Visual scoring of osteoporosis on low-dose CT in lung cancer screening population.

OBJECTIVES: The risk factors for lung cancer screening eligibility, age as well as smoking history, are also present for osteoporosis. This study aims to develop a visual scoring system to identify osteoporosis that can be applied to low-dose CT scans obtained for lung cancer screening. MATERIALS AND METHODS: We retrospectively reviewed 1000 prospectively enrolled participants in the lung cancer screening program at the Mount Sinai Hospital. Optimal window width and level settings for the visual assessment were chosen based on a previously described approach. Visual scoring of osteoporosis and automated measurement using dedicated software were compared. Inter-reader agreement was conducted using six readers with different levels of experience who independently visually assessed 30 CT scans. RESULTS: Based on previously validated formulas for choosing window and level settings, we chose osteoporosis settings of Width = 230 and Level = 80. Of the 1000 participants, automated measurement was successfully performed on 774 (77.4 %). Among these, 138 (17.8 %) had osteoporosis. There was a significant correlation between the automated measurement and the visual score categories for osteoporosis (Kendall's Tau = -0.64, p < 0.0001; Spearman's rho = -0.77, p < 0.0001). We also found substantial to excellent inter-reader agreement on the osteoporosis classification among the 6 radiologists (Fleiss κ = 0.91). CONCLUSIONS: Our study shows that a simple approach of applying specific window width and level settings to already reconstructed sagittal images obtained in the context of low-dose CT screening for lung cancer is highly feasible and useful in identifying osteoporosis.

Evaluation of emphysema on thoracic low-dose CTs through attention-based multiple instance deep learning.

In addition to lung cancer, other thoracic abnormalities, such as emphysema, can be visualized within low-dose CT scans that were initially obtained in cancer screening programs, and thus, opportunistic evaluation of these diseases may be highly valuable. However, manual assessment for each scan is tedious and often subjective, thus we have developed an automatic, rapid computer-aided diagnosis system for emphysema using attention-based multiple instance deep learning and 865 LDCTs. In the task of determining if a CT scan presented with emphysema or not, our novel Transfer AMIL approach yielded an area under the ROC curve of 0.94 ± 0.04, which was a statistically significant improvement compared to other methods evaluated in our study following the Delong Test with correction for multiple comparisons. Further, from our novel attention weight curves, we found that the upper lung demonstrated a stronger influence in all scan classes, indicating that the model prioritized upper lobe information. Overall, our novel Transfer AMIL method yielded high performance and provided interpretable information by identifying slices that were most influential to the classification decision, thus demonstrating strong potential for clinical implementation.

Combined aortic valve and coronary artery calcifications in lung cancer screening as predictors of death from cardiovascular disease.

OBJECTIVES: Smoking is a major risk factor for both cardiovascular disease (CVD) and lung cancer. Aortic valve calcification (AVC) and coronary artery calcification (CAC) are both due to atherosclerotic disease. We aim to investigate whether AVC on low-dose CT (LDCT) predicts death from CVD in smokers beyond that provided by CAC. METHODS: We reviewed a prospective cohort of 8618 smokers enrolled in LDCT screening for lung cancer in New York State between June 2000 and December 2005. As of December 2009, 169 of the 643 deaths were due to CVD; median follow-up time was 96.4 months. Visual AVC was assessed as being absent (AVC = 0) or present (AVC > 0). CAC ordinal scores of 0-12 were categorized into three validated prognostic categories (0, 1-3, and 4-12). Cox proportional hazards regression analysis was used to assess whether AVC > 0 increased the risk of CVD death, after adjustment for CAC categories and other risk factors. RESULTS: The prevalence of AVC significantly increased (p < 0.0001) with the increasing severity of the CAC categories; Pearson, Spearman, and Kendall's correlation coefficients showed a significant correlation between AVC and CAC with r = 0.29, ρ = 0.32, and τB = 0.28 (all p values < 0.0001), respectively. CAC and AVC were significant predictors of CVD death when considered alone using multivariable Cox regression analysis (adjusted HR of CAC = 1.57, p = 0.04; adjusted HR of AVC = 1.39, p = 0.045). When AVC > 0 and CAC ≥ 4, the hazard ratio was 2.35 (95%CI 1.57-3.50) compared with the reference group of AVC = 0 and CAC < 4, when adjusted for other risk factors. CONCLUSIONS: The presence of AVC identified on LDCT is a significant predictor of future CVD death, particularly for those with ordinal CAC score ≥ 4. KEY POINTS: • Aortic valve calcification (AVC) and coronary artery calcification (CAC) are both due to atherosclerotic disease. The prevalence of AVC in lung cancer screening cohort significantly increased with the increasing severity of CAC. • CAC and AVC were significant predictors of cardiovascular disease (CVD) death when considered alone. Participants who underwent lung cancer screening with AVC > 0 and CAC ≥ 4 had more than a 2-fold increased risk of CVD death than the group with AVC = 0 and CAC < 4, when adjusted for other risk factors.

Visual scoring of aortic valve calcifications on low-dose CT in lung cancer screening.

OBJECTIVES: To evaluate risk factors for prevalence and progression of aortic valve calcification (AVC) in lung cancer screening participants and also to assess the sensitivity and reliability of visual AVCs on low-dose CT (LDCT) for predicting aortic stenosis (AS) in high-risk smokers. METHODS: We reviewed 1225 consecutive participants in annual LDCT screening for lung cancer at the Mount Sinai Hospital between 2010 and 2017. Sensitivity and specificity of moderate/severe AVC score on LDCT to identify AS on echocardiogram were calculated for 126 participants who had both within 12 months. Using regression analyses, risk factors for AVC at baseline, for progression, and for new AVC on annual rounds of screening were identified. Reliability of AVC assessment on LDCT was assessed by comparing visual AVC scores (1) with standard-dose, electrocardiography (ECG)-gated CT for 31 participants who had both within 12 months and (2) with Agatston scores of 1225 participants and by determining (3) the intra-reader agreement of 1225 participants. RESULTS: Visual AVC scores on LDCT had substantial agreement with the severity of AS on echocardiography and substantial inter-observer and excellent intra-observer agreement. Sensitivity and specificity of moderate/severe visual AVC scores for moderate/severe AS on echocardiogram were 100% and 94%, respectively. Significant predictors for baseline AVC were male sex (OR = 2.52), age (OR10 years = 2.87), and coronary artery calcification score (OR = 1.18), the significant predictor for AVC progression after baseline was pack-years of smoking (HR10 packyears = 1.14), and significant predictors for new AVC on annual LDCT were male sex (HR = 1.51), age (HR10 years = 2.17), CAC (HR = 1.09)  and BMI (HR = 1.06). CONCLUSIONS: AVC scores on LDCT should be documented, especially in lung cancer screening program. KEY POINTS: • LDCT screening for lung cancer provides an opportunity to identify lung cancer and cardiovascular disease in asymptomatic smokers. • Visual aortic valve calcification scores could be reliably evaluated on LDCT and had substantial agreement with the severity of aortic valve stenosis on echocardiography. • Sensitivity and specificity of moderate/severe visual AVC scores on LDCT for moderate/severe AS on echocardiogram were 100% and 94%, respectively.

Low-Dose CT Screening for Lung Cancer: Computer-aided Detection of Missed Lung Cancers.

Purpose To update information regarding the usefulness of computer-aided detection (CAD) systems with a focus on the most critical category, that of missed cancers at earlier imaging, for cancers that manifest as a solid nodule. Materials and Methods By using a HIPAA-compliant institutional review board-approved protocol where informed consent was obtained, 50 lung cancers that manifested as a solid nodule on computed tomographic (CT) scans in annual rounds of screening (time 1) were retrospectively identified that could, in retrospect, be identified on the previous CT scans (time 0). Four CAD systems were compared, which were referred to as CAD 1, CAD 2, CAD 3, and CAD 4. The total number of accepted CAD-system-detected nodules at time 0 was determined by consensus of two radiologists and the number of CAD-system-detected nodules that were rejected by the radiologists was also documented. Results At time 0 when all the cancers had been missed, CAD system detection rates for the cancers were 56%, 70%, 68%, and 60% (κ = 0.45) for CAD systems 1, 2, 3, and 4, respectively. At time 1, the rates were 74%, 82%, 82%, and 78% (κ = 0.32), respectively. The average diameter of the 50 cancers at time 0 and time 1 was 4.8 mm and 11.4 mm, respectively. The number of CAD-system-detected nodules that were rejected per CT scan for CAD systems 1-4 at time 0 was 7.4, 1.7, 0.6, and 4.5 respectively. Conclusion CAD systems detected up to 70% of lung cancers that were not detected by the radiologist but failed to detect about 20% of the lung cancers when they were identified by the radiologist, which suggests that CAD may be useful in the role of second reader. (©) RSNA, 2016.

Segmentation of juxtapleural pulmonary nodules using a robust surface estimate.

An algorithm was developed to segment solid pulmonary nodules attached to the chest wall in computed tomography scans. The pleural surface was estimated and used to segment the nodule from the chest wall. To estimate the surface, a robust approach was used to identify points that lie on the pleural surface but not on the nodule. A 3D surface was estimated from the identified surface points. The segmentation performance of the algorithm was evaluated on a database of 150 solid juxtapleural pulmonary nodules. Segmented images were rated on a scale of 1 to 4 based on visual inspection, with 3 and 4 considered acceptable. This algorithm offers a large improvement in the success rate of juxtapleural nodule segmentation, successfully segmenting 98.0% of nodules compared to 81.3% for a previously published plane-fitting algorithm, which will provide for the development of more robust automated nodule measurement methods.

Zone of transition: a potential source of error in tumor volume estimation.

PURPOSE: To measure the width of the zone of transition (ZOT) between nonaerated solid tumor and surrounding nonneoplastic lung parenchyma and determine the extent to which ZOT influences computer-derived estimates of tumor volume based on computed tomographic (CT) images. MATERIALS AND METHODS: This HIPAA-compliant study was approved by the institutional research board. The histologic slide containing the maximum tumor area was digitized for 20 consecutive patients with solid adenocarcinoma. The outer border of the tumor (A2) was marked; it included all lung parenchyma having any tumor cells. The inner border of the tumor (A1) was marked; it included only solid tumor where lung parenchyma was no longer preserved. Assuming two circles with areas of A2 and A1, the corresponding two radii, R2 and R1, were calculated. The average width of the ZOT was defined as R2 minus R1. The relationship between ZOT and tumor diameter on the CT images prior to surgery was assessed by using regression analysis. The relationship between ZOT and tumor volume was assessed by using a theoretical model of idealized spheres with varying diameters. RESULTS: The mean width of the ZOT was 0.78 mm (median, 0.48 mm). The proportional effect of ZOT on tumor volume estimates decreased with increasing tumor diameter and increased with increasing width of ZOT. Correlation between ZOT and tumor diameter was not significant (P = .87). CONCLUSION: The average width of ZOT is about a single pixel width on a full field of view CT scan; thus, the ZOT can have a large effect on volume estimates, particularly for small tumors.

A comparison of ground truth estimation methods.

PURPOSE: Knowledge of the exact shape of a lesion, or ground truth (GT), is necessary for the development of diagnostic tools by means of algorithm validation, measurement metric analysis, accurate size estimation. Four methods that estimate GTs from multiple readers' documentations by considering the spatial location of voxels were compared: thresholded Probability-Map at 0.50 (TPM(0.50)) and at 0.75 (TPM(0.75)), simultaneous truth and performance level estimation (STAPLE) and truth estimate from self distances (TESD). METHODS: A subset of the publicly available Lung Image Database Consortium archive was used, selecting pulmonary nodules documented by all four radiologists. The pair-wise similarities between the estimated GTs were analyzed by computing the respective Jaccard coefficients. Then, with respect to the readers' marking volumes, the estimated volumes were ranked and the sign test of the differences between them was performed. RESULTS: (a) the rank variations among the four methods and the volume differences between STAPLE and TESD are not statistically significant, (b) TPM(0.50) estimates are statistically larger (c) TPM(0.75) estimates are statistically smaller (d) there is some spatial disagreement in the estimates as the one-sided 90% confidence intervals between TPM(0.75) and TPM(0.50), TPM(0.75) and STAPLE, TPM(0.75) and TESD, TPM(0.50) and STAPLE, TPM(0.50) and TESD, STAPLE and TESD, respectively, show: [0.67, 1.00], [0.67, 1.00], [0.77, 1.00], [0.93, 1.00], [0.85, 1.00], [0.85, 1.00]. CONCLUSIONS: The method used to estimate the GT is important: the differences highlighted that STAPLE and TESD, notwithstanding a few weaknesses, appear to be equally viable as a GT estimator, while the increased availability of computing power is decreasing the appeal afforded to TPMs. Ultimately, the choice of which GT estimation method, between the two, should be preferred depends on the specific characteristics of the marked data that is used with respect to the two elements that differentiate the method approaches: relative reliabilities of the readers and the reliability of the region boundaries.

Automated nodule location and size estimation using a multi-scale Laplacian of Gaussian filtering approach.

Estimation of nodule location and size is an important pre-processing step in some nodule segmentation algorithms to determine the size and location of the region of interest. Ideally, such estimation methods will consistently find the same nodule location regardless of where the the seed point (provided either manually or by a nodule detection algorithm) is placed relative to the "true" center of the nodule, and the size should be a reasonable estimate of the true nodule size. We developed a method that estimates nodule location and size using multi-scale Laplacian of Gaussian (LoG) filtering. Nodule candidates near a given seed point are found by searching for blob-like regions with high filter response. The candidates are then pruned according to filter response and location, and the remaining candidates are sorted by size and the largest candidate selected. This method was compared to a previously published template-based method. The methods were evaluated on the basis of stability of the estimated nodule location to changes in the initial seed point and how well the size estimates agreed with volumes determined by a semi-automated nodule segmentation method. The LoG method exhibited better stability to changes in the seed point, with 93% of nodules having the same estimated location even when the seed point was altered, compared to only 52% of nodules for the template-based method. Both methods also showed good agreement with sizes determined by a nodule segmentation method, with an average relative size difference of 5% and -5% for the LoG and template-based methods respectively.

Vessel diameter measurement from intravital microscopy.

The blood vessel diameter is often measured in microcirculation studies to quantify the effects of various stimuli. Intravital video microscopy is used to measure the change in vessel diameter by first recording the video and analyzing it using electronic calipers or by using image shearing technique. Manual measurement using electronic calipers or image shearing is time-consuming and prone to measurement error, and automated measurement can serve as an alternative that is faster and more reliable. In this paper, a new feature-based tracking algorithm is presented for automatically measuring diameter of vessels in intravital video microscopy image sequences. Our method tracks the vessel diameter throughout the entire image sequence once the diameter is marked in the first image. The parameters were calibrated using the intravital videos with manual ground truth measurements. The experiment with 10 synthetic videos and 20 intravital microscopy videos, including 10 fluorescence confocal and 10 non-confocal transmission, shows that the measurement can be performed accurately.