Smita Patel

University of Michigan, Ann Arbor, Michigan, United States

Are you Smita Patel?

Claim your profile

Publications (61)104.08 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: The buildup of noncalcified plaques (NCPs) that are vulnerable to rupture in coronary arteries is a risk for myocardial infarction. Interpretation of coronary CT angiography (cCTA) to search for NCP is a challenging task for radiologists due to the low CT number of NCP, the large number of coronary arteries, and multiple phase CT acquisition. The authors conducted a preliminary study to develop machine learning method for automated detection of NCPs in cCTA.
    Medical physics. 08/2014; 41(8):081901.
  • [Show abstract] [Hide abstract]
    ABSTRACT: The authors are developing a computer-aided detection system to assist radiologists in analysis of coronary artery disease in coronary CT angiograms (cCTA). This study evaluated the accuracy of the authors' coronary artery segmentation and tracking method which are the essential steps to define the search space for the detection of atherosclerotic plaques.
    Medical physics. 08/2014; 41(8):081912.
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: IntroductionLeft ventricular (LV) and right ventricular (RV) volumetric and functional parameters are important biomarkers for morbidity and mortality in patients with heart failure.PurposeTo retrospectively determine reference mean values of LV and RV volume, function and mass normalised by age, gender and body surface area (BSA) from retrospectively electrocardiographically gated 64-slice cardiac computed tomography (CCT) by using automated analysis software in healthy adults.Materials and Methods The study was approved by the institutional review board with a waiver of informed consent. Seventy-four healthy subjects (49% female, mean age 49.6 ± 11) free of hypertension and hypercholesterolaemia with a normal CCT formed the study population. Analyses of LV and RV volume (end-diastolic, end-systolic and stroke volumes), function (ejection fraction), LV mass and inter-rater reproducibility were performed with commercially available analysis software capable of automated contour detection. General linear model analysis was performed to assess statistical significance by age group after adjustment for gender and BSA. Bland–Altman analysis assessed the inter-rater agreement.ResultsThe reference range for LV and RV volume, function, and LV mass was normalised to age, gender and BSA. Statistically significant differences were noted between genders in both LV mass and RV volume (P-value < 0.0001). Age, in concert with gender, was associated with significant differences in RV end-diastolic volume and LV ejection fraction (P-values 0.027 and 0.03). Bland–Altman analysis showed acceptable limits of agreement (±1.5% for ejection fraction) without systematic error.ConclusionLV and RV volume, function and mass normalised to age, gender and BSA can be reported from CCT datasets, providing additional information important for patient management.
    Journal of Medical Imaging and Radiation Oncology 06/2014; · 0.98 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Non-calcified plaque (NCP) detection in coronary CT angiography (cCTA) is challenging due to the low CT number of NCP, the large number of coronary arteries and multiple phase CT acquisition. We are developing computer-vision methods for automated detection of NCPs in cCTA. A data set of 62 cCTA scans with 87 NCPs was collected retrospectively from patient files. Multiscale coronary vessel enhancement and rolling balloon tracking were first applied to each cCTA volume to extract the coronary artery trees. Each extracted vessel was reformatted to a straightened volume composed of cCTA slices perpendicular to the vessel centerline. A topological soft-gradient (TSG) detection method was developed to prescreen for both positive and negative remodeling candidates by analyzing the 2D topological features of the radial gradient field surface along the vessel wall. A quantitative luminal analysis was newly designed for feature extraction and false positive (FP) reduction. We extracted 9 geometric features and 6 gray-level features, to quantify the differences between NCPs and FPs. The gray-level features included 4 features to measure local statistical characteristics and 2 asymmetry features to measure the asymmetric spatial location of gray-level density along the vessel centerline. The geometric features included a radius differential feature and 8 features extracted from two transformed volumes: the volumetric shape indexing and the gradient direction mapping volumes. With a machine learning algorithm and feature selection method, useful features were selected and combined into an NCP likelihood measure to differentiate TPs from FPs. With the NCP likelihood measure as a decision variable in the receiver operating characteristic (ROC) analysis, the area under the curve achieved a value of 0.85±0.01, indicating that the luminal analysis is effective in reducing FPs for NCP detection.
    SPIE Medical Imaging; 03/2014
  • [Show abstract] [Hide abstract]
    ABSTRACT: A 3D multiscale intensity homogeneity transformation (MIHT) method was developed to reduce false positives (FPs) in our previously developed CAD system for pulmonary embolism (PE) detection. In MIHT, the voxel intensity of a PE candidate region was transformed to an intensity homogeneity value (IHV) with respect to the local median intensity. The IHVs were calculated in multiscales (MIHVs) to measure the intensity homogeneity, taking into account vessels of different sizes and different degrees of occlusion. Seven new features including the entropy, gradient, and moments that characterized the intensity distributions of the candidate regions were derived from the MIHVs and combined with the previously designed features that described the shape and intensity of PE candidates for the training of a linear classifier to reduce the FPs. 59 CTPA PE cases were collected from our patient files (UM set) with IRB approval and 69 cases from the PIOPED II data set with access permission. 595 and 800 PEs were identified as reference standard by experienced thoracic radiologists in the UM and PIOPED set, respectively. FROC analysis was used for performance evaluation. Compared with our previous CAD system, at a test sensitivity of 80%, the new method reduced the FP rate from 18.9 to 14.1/scan for the PIOPED set when the classifier was trained with the UM set and from 22.6 to 16.0/scan vice versa. The improvement was statistically significant (p<0.05) by JAFROC analysis. This study demonstrated that the MIHT method is effective in reducing FPs and improving the performance of the CAD system.
    Proc SPIE 04/2013;
  • [Show abstract] [Hide abstract]
    ABSTRACT: We are developing an automated registration method for coronary arterial trees from multiple-phase cCTA to build a best-quality tree to facilitate detection of stenotic plaques. Cubic B-spline with fast localized optimization (CBSO) is designed to register the initially segmented left and right coronary arterial trees (LCA or RCA) separately in adjacent phase pairs where displacements are small. First, the corresponding trees in phase 1 and 2 are registered. The phase 3 tree is then registered to the combined tree. Similarly the trees in phases 4, 5, and 6 are registered. An affine transform with quadratic terms and nonlinear simplex optimization (AQSO) is designed to register the trees between phases with large displacements, namely, registering the combined tree from phases 1, 2, and 3 to that from phases 4, 5, and 6. Finally, CBSO is again applied to the AQSO registered volumes for final refinement. The costs determined by the distances between the vessel centerlines, bifurcation points and voxels of the trees are minimized to guide both CBSO and AQSO registration. The registration performance was evaluated on 22 LCA and 22 RCA trees on 22 CTA scans with 6 phases from 22 patients. The average distance between the centerlines of the registered trees was used as a registration quality index. The average distances for LCA and RCA registration for 6 phases and 22 patients were 1.49 and 1.43 pixels, respectively. This study demonstrates the feasibility of using automated method for registration of coronary arterial trees from multiple cCTA phases.
    Proc SPIE 03/2013; 59(16).
  • [Show abstract] [Hide abstract]
    ABSTRACT: The curved planar reformation (CPR) method re-samples the vascular structures along the vessel centerline to generate longitudinal cross-section views. The CPR technique has been commonly used in coronary CTA workstation to facilitate radiologists' visual assessment of coronary diseases, but has not yet been used for pulmonary vessel analysis in CTPA due to the complicated tree structures and the vast network of pulmonary vasculature. In this study, a new curved planar reformation and optimal path tracing (CROP) method was developed to facilitate feature extraction and false positive (FP) reduction and improve our PE detection system. PE candidates are first identified in the segmented pulmonary vessels at prescreening. Based on Dijkstra's algorithm, the optimal path (OP) is traced from the pulmonary trunk bifurcation point to each PE candidate. The traced vessel is then straightened and a reformatted volume is generated using CPR. Eleven new features that characterize the intensity, gradient, and topology are extracted from the PE candidate in the CPR volume and combined with the previously developed 9 features to form a new feature space for FP classification. With IRB approval, CTPA of 59 PE cases were retrospectively collected from our patient files (UM set) and 69 PE cases from the PIOPED II data set with access permission. 595 and 800 PEs were manually marked by experienced radiologists as reference standard for the UM and PIOPED set, respectively. At a test sensitivity of 80%, the average FP rate was improved from 18.9 to 11.9 FPs/case with the new method for the PIOPED set when the UM set was used for training. The FP rate was improved from 22.6 to 14.2 FPs/case for the UM set when the PIOPED set was used for training. The improvement in the free response receiver operating characteristic (FROC) curves was statistically significant (p<0.05) by JAFROC analysis, indicating that the new features extracted from the CROP method are useful for FP reduction.
    Proc SPIE 03/2013;
  • [Show abstract] [Hide abstract]
    ABSTRACT: Non-calcified plaque (NCP) detection in coronary CT angiography (cCTA) is challenging due to the low CT number of NCP, the large number of coronary arteries and multiple phase CT acquisition. We are developing computervision methods for automated detection of NCPs in cCTA. A data set of 62 cCTA scans with 87 NCPs was collected retrospectively from patient files. Multiscale coronary vessel enhancement and rolling balloon tracking were first applied to each cCTA volume to extract the coronary artery trees. Each extracted vessel was reformatted to a straightened volume composed of cCTA slices perpendicular to the vessel centerline. A new topological soft-gradient (TSG) detection method was developed to prescreen for both positive and negative remodeling candidates by analyzing the 2D topological features of the radial gradient field surface along the vessel wall. Nineteen features were designed to describe the relative location along the coronary artery, shape, distribution of CT values, and radial gradients of each NCP candidate. With a machine learning algorithm and a two-loop leave-one-case-out training and testing resampling method, useful features were selected and combined into an NCP likelihood measure to differentiate TPs from FPs. The detection performance was evaluated by FROC analysis. Our TSG method achieved a sensitivity of 96.6% with 35.4 FPs/scan at prescreening. Classification with the NCP likelihood measure reduced the FP rates to 13.1, 10.0 and 6.7 FPs/scan at sensitivities of 90%, 80%, and 70%, respectively. These results demonstrated that the new TSG method is useful for computerized detection of NCPs in cCTA.
    Proc SPIE 03/2013;
  • [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE We conducted a preliminary study to develop machine learning methods for automated detection of non-calcified plaques (NCPs) in coronary CTA (cCTA). METHOD AND MATERIALS With IRB approval, a data set of 38 cCTA scans was retrospectively collected from 38 patient files. Experienced cardiothoracic radiologists marked the location of NCPs and rated their conspicuity on a 10-point scale (10=most obvious). A total of 55 NCPs with conspicuity of 5.2±2.8 was identified and 31 were assessed as positive vessel remodeling (PVR). Our previously developed method that combined multiscale vessel enhancement with 3D rolling balloon vessel tracking was applied to each cCTA volume to extract the coronary artery trees. Each extracted vessel was reformatted to a straightened volume composed of cCTA slices perpendicular to the vessel centerline using multi-planar reformation. The vessel wall gradient (VWG) at a given point was calculated as the maximum gradient along the radial direction from the vessel center to the wall and averaged over a ±45o sector. The minimum VWG around the vessel circumference was then found on each reformatted slice. The local minima of the VWG profile along each vessel were labeled as NCP candidates. Eleven features were designed to characterize the relative location along the coronary artery, shape, density, and VWGs of each NCP candidate. With a machine learning algorithm and a two-loop leave-one-case-out resampling method for training and testing, useful features were selected and combined into an NCP likelihood measure to differentiate true NCPs from FPs and to evaluate the performance of the detection system. RESULTS Our NCP prescreening method achieved a sensitivity of 80% (44/55) at an FP rate of 19.7 FPs/scan. An average of 3 features was selected. With the NCP likelihood measure, the FP rates reduced to 11.6, 5.8 and 3.7 FPs/scan at sensitivities of 80, 70, and 60%, respectively. At the same FP rates, the corresponding sensitivities for PVR were 77, 70, and 59%, respectively. CONCLUSION The study demonstrates the feasibility of our approach to the detection of NCPs in cCTA. Further investigation is underway to improve the accuracy and to validate the performance in a larger data set. CLINICAL RELEVANCE/APPLICATION NCP detection remains challenging due to the large number of coronary arteries and multiple phase CT acquisition. CAD may be useful for radiologists in cCTA interpretation and risk assessment.
    Radiological Society of North America 2012 Scientific Assembly and Annual Meeting; 11/2012
  • [Show abstract] [Hide abstract]
    ABSTRACT: The purpose of the study was to determine the effectiveness of a collaborative educational, continuous quality improvement (CQI) initiative to increase appropriate use of coronary computed tomography angiography (CCTA). Potential overuse of CCTA has prompted multisociety appropriate use criteria (AUC) publications. This prospective, observational study was conducted with pre-intervention (July 2007 to June 2008), intervention (July 2008 to June 2010), and follow-up (July 2010 to December 2010) periods during which patients were enrolled in the Advanced Cardiovascular Imaging Consortium (ACIC) at 47 Michigan hospitals. Continuous education was provided to referring physicians. The possibility of losing third-party payer coverage in the absence of a measurable change in AUC was emphasized. AUC was compared between the 3 periods. The study group included 25,387 patients. Compared with the pre-intervention period, there was a 23.4% increase in appropriate (61.3% to 80%, p < 0.0001), 60.3% decrease in inappropriate (14.6% to 5.8%, p < 0.0001), 40.8% decrease in uncertain (10.3% to 6.1%, p < 0.0001), and 41.7% decrease in unclassifiable (13.9% to 8.1%, p < 0.0001) scans during follow-up. Between pre-intervention and follow-up, change in CCTA referrals by provider specialty were cardiology (appropriate: 60.4% to 79.5%; inappropriate: 13% to 5.2%; p < 0.0001), internal medicine/family practice (appropriate: 51.1% to 70.4%; inappropriate: 20.2% to 12.5%; p < 0.0001), emergency medicine (appropriate: 83.6% to 91.6%; inappropriate: 9.1% to 0.6%; p < 0.0001), and other (appropriate: 61.1% to 83.2%; inappropriate: 18.6% to 5.9%; p < 0.0001). Application of a systematic CQI and emphasis on possible loss of coverage were associated with a significant improvement in the proportion of CCTA examinations meeting AUC across referring physician specialties.
    Journal of the American College of Cardiology 08/2012; 60(13):1185-91. · 14.09 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Coronary computed tomography angiography (CCTA) is an emerging noninvasive anatomical method for evaluation of patients with suspected coronary artery disease (CAD). Multicenter clinical registries are key to efforts to establish the role of CCTA in CAD diagnosis and management. The Advanced Cardiovascular Imaging Consortium (ACIC) is a statewide, multicenter collaborative quality initiative with the intent to establish quality and appropriate use of CCTA in Michigan. The ACIC is sponsored by the Blue Cross Blue Shield of Michigan/Blue Care Network, and its 47 sites include imaging centers that offer CCTA and meet established structure and process standards for participation. Patients enrolled include those with suspected ischemia with or without known CAD, and individuals across the entire spectrum of CAD risk. Patient demographics, history, CCTA scan-related data and findings, and 90-day follow-up data are entered prospectively into a centralized database with strict validation tools and processes. Collaborative quality initiatives include radiation dose reduction and appropriate CCTA use by education and feedback to participating sites and referring physicians. Across a wide range of institutions, the ACIC permits evaluation of "real-world" utilization and effectiveness of CCTA and examines an alternative, nontraditional approach to utilization management wherein physicians and payers collaborate to address the growing problem of cardiac imaging overutilization.
    American heart journal 03/2012; 163(3):346-53. · 4.65 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Vessel segmentation is a fundamental step in an automated pulmonary embolism (PE) detection system. The purpose of this study is to improve the segmentation scheme for pulmonary vessels affected by PE and other lung diseases. We have developed a multiscale hierarchical vessel enhancement and segmentation (MHES) method for pulmonary vessel tree extraction based on the analysis of eigenvalues of Hessian matrices. However, it is difficult to segment the pulmonary vessels accurately under suboptimal conditions, such as vessels occluded by PEs, surrounded by lymphoid tissues or lung diseases, and crossing with other vessels. In this study, we developed a new vessel refinement method utilizing curved planar reformation (CPR) technique combined with optimal path finding method (MHES-CROP). The MHES segmented vessels straightened in the CPR volume was refined using adaptive gray level thresholding where the local threshold was obtained from least-square estimation of a spline curve fitted to the gray levels of the vessel along the straightened volume. An optimal path finding method based on Dijkstra's algorithm was finally used to trace the correct path for the vessel of interest. Two and eight CTPA scans were randomly selected as training and test data sets, respectively. Forty volumes of interest (VOIs) containing "representative" vessels were manually segmented by a radiologist experienced in CTPA interpretation and used as reference standard. The results show that, for the 32 test VOIs, the average percentage volume error relative to the reference standard was improved from 32.9+/-10.2% using the MHES method to 9.9+/-7.9% using the MHES-CROP method. The accuracy of vessel segmentation was improved significantly (p<0.05). The intraclass correlation coefficient (ICC) of the segmented vessel volume between the automated segmentation and the reference standard was improved from 0.919 to 0.988. Quantitative comparison of the MHES method and the MHES-CROP method with the reference standard was also evaluated by the Bland-Altman plot. This preliminary study indicates that the MHES-CROP method has the potential to improve PE detection.
    Proc SPIE 02/2012;
  • [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE/AIM Highlight the Dana Point Classification of Pulmonary Hypertension with an emphasis on the CT manifestation of disease. CONTENT ORGANIZATION 1. Introduction 2. Diagnosis and Classification System (Dana Point) 3. CT Findings of Pulmonary Hypertension 4. Treatment and Management 5. Summary and Conclusions SUMMARY The Dana Point criteria offer a more clear pathophysiologic-based approach to clinical pulmonary hypertension. In the era of rapidly developing treatment for this spectrum of disease, it is increasingly important for the radiologist to be familiar with the CT characteristics of Pulmonary Hypertension in order to become an integral part of the management team.
    Radiological Society of North America 2011 Scientific Assembly and Annual Meeting; 12/2011
  • [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE/AIM 1. Review classification, clinical and imaging features of congenital conditions that may lead to sudden cardiac death 2. Review cardiac CT/MRI techniques used to evaluate patients with an emphasis on technical advances 3. Discuss the clinical potential of CT and MRI in evaluating these patients and to correlate clinical and imaging findings CONTENT ORGANIZATION The spectrum of congenital causes of sudden cardiac death includes genetic 1° cardiomyopathies (non-compaction, hypertrophic and arrhthymogenic right ventricular cardiomyopathies), and anomalous coronary arteries (interarterial or intramural course), and will be presented as clinical case series. The following will be discussed: Clinical significance, treatment and prognosis CT/MRI evaluation (protocols, imaging findings, and recent advances) Alternative diagnostic methods SUMMARY Cardiac MR is an important tool for evaluation of genetic 1° cardiomyopathies providing assessment of myocardial morphology, function and tissue composition that guides clinical management in certain entities. CT is considered imaging modality of choice for non-invasive delineation of normal and anomalous coronary anatomy, and is useful for surgical planning vs. medical management. These imaging modalities can effectively identify causes of sudden cardiac death and alter clinical outcome.
    Radiological Society of North America 2011 Scientific Assembly and Annual Meeting; 11/2011
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The purpose of this study was to retrospectively determine CT-based normal reference values of left atrial volume, function, and diameter normalized by age, sex, and body surface area. The study group consisted of 74 subjects with normal findings at ECG-gated coronary CT angiography performed with retrospective gating. Analysis of left atrial volume (end-diastolic, end-systolic, and stroke volume) and function (ejection fraction) was performed with the Simpson method. Left atrial diameter was measured in the anteroposterior dimension. General linear model analysis was performed to model the data and assess statistical significance by age group after adjustment for sex and body surface area. The reference range for left atrial volume, function, and diameter was normalized (indexed) to age, sex, and body surface area in healthy subjects. A statistically significant difference was noted between left atrial volume and age without adjustment for sex and body surface area, but no statistically significant difference was found after adjustment for these variables. Sex and body surface area had a significant influence on left atrial volume, function, and diameter. Left atrial volume, function, and diameter normalized to age, sex, and body surface area can be reported from CTA datasets and may provide information important for patient care.
    American Journal of Roentgenology 09/2011; 197(3):631-7. · 2.90 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: BACKGROUND: Transesophageal echocardiography (TEE) is the standard for evaluating cardioembolic sources of stroke, although many strokes remain cryptogenic after TEE. Cardiac magnetic resonance (CMR) imaging may have advantages over TEE. We performed a prospective pilot study comparing CMR to TEE after stroke to assist in planning future definitive studies. METHODS: Individuals with nonlacunar stroke within 90 days of undergoing clinical TEE were prospectively identified and underwent a 1.5 Tesla research CMR scan. Exclusion criteria included >50% relevant cervical vessel stenosis and inability to undergo nonsedated CMR. A descriptive comparison of cardioembolic source (intracardiac thrombus/mass, aortic atheroma ≥4 mm, or patent foramen ovale [PFO]) by study type was performed. RESULTS: Twenty patients underwent CMR and TEE a median of 6 days apart. The median age was 51 years (interquartile range [IQR] 40, 63.5), 40% had hypertension, 15% had diabetes, 25% had a previous stroke/transient ischemic attack, 5% had atrial fibrillation, and none had coronary disease or heart failure. No patient had intracardiac thrombus or mass detected on either study. Aortic atheroma ≥4 mm thick was identified by TEE in 1 patient. CMR identified aortic atheroma as <4 mm in this patient (3 mm on CMR compared with 5 mm on TEE). PFO was identified in 6 of 20 patients on TEE; CMR found only 1 of these. CONCLUSIONS: In this pilot study, TEE identified more potential cardioembolic sources than CMR imaging. Future studies comparing TEE and CMR after stroke should focus on older subjects at higher risk for cardiac disease to determine whether TEE, CMR, or both can best elucidate potential cardioembolic sources.
    Journal of stroke and cerebrovascular diseases: the official journal of National Stroke Association 06/2011;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: To evaluate our prototype method for segmentation and tracking of the coronary arterial tree, which is the foundation for a computer-aided detection (CADe) system to be developed to assist radiologists in detecting non-calcified plaques in coronary CT angiography (cCTA) scans. The heart region was first extracted by a morphological operation and an adaptive thresholding method based on expectation-maximization (EM) estimation. The vascular structures within the heart region were enhanced and segmented using a multiscale coronary response (MSCAR) method that combined 3D multiscale filtering, analysis of the eigenvalues of Hessian matrices and EM estimation segmentation. After the segmentation of vascular structures, the coronary arteries were tracked by a 3D dynamic balloon tracking (DBT) method. The DBT method started at two manually identified seed points located at the origins of the left and right coronary arteries (LCA and RCA) for extraction of the arterial trees. The coronary arterial trees of a data set containing 20 ECG-gated contrast-enhanced cCTA scans were extracted by our MSCAR-DBT method and a clinical GE Advantage workstation. Two experienced thoracic radiologists visually examined the coronary arteries on the original cCTA scans and the rendered volume of segmented vessels to count the untracked false-negative (FN) segments and false positives (FPs) for both methods. For the visible coronary arterial segments in the 20 cases, the radiologists identified that 25 segments were missed by our MSCAR-DBT method, ranging from 0 to 5 FN segments in individual cases, and that 55 artery segments were missed by the GE software, ranging from 0 to 7 FN segments in individual cases. 19 and 15 FPs were identified in our and the GE coronary trees, ranging from 0 to 4 FPs for both methods in individual cases, respectively. The preliminary study demonstrates the feasibility of our MSCAR-DBT method for segmentation and tracking coronary artery trees. The results indicated that both our method and GE software can extract coronary artery trees reasonably well and the performance of our method is superior to that of GE software in this small data set. Further studies are underway to develop methods for improvement of the segmentation and tracking accuracy.
    Computerized medical imaging and graphics: the official journal of the Computerized Medical Imaging Society 05/2011; 36(1):1-10. · 1.04 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Vessel segmentation is a fundamental step in an automated pulmonary embolism (PE) detection system. The purpose of this study is to improve the segmentation scheme for pulmonary vessels affected by PE and other lung diseases. We have developed a multiscale hierarchical vessel enhancement and segmentation (MHES) method for pulmonary vessel tree extraction based on the analysis of eigenvalues of Hessian matrices. However, it is difficult to segment the pulmonary vessels accurately when the vessel is occluded by PEs and/or surrounded by lymphoid tissues or lung diseases. In this study, we developed a method that combines MHES with level set refinement (MHES-LSR) to improve vessel segmentation accuracy. The level set was designed to propagate the initial object contours to the regions with relatively high gray-level, high gradient, and high compactness as measured by the smoothness of the curvature along vessel boundaries. Two and eight CTPA scans were randomly selected as training and test data sets, respectively. Forty volumes of interest (VOI) containing "representative" vessels were manually segmented by a radiologist experienced in CTPA interpretation and used as reference standard. The results show that, for the 32 test VOIs, the average percentage volume error relative to the reference standard was improved from 31.7+/-10.9% using the MHES method to 7.7+/-4.7% using the MHES-LSR method. The correlation between the computer-segmented vessel volume and the reference standard was improved from 0.954 to 0.986. The accuracy of vessel segmentation was improved significantly (p
    Proc SPIE 03/2011;
  • [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE We are developing a computer-aided detection (CAD) system to assist radiologists in PE detection. A key step is to establish the reference standard by radiologists’ manual marking of PE in the CTPA cases for the training and testing of the CAD system. This study evaluated the variability of the radiologist-identified PE locations and the potential benefit of CAD. METHOD AND MATERIALS With IRB approval, 40 CTPA PE cases collected retrospectively from our patient files. Two experienced thoracic radiologists assessed each case independently. For a PE that occluded more than one branch of the arteries, the radiologist virtually split the single PE volume by marking the PE segment in each branch as a separate PE. For markings that the two radiologists did not agree, the case was re-read by each radiologist to obtain a consensus. Our prototype CAD system was applied to the 40 cases and the CAD marks were examined by the radiologists to evaluate whether there were true PEs not included in the reference standard. RESULTS For the 40 PE cases, 444 and 500 PEs were identified by radiologists R1 and R2, respectively. The two radiologists’ reading did not agree on 101 marks in 36 cases. After consensus, 65 (64.4%) and 36 (35.6%) of the 101 marks were determined to be true PEs and false positives (FPs), respectively. Of these, 48 and 17 were false negatives (FNs) by R1 and R2, and 14 and 22 were FPs in R1 and R2’s reading, respectively. The interobserver agreement between the two radiologists’ initial readings reached 0.903±0.04 with a kappa value of 0.613 using a ĸ statistic. At a test sensitivity of 80% and an average of 22.6 FPs/case, our CAD system detected 17 PEs that were FNs for both radiologists, and 6 PEs that were FNs for one of the radiologists. CONCLUSION There was substantial agreement (kappa>0.610) between two radiologists’ markings of PEs but consensus with multiple radiologists improved the reference standard. The results also demonstrated the feasibility of the CAD system for assisting radiologists in PE detection. CLINICAL RELEVANCE/APPLICATION An accurate CAD system may improve the efficacy of CTPA for PE detection. Early detection may reduce the mortality rate and complications caused by PE.
    Radiological Society of North America 2010 Scientific Assembly and Annual Meeting; 11/2010
  • [Show abstract] [Hide abstract]
    ABSTRACT: Although numerous strategies for radiation dose decrease in coronary computed tomographic angiography are effective, their combined impact on diagnostic performance is not known. We therefore assessed the effect of a standardized coronary computed tomographic angiographic protocol on diagnostic accuracy. We evaluated 80 consecutive patients from 3 sites with coronary computed tomographic angiography and quantitative coronary angiography. All sites initially used nonstandardized protocols; 2 sites then initiated a standardized protocol, and 1 site continued its nonstandardized protocol as a time-overlapping control. Two blinded readers interpreted coronary computed tomographic angiographic studies; a third obtained consensus. A blinded core laboratory performed quantitative coronary angiography. Each segment was graded as <50% or > or =50% diameter stenosis. Compared to those using nonstandardized protocols (n = 35), studies using standardized protocols (n = 45) had a trend to increased use of prospective gating (p = 0.09), lower voltage (p <0.01), decreased current (p <0.01), and shorter scan length (p <0.01). Median (interquartile range) radiation dose decreased from 5.7 mSv (4.0 to 10.8) to 2.0 mSv (1.3 to 3.4, p <0.001). There were no significant differences in sensitivity (100%, 20 of 20, vs 100%, 18 of 18, p = 1.0), specificity (93%, 14 of 15, vs 85%, 23 of 27, p = 0.61), or accuracy (97%, 34 of 35, vs 91%, 41 of 45, p = 0.27) by patient; sensitivity (83%, 33 of 40, vs 83%, 25 of 30, p = 0.93), specificity (92%, 86 of 93, vs 92%, 134 of 146, p = 0.85), or accuracy (89%, 119 of 133, vs 90%, 159 of 176, p = 0.80) by artery; or sensitivity (80%, 44 of 55, vs 72%, 26 of 36, p = 0.74), specificity (94%, 332 of 353, vs 94%, 499 of 531, p = 0.96), or accuracy (92%, 376 of 408, vs 93%, 525 of 567, p = 0.80) by segment. In conclusion, a standardized dose-decrease protocol for coronary computed tomographic angiography decreases radiation dose without affecting diagnostic performance.
    The American journal of cardiology 07/2010; 106(2):287-92. · 3.58 Impact Factor