Joseph K. Leader
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Joseph K. Leader has a total of 96 co-authors. Here are the top co-authors by number of shared publications:
Bin Zheng
Univ. of Pittsburgh (United States)
5 shared publications- Bin Zheng
School of Electrical and Computer Engineering, University of Oklahoma, Norman, OK, 73019, USA
4 shared publications - David Gur
Univ. of Pittsburgh (United States)
4 shared publications - Frank C. Sciurba
Division of Pulmonary, Allergy, and Critical Care Medicine, University of Pittsburgh, Pittsburgh, PA, USA
3 shared publications - Jiantao Pu
Department of Radiology, University of Pittsburgh, Pittsburgh, USA
3 shared publications Jun Tan
Univ. of Pittsburgh (United States)
3 shared publications
Latest Publications
- 13 October 2020
- Published by SPIE-Intl Soc Optical Eng in Journal of Medical Imaging
- Vol. 7 (05) , 051202-11
2D CNN versus 3D CNN for false-positive reduction in lung cancer screening
Purpose: To clarify whether and to what extent three-dimensional (3D) convolutional neural network (CNN) is superior to 2D CNN when applied to reduce false-positive nodule detections in the scenario of low-dose computed tomography (CT) lung cancer screening. Approach: We established a dataset consisting of 1600 chest CT examinations acquired on different subjects from various sources. There were in total 18,280 candidate nodules in these CT examinations, among which 9185 were nodules and 9095 were not nodules. For each candidate nodule, we extracted a number of cubic subvolumes with a dimension of 72 × 72 × 72 mm3 by rotating the CT examinations randomly for 25 times prior to the extraction of the axis-aligned subvolumes. These subvolumes were split into three groups in a ratio of 8 ∶ 1 ∶ 1 for training, validation, and independent testing purposes. We developed a multiscale CNN architecture and implemented its 2D and 3D versions to classify pulmonary nodules into two categories, namely true positive and false positive. The performance of the 2D/3D-CNN classification schemes was evaluated using the area under the receiver operating characteristic curves (AUC). The p-values and the 95% confidence intervals (CI) were calculated. Results: The AUC for the optimal 2D-CNN model is 0.9307 (95% CI: 0.9285 to 0.9330) with a sensitivity of 92.70% and a specificity of 76.21%. The 3D-CNN model with the best performance had an AUC of 0.9541 (95% CI: 0.9495 to 0.9583) with a sensitivity of 89.98% and a specificity of 87.30%. The developed multiscale CNN architecture had a better performance than the vanilla architecture did. Conclusions: The 3D-CNN model has a better performance in false-positive reduction compared with its 2D counterpart; however, the improvement is relatively limited and demands more computational resources for training purposes.
- 13 October 2020
- Published by SPIE-Intl Soc Optical Eng in Journal of Medical Imaging
- Vol. 7 (05) , 051202-11
2D CNN versus 3D CNN for false-positive reduction in lung cancer screening
Purpose: To clarify whether and to what extent three-dimensional (3D) convolutional neural network (CNN) is superior to 2D CNN when applied to reduce false-positive nodule detections in the scenario of low-dose computed tomography (CT) lung cancer screening. Approach: We established a dataset consisting of 1600 chest CT examinations acquired on different subjects from various sources. There were in total 18,280 candidate nodules in these CT examinations, among which 9185 were nodules and 9095 were not nodules. For each candidate nodule, we extracted a number of cubic subvolumes with a dimension of 72 × 72 × 72 mm3 by rotating the CT examinations randomly for 25 times prior to the extraction of the axis-aligned subvolumes. These subvolumes were split into three groups in a ratio of 8 ∶ 1 ∶ 1 for training, validation, and independent testing purposes. We developed a multiscale CNN architecture and implemented its 2D and 3D versions to classify pulmonary nodules into two categories, namely true positive and false positive. The performance of the 2D/3D-CNN classification schemes was evaluated using the area under the receiver operating characteristic curves (AUC). The p-values and the 95% confidence intervals (CI) were calculated. Results: The AUC for the optimal 2D-CNN model is 0.9307 (95% CI: 0.9285 to 0.9330) with a sensitivity of 92.70% and a specificity of 76.21%. The 3D-CNN model with the best performance had an AUC of 0.9541 (95% CI: 0.9495 to 0.9583) with a sensitivity of 89.98% and a specificity of 87.30%. The developed multiscale CNN architecture had a better performance than the vanilla architecture did. Conclusions: The 3D-CNN model has a better performance in false-positive reduction compared with its 2D counterpart; however, the improvement is relatively limited and demands more computational resources for training purposes.
- Open Access
Transcriptomics of Bronchoalveolar Lavage Cells Identifies New Molecular Endotypes of Sarcoidosis
Sarcoidosis is a multisystem granulomatous disease of unknown origin with a variable and often unpredictable course and pattern of organ involvement. In this study we sought to identify specific bronchoalveolar lavage (BAL) cell gene expression patterns indicative of distinct disease phenotypic traits. RNA sequencing by Ion Torrent Proton was performed on BAL cells obtained from 215 well characterized patients with pulmonary sarcoidosis enrolled in the multicenter Genomic Research in Alpha-1 Antitrypsin Deficiency and Sarcoidosis (GRADS) study. Weighted Gene Co-expression Network Analysis (WGCNA) and non-parametric statistics were used to analyze genome wide BAL transcriptome. Validation of results was performed using a microarray expression data set of an independent sarcoidosis cohort (Freiburg, Germany (n=50)). Our supervised analysis found associations between distinct transcriptional programs and major pulmonary phenotypic manifestations of sarcoidosis including; TH1 and TH17 pathways associated with hilar lymphadenopathy; TGFB1 and MTOR signaling with parenchymal involvement, and IL7 and IL2 with airway involvement. Our unsupervised analysis revealed gene modules that uncovered four potential sarcoidosis endotypes including hilar lymphadenopathy with increased acute T cell immune response; extraocular organ involvement with PI3K activation pathways; chronic and multiorgan disease with increased immune response pathways; and multiorgan with increased IL-1 and IL-18 immune and inflammatory responses. We validated the occurrence of these endotypes using gene expression, pulmonary function tests and cell differentials from Freiburg. Taken together our results identify BAL gene expression programs that characterize major pulmonary sarcoidosis phenotypes and suggest the presence of distinct disease molecular endotypes. Take home message: Genome wide BAL transcriptomics identified novel gene expression profiles associated with distinct phenotypic traits in sarcoidosis and is suggestive of the presence of novel molecular and clinical sarcoidosis endotypes that could help with further understanding of this heterogenous disease.
Open AccessTranscriptomics of Bronchoalveolar Lavage Cells Identifies New Molecular Endotypes of Sarcoidosis
Sarcoidosis is a multisystem granulomatous disease of unknown origin with a variable and often unpredictable course and pattern of organ involvement. In this study we sought to identify specific bronchoalveolar lavage (BAL) cell gene expression patterns indicative of distinct disease phenotypic traits. RNA sequencing by Ion Torrent Proton was performed on BAL cells obtained from 215 well characterized patients with pulmonary sarcoidosis enrolled in the multicenter Genomic Research in Alpha-1 Antitrypsin Deficiency and Sarcoidosis (GRADS) study. Weighted Gene Co-expression Network Analysis (WGCNA) and non-parametric statistics were used to analyze genome wide BAL transcriptome. Validation of results was performed using a microarray expression data set of an independent sarcoidosis cohort (Freiburg, Germany (n=50)). Our supervised analysis found associations between distinct transcriptional programs and major pulmonary phenotypic manifestations of sarcoidosis including; TH1 and TH17 pathways associated with hilar lymphadenopathy; TGFB1 and MTOR signaling with parenchymal involvement, and IL7 and IL2 with airway involvement. Our unsupervised analysis revealed gene modules that uncovered four potential sarcoidosis endotypes including hilar lymphadenopathy with increased acute T cell immune response; extraocular organ involvement with PI3K activation pathways; chronic and multiorgan disease with increased immune response pathways; and multiorgan with increased IL-1 and IL-18 immune and inflammatory responses. We validated the occurrence of these endotypes using gene expression, pulmonary function tests and cell differentials from Freiburg. Taken together our results identify BAL gene expression programs that characterize major pulmonary sarcoidosis phenotypes and suggest the presence of distinct disease molecular endotypes. Take home message: Genome wide BAL transcriptomics identified novel gene expression profiles associated with distinct phenotypic traits in sarcoidosis and is suggestive of the presence of novel molecular and clinical sarcoidosis endotypes that could help with further understanding of this heterogenous disease.
- Open Access
- 1 May 2018
- Published by Oxford University Press (OUP) in Annals of the American Thoracic Society
- Vol. 15 (5) , 615-621
Radiographic Emphysema, Circulating Bone Biomarkers, and Progressive Bone Mineral Density Loss in Smokers
Rationale: Osteoporosis is common in individuals with chronic obstructive pulmonary disease. Lung-specific factors, including radiographic emphysema, independently associate with low bone mineral density in cross-sectional smoking cohorts. However, factors associated with progressive bone loss in smokers are understudied and largely unknown. Objectives: To determine the relationship between radiographic emphysema, circulating bone metabolism markers, and pulmonary function and accelerated bone mineral density loss in smokers. Methods: Two hundred and forty male and female current and former smokers, 40 years of age or older, underwent baseline and 2-year assessments of pulmonary function, computed tomography–assessed emphysema, dual X-ray absorptiometry–measured bone mineral density, and circulating bone metabolism biomarker levels (type I collagen C-telopeptide [CTX], amino-terminal propeptide of type I procollagen [P1NP]). The association of radiographic emphysema, bone metabolism biomarker levels, and pulmonary function with accelerated hip bone mineral density loss, defined by the 75th percentile of annual hip bone mineral density decline, was determined by logistic regression modeling with adjustment for age, sex, inhaled and intermittent steroid use, active smoking, body mass index, and the presence of baseline low hip bone mineral density. Results: Of those participants with accelerated hip bone mineral density loss, 22% had moderate or severe visually assessed emphysema compared with 7.2% of smokers without accelerated bone mineral density decline. Moderate to severe visually assessed emphysema (odds ratio, 2.84; 95% confidence interval, 1.01–7.98 compared with trace/mild or no visually assessed emphysema) and the 75th percentile of CTX levels (odds ratio, 2.38; 95% confidence interval, 1.20–4.72 compared with CTX levels below the 75th percentile), a marker of bone resorption, were associated with accelerated hip bone mineral density decline after adjustment for covariates and the presence of baseline low hip bone mineral density. FEV1% predicted was not associated with accelerated bone mineral density decline after adjustment for covariates. Multivariate modeling showed moderate to severe visually assessed emphysema, and the 75th percentiles of CTX were independently associated with accelerated hip bone mineral density decline after adjustment for covariates. Conclusions: Emphysema and elevated markers of bone resorption are independently associated with progressive bone mineral density loss in smokers. These clinical markers may guide targeted bone mineral density screening and monitoring in smokers at highest risk.
- 1 May 2018
- Published by Oxford University Press (OUP) in Annals of the American Thoracic Society
- Vol. 15 (5) , 615-621
Open AccessRadiographic Emphysema, Circulating Bone Biomarkers, and Progressive Bone Mineral Density Loss in Smokers
Rationale: Osteoporosis is common in individuals with chronic obstructive pulmonary disease. Lung-specific factors, including radiographic emphysema, independently associate with low bone mineral density in cross-sectional smoking cohorts. However, factors associated with progressive bone loss in smokers are understudied and largely unknown. Objectives: To determine the relationship between radiographic emphysema, circulating bone metabolism markers, and pulmonary function and accelerated bone mineral density loss in smokers. Methods: Two hundred and forty male and female current and former smokers, 40 years of age or older, underwent baseline and 2-year assessments of pulmonary function, computed tomography–assessed emphysema, dual X-ray absorptiometry–measured bone mineral density, and circulating bone metabolism biomarker levels (type I collagen C-telopeptide [CTX], amino-terminal propeptide of type I procollagen [P1NP]). The association of radiographic emphysema, bone metabolism biomarker levels, and pulmonary function with accelerated hip bone mineral density loss, defined by the 75th percentile of annual hip bone mineral density decline, was determined by logistic regression modeling with adjustment for age, sex, inhaled and intermittent steroid use, active smoking, body mass index, and the presence of baseline low hip bone mineral density. Results: Of those participants with accelerated hip bone mineral density loss, 22% had moderate or severe visually assessed emphysema compared with 7.2% of smokers without accelerated bone mineral density decline. Moderate to severe visually assessed emphysema (odds ratio, 2.84; 95% confidence interval, 1.01–7.98 compared with trace/mild or no visually assessed emphysema) and the 75th percentile of CTX levels (odds ratio, 2.38; 95% confidence interval, 1.20–4.72 compared with CTX levels below the 75th percentile), a marker of bone resorption, were associated with accelerated hip bone mineral density decline after adjustment for covariates and the presence of baseline low hip bone mineral density. FEV1% predicted was not associated with accelerated bone mineral density decline after adjustment for covariates. Multivariate modeling showed moderate to severe visually assessed emphysema, and the 75th percentiles of CTX were independently associated with accelerated hip bone mineral density decline after adjustment for covariates. Conclusions: Emphysema and elevated markers of bone resorption are independently associated with progressive bone mineral density loss in smokers. These clinical markers may guide targeted bone mineral density screening and monitoring in smokers at highest risk.
- Open Access
- 4 August 2016
- Published by Springer Nature in BMC Pulmonary Medicine
- Vol. 16 (1) , 1-10
Adiposity influences airway wall thickness and the asthma phenotype of HIV-associated obstructive lung disease: a cross-sectional study
Airflow obstruction, which encompasses several phenotypes, is common among HIV-infected individuals. Obesity and adipose-related inflammation are associated with both COPD (fixed airflow obstruction) and asthma (reversible airflow obstruction) in HIV-uninfected persons, but the relationship to airway inflammation and airflow obstruction in HIV-infected persons is unknown. The objective of this study was to determine if adiposity and adipose-associated inflammation are associated with airway obstruction phenotypes in HIV-infected persons. We performed a cross-sectional analysis of 121 HIV-infected individuals assessed with pulmonary function testing, chest CT scans for measures of airway wall thickness (wall area percent [WA%]) and adipose tissue volumes (mediastinal and subcutaneous), as well as HIV- and adipose-related inflammatory markers. Participants were defined as COPD phenotype (post-bronchodilator FEV1/FVC < lower limit of normal) or asthma phenotype (doctor-diagnosed asthma or bronchodilator response). Pearson correlation coefficients were calculated between adipose measurements, WA%, and pulmonary function. Multivariable logistic and linear regression models were used to determine associations of airflow obstruction and airway remodeling (WA%) with adipose measurements and participant characteristics. Twenty-three (19 %) participants were classified as the COPD phenotype and 33 (27 %) were classified as the asthma phenotype. Body mass index (BMI) was similar between those with and without COPD, but higher in those with asthma compared to those without (mean [SD] 30.7 kg/m2 [8.1] vs. 26.5 kg/m2 [5.3], p = 0.008). WA% correlated with greater BMI (r = 0.55, p < 0.001) and volume of adipose tissue (subcutaneous, r = 0.40; p < 0.001; mediastinal, r = 0.25; p = 0.005). Multivariable regression found the COPD phenotype associated with greater age and pack-years smoking; the asthma phenotype with younger age, female gender, smoking history, and lower adiponectin levels; and greater WA% with greater BMI, younger age, higher soluble CD163, and higher CD4 counts. Adiposity and adipose-related inflammation are associated with an asthma phenotype, but not a COPD phenotype, of obstructive lung disease in HIV-infected persons. Airway wall thickness is associated with adiposity and inflammation. Adipose-related inflammation may play a role in HIV-associated asthma.
- 4 August 2016
- Published by Springer Nature in BMC Pulmonary Medicine
- Vol. 16 (1) , 1-10
Open AccessAdiposity influences airway wall thickness and the asthma phenotype of HIV-associated obstructive lung disease: a cross-sectional study
Airflow obstruction, which encompasses several phenotypes, is common among HIV-infected individuals. Obesity and adipose-related inflammation are associated with both COPD (fixed airflow obstruction) and asthma (reversible airflow obstruction) in HIV-uninfected persons, but the relationship to airway inflammation and airflow obstruction in HIV-infected persons is unknown. The objective of this study was to determine if adiposity and adipose-associated inflammation are associated with airway obstruction phenotypes in HIV-infected persons. We performed a cross-sectional analysis of 121 HIV-infected individuals assessed with pulmonary function testing, chest CT scans for measures of airway wall thickness (wall area percent [WA%]) and adipose tissue volumes (mediastinal and subcutaneous), as well as HIV- and adipose-related inflammatory markers. Participants were defined as COPD phenotype (post-bronchodilator FEV1/FVC < lower limit of normal) or asthma phenotype (doctor-diagnosed asthma or bronchodilator response). Pearson correlation coefficients were calculated between adipose measurements, WA%, and pulmonary function. Multivariable logistic and linear regression models were used to determine associations of airflow obstruction and airway remodeling (WA%) with adipose measurements and participant characteristics. Twenty-three (19 %) participants were classified as the COPD phenotype and 33 (27 %) were classified as the asthma phenotype. Body mass index (BMI) was similar between those with and without COPD, but higher in those with asthma compared to those without (mean [SD] 30.7 kg/m2 [8.1] vs. 26.5 kg/m2 [5.3], p = 0.008). WA% correlated with greater BMI (r = 0.55, p < 0.001) and volume of adipose tissue (subcutaneous, r = 0.40; p < 0.001; mediastinal, r = 0.25; p = 0.005). Multivariable regression found the COPD phenotype associated with greater age and pack-years smoking; the asthma phenotype with younger age, female gender, smoking history, and lower adiponectin levels; and greater WA% with greater BMI, younger age, higher soluble CD163, and higher CD4 counts. Adiposity and adipose-related inflammation are associated with an asthma phenotype, but not a COPD phenotype, of obstructive lung disease in HIV-infected persons. Airway wall thickness is associated with adiposity and inflammation. Adipose-related inflammation may play a role in HIV-associated asthma.
- 1 April 2016
- Published by Wolters Kluwer Health in JAIDS Journal of Acquired Immune Deficiency Syndromes
- Vol. 71 (4) , 420-427
Risk Factors Associated With Quantitative Evidence of Lung Emphysema and Fibrosis in an HIV-Infected Cohort
Introduction: The disease spectrum for HIV-infected individuals has shifted toward comorbid non-AIDS conditions including chronic lung disease, but quantitative image analysis of lung disease has not been performed. Objectives: To quantify the prevalence of structural changes of the lung indicating emphysema or fibrosis on radiographic examination. Methods: A cross-sectional analysis of 510 HIV-infected participants in the multicenter Lung-HIV study was performed. Data collected included demographics, biological markers of HIV, pulmonary function testing, and chest computed tomographic examinations. Emphysema and fibrosis-like changes were quantified on computed tomographic images based on threshold approaches. Results: In our cohort, 69% was on antiretroviral therapy, 13% had a current CD4 cell count less than 200 cells per microliter, 39% had an HIV viral load greater than 500 copies per milliliter, and 25% had at least a trace level of emphysema (defined as >2.5% of voxels <−950HU). Trace emphysema was significantly correlated with age, smoking, and pulmonary function. Neither current CD4 cell count nor HIV viral load was significantly correlated with emphysema. Fibrosis-like changes were detected in 29% of the participants and were significantly correlated with HIV viral load (Pearson correlation coefficient = 0.210; P < 0.05); current CD4 cell count was not associated with fibrosis. In multivariable analyses including age, race, and smoking status, HIV viral load remained significantly correlated with fibrosis-like changes (coefficient = 0.107; P = 0.03). Conclusions: A higher HIV viral load was significantly associated with fibrosis-like changes, possibly indicating early interstitial lung disease, but emphysematous changes were not related to current CD4 cell count or HIV viral load.
- 1 April 2016
- Published by Wolters Kluwer Health in JAIDS Journal of Acquired Immune Deficiency Syndromes
- Vol. 71 (4) , 420-427
Risk Factors Associated With Quantitative Evidence of Lung Emphysema and Fibrosis in an HIV-Infected Cohort
Introduction: The disease spectrum for HIV-infected individuals has shifted toward comorbid non-AIDS conditions including chronic lung disease, but quantitative image analysis of lung disease has not been performed. Objectives: To quantify the prevalence of structural changes of the lung indicating emphysema or fibrosis on radiographic examination. Methods: A cross-sectional analysis of 510 HIV-infected participants in the multicenter Lung-HIV study was performed. Data collected included demographics, biological markers of HIV, pulmonary function testing, and chest computed tomographic examinations. Emphysema and fibrosis-like changes were quantified on computed tomographic images based on threshold approaches. Results: In our cohort, 69% was on antiretroviral therapy, 13% had a current CD4 cell count less than 200 cells per microliter, 39% had an HIV viral load greater than 500 copies per milliliter, and 25% had at least a trace level of emphysema (defined as >2.5% of voxels <−950HU). Trace emphysema was significantly correlated with age, smoking, and pulmonary function. Neither current CD4 cell count nor HIV viral load was significantly correlated with emphysema. Fibrosis-like changes were detected in 29% of the participants and were significantly correlated with HIV viral load (Pearson correlation coefficient = 0.210; P < 0.05); current CD4 cell count was not associated with fibrosis. In multivariable analyses including age, race, and smoking status, HIV viral load remained significantly correlated with fibrosis-like changes (coefficient = 0.107; P = 0.03). Conclusions: A higher HIV viral load was significantly associated with fibrosis-like changes, possibly indicating early interstitial lung disease, but emphysematous changes were not related to current CD4 cell count or HIV viral load.
- 3 March 2011
- Published by SPIE-Intl Soc Optical Eng
- Vol. 7965, 796521
Lung registration using airway tree morphometry
This paper describes a non-linear medical image registration algorithm that aligns lung CT images scanned at different respiratory phases. The method uses landmarks obtained from the airway tree to find the airway branch extension lines and where the lines intersect the lung surface. The branch extension and lung intersection voxels on the surface were the crucial landmarks that initialize the non-rigid registration process. The advantage of these landmarks is that they have high correspondence between the matching patterns in the template images and deformed images. This method was developed and tested on CT examinations from participants in an asthma study. The registration accuracy was evaluated by the average distance between the corresponding airway tree branch points in the pair of images. The mean value of the distance between landmarks in template images and deformed matching images for subjects 1 and 2 were 8.44 mm (±4.46 mm) and 4.33 mm (± 3.78 mm), respectively. The results show that the lung image registration technique developed in this study may prove useful in quantifying longitudinal changes, performing regional analysis, tracking lung tumors, and compensating for subject motion across CT images.
- 3 March 2011
- Published by SPIE-Intl Soc Optical Eng
- Vol. 7965, 796521
Lung registration using airway tree morphometry
This paper describes a non-linear medical image registration algorithm that aligns lung CT images scanned at different respiratory phases. The method uses landmarks obtained from the airway tree to find the airway branch extension lines and where the lines intersect the lung surface. The branch extension and lung intersection voxels on the surface were the crucial landmarks that initialize the non-rigid registration process. The advantage of these landmarks is that they have high correspondence between the matching patterns in the template images and deformed images. This method was developed and tested on CT examinations from participants in an asthma study. The registration accuracy was evaluated by the average distance between the corresponding airway tree branch points in the pair of images. The mean value of the distance between landmarks in template images and deformed matching images for subjects 1 and 2 were 8.44 mm (±4.46 mm) and 4.33 mm (± 3.78 mm), respectively. The results show that the lung image registration technique developed in this study may prove useful in quantifying longitudinal changes, performing regional analysis, tracking lung tumors, and compensating for subject motion across CT images.
- 3 March 2011
- Published by SPIE-Intl Soc Optical Eng
- Vol. 7965, 79651X
Quantitative computed tomography of lung parenchyma in patients with emphysema: analysis of higher-density lung regions
Quantitative computed tomography (CT) has been widely used to detect and evaluate the presence (or absence) of emphysema applying the density masks at specific thresholds, e.g., -910 or -950 Hounsfield Unit (HU). However, it has also been observed that subjects with similar density-mask based emphysema scores could have varying lung function, possibly indicating differences of disease severity. To assess this possible discrepancy, we investigated whether density distribution of "viable" lung parenchyma regions with pixel values > -910 HU correlates with lung function. A dataset of 38 subjects, who underwent both pulmonary function testing and CT examinations in a COPD SCCOR study, was assembled. After the lung regions depicted on CT images were automatically segmented by a computerized scheme, we systematically divided the lung parenchyma into different density groups (bins) and computed a number of statistical features (i.e., mean, standard deviation (STD), skewness of the pixel value distributions) in these density bins. We then analyzed the correlations between each feature and lung function. The correlation between diffusion lung capacity (DLCO) and STD of pixel values in the bin of -910HU ≤ PV < -750HU was -0.43, as compared with a correlation of -0.49 obtained between the post-bronchodilator ratio (FEV1/FVC) measured by the forced expiratory volume in 1 second (FEV1) dividing the forced vital capacity (FVC) and the STD of pixel values in the bin of -1024HU ≤ PV < -910HU. The results showed an association between the distribution of pixel values in "viable" lung parenchyma and lung function, which indicates that similar to the conventional density mask method, the pixel value distribution features in "viable" lung parenchyma areas may also provide clinically useful information to improve assessments of lung disease severity as measured by lung functional tests.
- 3 March 2011
- Published by SPIE-Intl Soc Optical Eng
- Vol. 7965, 79651X
Quantitative computed tomography of lung parenchyma in patients with emphysema: analysis of higher-density lung regions
Quantitative computed tomography (CT) has been widely used to detect and evaluate the presence (or absence) of emphysema applying the density masks at specific thresholds, e.g., -910 or -950 Hounsfield Unit (HU). However, it has also been observed that subjects with similar density-mask based emphysema scores could have varying lung function, possibly indicating differences of disease severity. To assess this possible discrepancy, we investigated whether density distribution of "viable" lung parenchyma regions with pixel values > -910 HU correlates with lung function. A dataset of 38 subjects, who underwent both pulmonary function testing and CT examinations in a COPD SCCOR study, was assembled. After the lung regions depicted on CT images were automatically segmented by a computerized scheme, we systematically divided the lung parenchyma into different density groups (bins) and computed a number of statistical features (i.e., mean, standard deviation (STD), skewness of the pixel value distributions) in these density bins. We then analyzed the correlations between each feature and lung function. The correlation between diffusion lung capacity (DLCO) and STD of pixel values in the bin of -910HU ≤ PV < -750HU was -0.43, as compared with a correlation of -0.49 obtained between the post-bronchodilator ratio (FEV1/FVC) measured by the forced expiratory volume in 1 second (FEV1) dividing the forced vital capacity (FVC) and the STD of pixel values in the bin of -1024HU ≤ PV < -910HU. The results showed an association between the distribution of pixel values in "viable" lung parenchyma and lung function, which indicates that similar to the conventional density mask method, the pixel value distribution features in "viable" lung parenchyma areas may also provide clinically useful information to improve assessments of lung disease severity as measured by lung functional tests.
- 29 June 2010
- Published by Wiley in Medical Physics
- Vol. 37 (7Part1) , 3802-3808
Computerized comprehensive data analysis of Lung Imaging Database Consortium (LIDC)
Purpose: Lung Image Database Consortium (LIDC) is the largest public CT image database of lung nodules. In this study, the authors present a comprehensive and the most updated analysis of this dynamically growing database under the help of a computerized tool, aiming to assist researchers to optimally use this database for lung cancer related investigations.Methods: The authors developed a computer scheme to automatically match the nodule outlines marked manually by radiologists on CT images. A large variety of characteristics regarding the annotated nodules in the database including volume, spiculation level, elongation, interobserver variability, as well as the intersection of delineated nodule voxels and overlapping ratio between the same nodules marked by different radiologists are automatically calculated and summarized. The scheme was applied to analyze all 157 examinations with complete annotation data currently available in LIDC dataset.Results: The scheme summarizes the statistical distributions of the abovementioned geometric and diagnosis features. Among the 391 nodules, (1) 365 (93.35%) have principal axis length ; (2) 120, 75, 76, and 120 were marked by one, two, three, and four radiologists, respectively; and (3) 122 (32.48%) have the maximum volume overlapping ratios for the delineations of two radiologists, while 198 (50.64%) have the maximum volume overlapping ratios . The results also showed that 72.89% of the nodules were assessed with malignancy score between 2 and 4, and only 7.93% of these nodules were considered as severely malignant (malignancy ).Conclusions: This study demonstrates that LIDC contains examinations covering a diverse distribution of nodule characteristics and it can be a useful resource to assess the performance of the nodule detection and/or segmentation schemes.
- 29 June 2010
- Published by Wiley in Medical Physics
- Vol. 37 (7Part1) , 3802-3808
Computerized comprehensive data analysis of Lung Imaging Database Consortium (LIDC)
Purpose: Lung Image Database Consortium (LIDC) is the largest public CT image database of lung nodules. In this study, the authors present a comprehensive and the most updated analysis of this dynamically growing database under the help of a computerized tool, aiming to assist researchers to optimally use this database for lung cancer related investigations.Methods: The authors developed a computer scheme to automatically match the nodule outlines marked manually by radiologists on CT images. A large variety of characteristics regarding the annotated nodules in the database including volume, spiculation level, elongation, interobserver variability, as well as the intersection of delineated nodule voxels and overlapping ratio between the same nodules marked by different radiologists are automatically calculated and summarized. The scheme was applied to analyze all 157 examinations with complete annotation data currently available in LIDC dataset.Results: The scheme summarizes the statistical distributions of the abovementioned geometric and diagnosis features. Among the 391 nodules, (1) 365 (93.35%) have principal axis length ; (2) 120, 75, 76, and 120 were marked by one, two, three, and four radiologists, respectively; and (3) 122 (32.48%) have the maximum volume overlapping ratios for the delineations of two radiologists, while 198 (50.64%) have the maximum volume overlapping ratios . The results also showed that 72.89% of the nodules were assessed with malignancy score between 2 and 4, and only 7.93% of these nodules were considered as severely malignant (malignancy ).Conclusions: This study demonstrates that LIDC contains examinations covering a diverse distribution of nodule characteristics and it can be a useful resource to assess the performance of the nodule detection and/or segmentation schemes.
- 4 March 2010
- Published by SPIE-Intl Soc Optical Eng
- Vol. 7624, 76242E
Improving performance and reliability of interactive CAD schemes
An interactive computer-aided detection or diagnosis (ICAD) scheme allows observers to query suspicious abnormalities (lesions) depicted on medical images. Once a suspicious region is queried, ICAD segments the abnormal region, computes a set of image features, searches for and identifies the reference regions depicted on the verified lesions that are similar to the queried one. Based on the distribution of the selected similar regions, ICAD generates a detection (or classification) score of the queried region depicting true-positive disease. In this study, we assessed the performance and reliability of an ICAD scheme when using a database including total 1500 positive images depicted verified breast masses and 1500 negative images depicted ICAD-cued false-positive regions as well as the leave-one-out testing method. We conducted two experiments. In the first experiment, we tested the relationship between ICAD performance and the size of reference database by systematically increasing the size of reference database from 200 to 3000 images. In the second experiment, we tested the relationship between ICAD performance and the similarity level between the queried image and the retrieved similar references by applying a set of thresholds to systematically remove the queried images whose similarity level to their most "similar" reference images are lower than threshold. The performance was compared based on the areas under ROC curves (AUC). The results showed that (1) as the increase of reference database, AUC value monotonically increased from 0.636±0.041 to 0.854±0.004 and (2) as the increase of similarity threshold values, AUC value also monotonically increased from 0.854±0.004 to 0.932±0.016. The increase of AUC values and the decrease of their standard deviations indicate the improvement of both CAD performance and reliability. The study suggested that (1) assembling the large and diverse reference databases and (2) assessing and reporting the reliability of ICAD-generated results based on the similarity measurement are important in development and application of the ICAD schemes.
- 4 March 2010
- Published by SPIE-Intl Soc Optical Eng
- Vol. 7624, 76242E
Improving performance and reliability of interactive CAD schemes
An interactive computer-aided detection or diagnosis (ICAD) scheme allows observers to query suspicious abnormalities (lesions) depicted on medical images. Once a suspicious region is queried, ICAD segments the abnormal region, computes a set of image features, searches for and identifies the reference regions depicted on the verified lesions that are similar to the queried one. Based on the distribution of the selected similar regions, ICAD generates a detection (or classification) score of the queried region depicting true-positive disease. In this study, we assessed the performance and reliability of an ICAD scheme when using a database including total 1500 positive images depicted verified breast masses and 1500 negative images depicted ICAD-cued false-positive regions as well as the leave-one-out testing method. We conducted two experiments. In the first experiment, we tested the relationship between ICAD performance and the size of reference database by systematically increasing the size of reference database from 200 to 3000 images. In the second experiment, we tested the relationship between ICAD performance and the similarity level between the queried image and the retrieved similar references by applying a set of thresholds to systematically remove the queried images whose similarity level to their most "similar" reference images are lower than threshold. The performance was compared based on the areas under ROC curves (AUC). The results showed that (1) as the increase of reference database, AUC value monotonically increased from 0.636±0.041 to 0.854±0.004 and (2) as the increase of similarity threshold values, AUC value also monotonically increased from 0.854±0.004 to 0.932±0.016. The increase of AUC values and the decrease of their standard deviations indicate the improvement of both CAD performance and reliability. The study suggested that (1) assembling the large and diverse reference databases and (2) assessing and reporting the reliability of ICAD-generated results based on the similarity measurement are important in development and application of the ICAD schemes.
- 30 November 2006
- Published by Elsevier in Academic Radiology
- Vol. 13 (11) , 1379-1386
Computerized Estimation of the Lung Volume Removed During Lung Volume Reduction Surgery
- 30 November 2006
- Published by Elsevier in Academic Radiology
- Vol. 13 (11) , 1379-1386
Computerized Estimation of the Lung Volume Removed During Lung Volume Reduction Surgery