged length of stay (LOS) following pulmonary lobectomy. LOS is typically described to have a direct relationship to postoperative complications. We sought to determine the LOS and factors associated with variability after uncomplicated pulmonary lobectomy. METHODS: Analysing the State Inpatient Databases, Healthcare Cost and Utilization Project, Agency for Healthcare Research and Quality database, we reviewed lobectomies performed (2009-11) on patients in California, Florida and New York. LOS and comorbidities were identified. Multivariable regression analysis (MVA) was used to determine factors associated with LOS greater than the median. Patients with postoperative complications or death were excluded. RESULTS: Among 22 647 lobectomies performed, we identified 13 099 patients (58%) with uncomplicated postoperative courses (mean age = 66 years; 56% female; 76% white, 57% Medicare; median DEYO comorbidity score = 3, 55% thoracotomy, 45% thoracoscopy/robotic). There was a wide distribution in LOS [median LOS = 5 days; interquartile range (IQR) 4-7]. By MVA, predictors of prolonged LOS included, age ≥ 75 years [odds ratio (OR) 1.7, 95% confidence interval (CI) 1.4-2.0], male gender (OR 1.2, 95% CI 1.1-1.2), chronic obstructive pulmonary disease (OR 1.6, 95% CI 1.5-1.7) and other comorbidities, Medicaid payer (OR 1.7, 95% CI 1.4-2.1) versus private insurance, thoracotomy (OR 3.0, 95% CI 2.8-3.3) versus video-assisted thoracoscopic surgery/robotic approach and low hospital volume (OR 2.4, 95% CI 2.1-2.6). CONCLUSIONS: Variability exists in LOS following even uncomplicated pulmonary lobectomy. Variability is driven by clinical factors such as age, gender, payer and comorbidities, but also by surgical approach and volume. All of these factors should be taken into account when designing clinical care pathways or when allocating payment resources. Attempts to define an optimal LOS depend heavily upon the patient population studied.

Incidence and Factors Associated With Hospital Readmission After Pulmonary Lobectomy.

BACKGROUND: Readmission rates after major procedures are used to benchmark quality of care. We sought to identify readmission diagnoses and factors associated with readmission in patients undergoing pulmonary lobectomy. METHODS: Analyzing the State Inpatient Databases (Healthcare Cost and Utilization Project), we reviewed all lobectomies performed from 2009 to 2011 in California, Florida, and New York. The group was subdivided into open (OL) versus minimally invasive lobectomy (MIL; thoracoscopic/robotic). We used unique identifiers to determine 30- and 90-day readmission rates and diagnoses and performed regression analysis to determine factors associated with readmission. RESULTS: A total of 22,647 lobectomies were identified (58.8% OL vs 41.2% MIL; median age, 68 years; median length of stay, 6 days). Most patients (59.8%) had routine discharge home (home health care, 29.4%; transfer to other facility, 8.8%; mortality, 1.9%). The 30-day readmission rate was 11.5% (OL 12.0% vs MIL 10.8%, p = 0.01), while the 90-day readmission rate was 19.8% (OL 21.1% vs MIL 17.9%, p < 0.001). The most common readmission diagnoses were pulmonary (24.1%), cardiovascular (16.3%), and complications related to surgical/medical procedures (15.1%). Preoperative factors associated with readmission included male gender (odds ratio, 1.19), Medicaid payer (odds ratio, 1.29), and several individual comorbidities. Surgical approach and postoperative complications were not independently associated with readmission. CONCLUSIONS: Readmission is a frequent event after pulmonary lobectomy and is strongly associated with preoperative demographic factors and comorbidities. Resources and services should be directed to patients at risk for readmission and multicomponent care pathways developed that may circumvent the need for repeat hospitalization.

A Hypergraph-Based Reduction for Higher-Order Binary Markov Random Fields.

Higher-order Markov Random Fields, which can capture important properties of natural images, have become increasingly important in computer vision. While graph cuts work well for first-order MRF's, until recently they have rarely been effective for higher-order MRF's. Ishikawa's graph cut technique [1], [2] shows great promise for many higher-order MRF's. His method transforms an arbitrary higher-order MRF with binary labels into a first-order one with the same minima. If all the terms are submodular the exact solution can be easily found; otherwise, pseudoboolean optimization techniques can produce an optimal labeling for a subset of the variables. We present a new transformation with better performance than [1], [2], both theoretically and experimentally. While [1], [2] transforms each higher-order term independently, we use the underlying hypergraph structure of the MRF to transform a group of terms at once. For n binary variables, each of which appears in terms with k other variables, at worst we produce n non-submodular terms, while [1], [2] produces O(nk). We identify a local completeness property under which our method perform even better, and show that under certain assumptions several important vision problems (including common variants of fusion moves) have this property. We show experimentally that our method produces smaller weight of non-submodular edges, and that this metric is directly related to the effectiveness of QPBO [3]. Running on the same field of experts dataset used in [1], [2] we optimally label significantly more variables (96 versus 80 percent) and converge more rapidly to a lower energy. Preliminary experiments suggest that some other higher-order MRF's used in stereo [4] and segmentation [5] are also locally complete and would thus benefit from our work.

Simultaneous multiple volume (SMV) acquisition algorithm for real-time navigator gating.

Navigator gating techniques can effectively reduce motion effects in MRI by accepting data only when the object is in a small range of positions at the cost of significantly prolonging scan time. A simultaneous multiple volume (SMV) algorithm is reported here that can substantially increase the scan efficiency while maintaining the effectiveness of motion suppression. This is achieved by acquiring different image volumes at different motion states. Initial experiments demonstrate that SMV can significantly increase the scan efficiency of navigator MRI.

What energy functions can be minimized via graph cuts?

In the last few years, several new algorithms based on graph cuts have been developed to solve energy minimization problems in computer vision. Each of these techniques constructs a graph such that the minimum cut on the graph also minimizes the energy. Yet, because these graph constructions are complex and highly specific to a particular energy function, graph cuts have seen limited application to date. In this paper, we give a characterization of the energy functions that can be minimized by graph cuts. Our results are restricted to functions of binary variables. However, our work generalizes many previous constructions and is easily applicable to vision problems that involve large numbers of labels, such as stereo, motion, image restoration, and scene reconstruction. We give a precise characterization of what energy functions can be minimized using graph cuts, among the energy functions that can be written as a sum of terms containing three or fewer binary variables. We also provide a general-purpose construction to minimize such an energy function. Finally, we give a necessary condition for any energy function of binary variables to be minimized by graph cuts. Researchers who are considering the use of graph cuts to optimize a particular energy function can use our results to determine if this is possible and then follow our construction to create the appropriate graph. A software implementation is freely available.

Dynamic programming and graph algorithms in computer vision.

Optimization is a powerful paradigm for expressing and solving problems in a wide range of areas, and has been successfully applied to many vision problems. Discrete optimization techniques are especially interesting since, by carefully exploiting problem structure, they often provide nontrivial guarantees concerning solution quality. In this paper, we review dynamic programming and graph algorithms, and discuss representative examples of how these discrete optimization techniques have been applied to some classical vision problems. We focus on the low-level vision problem of stereo, the mid-level problem of interactive object segmentation, and the high-level problem of model-based recognition.

Bayesian parallel imaging with edge-preserving priors.

Existing parallel MRI methods are limited by a fundamental trade-off in that suppressing noise introduces aliasing artifacts. Bayesian methods with an appropriately chosen image prior offer a promising alternative; however, previous methods with spatial priors assume that intensities vary smoothly over the entire image, resulting in blurred edges. Here we introduce an edge-preserving prior (EPP) that instead assumes that intensities are piecewise smooth, and propose a new approach to efficiently compute its Bayesian estimate. The estimation task is formulated as an optimization problem that requires a nonconvex objective function to be minimized in a space with thousands of dimensions. As a result, traditional continuous minimization methods cannot be applied. This optimization task is closely related to some problems in the field of computer vision for which discrete optimization methods have been developed in the last few years. We adapt these algorithms, which are based on graph cuts, to address our optimization problem. The results of several parallel imaging experiments on brain and torso regions performed under challenging conditions with high acceleration factors are shown and compared with the results of conventional sensitivity encoding (SENSE) methods. An empirical analysis indicates that the proposed method visually improves overall quality compared to conventional methods.

Statistical aspects of parallel imaging reconstruction.

A statistical interpretation of existing parallel magnetic resonance imaging methods reveals that the underlying noise model is of additive independent Gaussian noise. In reality MR imaging processes suffer from a variety of noise, errors and other uncertainties. A careful statistical analysis of these uncertainties can potentially allow significant improvement of the reconstruction process. In this paper we present such an analysis and describe a few very recent approaches to handle these statistical models. We show examples of simulation and in vivo reconstructed data which demonstrate the potential of the statistical approach.

Automatic algorithm for correcting motion artifacts in time-resolved two-dimensional magnetic resonance angiography using convex projections.

Time-resolved contrast enhanced magnetic resonance angiography (MRA) may suffer from involuntary patient motion. It is noted that while MR signal change associated with motion is large in magnitude and has smooth phase variation in k-phase, signal change associated with vascular enhancement is small in magnitude and has rapid phase variation in k-space. Based upon this observation, a novel projection onto convex sets (POCS) algorithm is developed as an automatic iterative method to remove motion artifacts. The presented POCS algorithm consists of high-pass phase filtering and convex projections in both k-space and image space. Without input of detailed motion knowledge, motion effects are filtered out, while vasculature information is preserved. The proposed method can be effective for a large class of nonrigid motions, including through-plane motion. The algorithm is stable and converges quickly, usually within five iterations. A double-blind evaluation on a set of clinical MRA cases shows that a completely unsupervised version of the algorithm produces significantly better rank scores (P=0.038) when compared to angiograms produced manually by an experienced radiologist.

Improved signal-to-noise ratio in parallel coronary artery magnetic resonance angiography using graph cuts based Bayesian reconstruction.

High resolution 3D coronary artery MR angiography is time-consuming and can benefit from accelerated data acquisition provided by parallel imaging techniques without sacrificing spatial resolution. Currently, popular maximum likelihood based parallel imaging reconstruction techniques such as the SENSE algorithm offer this advantage at the cost of reduced signal-to-noise ratio (SNR). Maximum a posteriori (MAP) reconstruction techniques that incorporate globally smooth priors have been developed to recover this SNR loss, but they tend to blur sharp edges in the target image. The objective of this study is to demonstrate the feasibility of employing edge-preserving Markov random field priors in a MAP reconstruction framework, which can be solved efficiently using a graph cuts based optimization algorithm. The preliminary human study shows that our reconstruction provides significantly better SNR than the SENSE reconstruction performed by a commercially available scanner for navigator gated steady state free precession 3D coronary magnetic resonance angiography images (n = 4).

Automatic selection of mask and arterial phase images for temporally resolved MR digital subtraction angiography.

For time-resolved background-subtracted contrast-enhanced magnetic resonance angiography, the bright and sparse arterial signal allows unique identification of contrast bolus arrival in the arteries. This article presents an automatic filtering algorithm using such arterial characterization for selecting arterial phase images and mask images to generate an optimal summary arteriogram. A paired double-blinded comparison demonstrated that this automatic algorithm is as effective as the manual process.

Multiprocessor scheduling implementation of the simultaneous multiple volume (SMV) navigator method.

The simultaneous multiple volume (SMV) approach in navigator-gated MRI allows the use of the whole motion range or the entire scan time for the reconstruction of final images by simultaneously acquiring different image volumes at different motion states. The motion tolerance range for each volume is kept small, thus SMV substantially increases the scan efficiency of navigator methods while maintaining the effectiveness of motion suppression. This article reports a general implementation of the SMV approach using a multiprocessor scheduling algorithm. Each motion state is regarded as a processor and each volume is regarded as a job. An efficient scheduling that completes all jobs in minimal time is maintained even when the motion pattern changes. Initial experiments demonstrated that SMV significantly increased the scan efficiency of navigator-gated MRI.