Search

Deep learning methods have been widely and successfully applied to the medical imaging field. Specifically, fully convolutional neural networks have become the state-of-the-art supervised segmentation method in a variety of biomedical segmentation problems. Two fully convolutional networks were proposed to sequentially achieve accurate lobar segmentation. Firstly, a 2D ResNet-101 based network is proposed for lung segmentation and 575 chest CT scans from multicenter clinical trials were used with radiologist approved lung segmentation. Secondly, a 3D DenseNet based network is applied to segment the 5 lobes and a total of 1280 different CT scans were used with radiologist approved lobar segmentation as ground truth. The dataset includes various pathological lung diseases and stratified sampling was used to form training and test sets following a ratio of 4:1 to ensure a balanced number and type of abnormality present. A 3D CNN segmentation model was also built for lung segmentation to investigate the feasibility using current hardware. Using 5-fold cross validation a mean Dice coefficient of 0.988 ± 0.012 and Average Surface Distance of 0.562 ± 0.49 mm was achieved by the proposed 2D CNN on lung segmentation. 3D DenseNet on lobar segmentation achieved Dice score of 0.959 ± 0.087 and Average surface distance of 0.873 ± 0.61 mm.

A clinical validation of the bone scan lesion area (BSLA) as a quantitative imaging biomarker was performed in metastatic castration-resistant prostate cancer (mCRPC). BSLA was computed from whole-body bone scintigraphy at baseline and week 12 posttreatment in a cohort of 198 mCRPC subjects (127 treated and 71 placebo) from a clinical trial involving a different drug from the initial biomarker development. BSLA computation involved automated image normalization, lesion segmentation, and summation of the total area of segmented lesions on bone scan AP and PA views as a measure of tumor burden. As a predictive biomarker, treated subjects with baseline BSLA [Formula: see text] had longer survival than those with higher BSLA ([Formula: see text] and [Formula: see text]). As a surrogate outcome biomarker, subjects were categorized as progressive disease (PD) if the BSLA increased by a prespecified 30% or more from baseline to week 12 and non-PD otherwise. Overall survival rates between PD and non-PD groups were statistically different ([Formula: see text] and [Formula: see text]). Subjects without PD at week 12 had longer survival than subjects with PD: median 398 days versus 280 days. BSLA has now been demonstrated to be an early surrogate outcome for overall survival in different prostate cancer drug treatments.

Rationale and Objectives Median survival of patients with idiopathic pulmonary fibrosis (IPF) is 2–5 years. Sensitive imaging metrics can play a role in detecting early changes in therapeutic development. The aim of the present study was to compare known computed tomography (CT) histogram kurtosis and a classifier-based quantitative score to assess baseline severity and change over time in patients with IPF. Materials and Methods A total of 57 patients with at least baseline and paired follow-up scans were selected from an imaging database of standardized CT scans obtained from patients with IPF. CT histogram measurement of kurtosis and quantitative lung fibrosis (QLF) and quantitative interstitial lung disease (QILD) scores from a classification algorithm were calculated. Spearman rank correlations were used to assess associations between baseline severity and changes for all CT-derived measures compared to forced vital capacity (FVC) and carbon monoxide diffusion capacity (DLCO) (percent predicted). Results At baseline, mean (±SD) of kurtosis was 2.43 (±1.83). Mean (±SD) values of QLF and QILD scores were 20.7% (±13.4) and 43.3% (±20.0), respectively. All baseline histogram indices and QLF and QILD scores were correlated well with baseline FVC and DLCO. When assessing associations with changes in FVC and DLCO over time, only QLF score was statistically significant (ρ = −0.57; P

Background: Patients with IPF experience varying degrees of decline in lung function over time. The rate of decline can vary between patients and across over time and at present there is no reliable way to predict which subjects will decline faster. Aim: This study explores whether short term change in QLF score can predict fast from vs slow FVC decline. Methods: In an open-label phase 2 trial in patients with IPF subjects underwent standardized HRCT scans (QLF) and spirometry (FVC) performed at baseline and every 24 and 12 weeks, respectively. QLF was measured whore for the whole lung and worst lobe. Mixed effect models were used to estimate and compare against the reduction in FVC over repeated measurements with linear or quadratic growth models. Results: The aAnnual rate of reduction in mean FVC was 3.5% for the Week 24week QLF changes of Conclusion: Change in QLF of 4 percent at 6 months isn predictive of rapid FVC decline within 12 to 15 months

Objectives: To investigate regional transitions of interstitial lung disease patterns on HRCT and their association with symptom scores in patients with Idiopathic Pulmonary Fibrosis (IPF). This work uses mapping of voxels between baseline and follow-up scans to assess highly localized transitions in lung patterns. Methods: Anonymized HRCT images were obtained from 25 IPF subjects who had a baseline and follow-up surveillance HRCT scan at 6 months. A transitional interstitial lung disease (TILD) matrix of pattern transitions was computed for each paired of subject scans using the following steps: (1) lobar segmentation and classification of each voxel as one of 4 lung patterns (fibrotic, honeycomb, ground glass, or normal); (2) registration (mapping) of corresponding voxels between the paired scans; (3) for each mapped pair of voxels recording a count in the appropriate cell of the 4x4 TILD matrix; (4) summarizing the pattern transitions as percentages of total voxels (TILD scores). Associations between TILD scores and symptom scores, as well as comparisons between worse and non-worse groups, were performed. Results: TILD scores indicating worsening from ground glass to fibrotic patterns were associated with increasing the shortness of breath (rho=0.48; p=0.016). Significant differences in TILD scores were found between worse and non-worse groups in the shortness of breath (p=0.014). When symptoms were not worse the TILD scores were minimal in the right and left upper lobes. Conclusions: Using voxel-by-voxel TILD scores on paired HRCT scans 6 months apart, there were obvious transitions to worsening lung patterns when symptoms got worse.

Background: There is increasing interest in determining treatment efficacy early in the setting of patients with IPF. This is important in design of clinical trials in which decline in FVC is used to assess PFS as an endpoint. Aim: This study explores whether change in Quantitative Lung Fibrosis (QLF) score or FVC at 24 weeks is more sensitive in predicting PFS (where progression is defined as a decline in of > 10%) . Methods: In an open-label phase 2 trial in patients with idiopathic pulmonary fibrosis subjects underwent standardized HRCT scans (QLF) and spirometry (FVC) at baseline and and every 24 and 12 weeks respectively. Cox regression models and Kaplan-Meier (KM) Progression-Free-Survival plots was were performed to evaluate 24 week change in QLF of 4%, 3%, and 2% and FVC of 5% as a predictor of PFS. Harrell’s C index test was performed to assess the validity of the models. Results: The nNumber of subjects with both QLF scores and FVC at baseline, weeks 24, 48, 72 and 96 were 73, 66, 32, and 24 respectively. For the whole lung the hazard ratios for change in QLF of 4, 3, and 2% wereas 3.65, 3.6 and 3.7 respectively (all p Conclusion: A 24 week change in QLF > 2 % (worst lobe) rather change in FVC is a stronger predictor of PFS defined by FVC decline defined PFS in clinical trials.

Author(s): Young, Stefano; Lo, Pechin; Kim, Grace; Brown, Matthew; Hoffman, John; Hsu, William; Wahi-Anwar, Wasil; Flores, Carlos; Lee, Grace; Noo, Frederic; Goldin, Jonathan; McNitt-Gray, Michael | Abstract: PurposeLung cancer screening with low-dose CT has recently been approved for reimbursement, heralding the arrival of such screening services worldwide. Computer-aided detection (CAD) tools offer the potential to assist radiologists in detecting nodules in these screening exams. In lung screening, as in all CT exams, there is interest in further reducing radiation dose. However, the effects of continued dose reduction on CAD performance are not fully understood. In this work, we investigated the effect of reducing radiation dose on CAD lung nodule detection performance in a screening population.MethodsThe raw projection data files were collected from 481 patients who underwent low-dose screening CT exams at our institution as part of the National Lung Screening Trial (NLST). All scans were performed on a multidetector scanner (Sensation 64, Siemens Healthcare, Forchheim Germany) according to the NLST protocol, which called for a fixed tube current scan of 25 effective mAs for standard-sized patients and 40 effective mAs for larger patients. The raw projection data were input to a reduced-dose simulation software to create simulated reduced-dose scans corresponding to 50% and 25% of the original protocols. All raw data files were reconstructed at the scanner with 1 mm slice thickness and B50 kernel. The lungs were segmented semi-automatically, and all images and segmentations were input to an in-house CAD algorithm trained on higher dose scans (75-300 mAs). CAD findings were compared to a reference standard generated by an experienced reader. Nodule- and patient-level sensitivities were calculated along with false positives per scan, all of which were evaluated in terms of the relative change with respect to dose. Nodules were subdivided based on size and solidity into categories analogous to the LungRADS assessment categories, and sub-analyses were performed.ResultsFrom the 481 patients in this study, 82 had at least one nodule (prevalence of 17%) and 399 did not (83%). A total of 118 nodules were identified. Twenty-seven nodules (23%) corresponded to LungRADS category 4 based on size and composition, while 18 (15%) corresponded to LungRADS category 3 and 73 (61%) corresponded to LungRADS category 2. For solid nodules ≥8 mm, patient-level median sensitivities were 100% at all three dose levels, and mean sensitivities were 72%, 63%, and 63% at original, 50%, and 25% dose, respectively. Overall mean patient-level sensitivities for nodules ranging from 3 to 45 mm were 38%, 37%, and 38% at original, 50%, and 25% dose due to the prevalence of smaller nodules and nonsolid nodules in our reference standard. The mean false-positive rates were 3, 5, and 13 per case.ConclusionsCAD sensitivity decreased very slightly for larger nodules as dose was reduced, indicating that reducing the dose to 50% of original levels may be investigated further for use in CT screening. However, the effect of dose was small relative to the effect of the nodule size and solidity characteristics. The number of false positives per scan increased substantially at 25% dose, illustrating the importance of tuning CAD algorithms to very challenging, high-noise screening exams.

For pediatric oncology patients, CT scans are performed to assess treatment response and disease progression. CAD may be used to detect lung nodules which would reflect metastatic disease. The purpose of this study was to investigate the effects of reducing radiation dose and varying slice thickness on CAD performance in the detection of solid lung nodules in pediatric patients. The dataset consisted of CT scans of 58 pediatric chest cases, from which 7 cases had lung nodules detected by radiologist, and a total of 28 nodules were marked. For each case, the original raw data (sinogram data) was collected and a noise addition model was used to simulate reduced-dose scans of 50%, 25% and 10% of the original dose. In addition, the original and reduced-dose raw data were reconstructed at slice thicknesses of 1.5 and 3 mm using a medium sharp (B45) kernel; the result was eight datasets (4 dose levels x 2 thicknesses) for each case An in-house CAD tool was applied on all reconstructed scans, and results were compared with the radiologist’s markings. Patient level mean sensitivities at 3mm thickness were 24%, 26%, 25%, 27%, and at 1.5 mm thickness were 23%, 29%, 35%, 36% for 10%, 25%, 50%, and 100% dose level, respectively. Mean FP numbers were 1.5, 0.9, 0.8, 0.7 at 3 mm and 11.4, 3.5, 2.8, 2.8 at 1.5 mm thickness for 10%, 25%, 50%, and 100% dose level respectively. CAD sensitivity did not change with dose level for 3mm thickness, but did change with dose for 1.5 mm. False Positives increased at low dose levels where noise values were high.

Background and objective Endobronchial valve (EBV) therapy is optimized in patients who demonstrate little or no collateral ventilation (CV). The accuracy of the Chartis System and visual assessment of high-resolution computerized tomography (HRCT) fissure completeness by a core radiology laboratory for classifying CV status was compared by evaluating the relationship of each method with target lobe volume reduction (TLVR) after EBV placement. Methods Retrospective HRCT fissure analysis of a study population who underwent catheter-based measurement of CV followed by complete occlusion of the targeted lobe by EBV. Accuracy, sensitivity, specificity, positive predictive value and negative predictive value of the HRCT fissure analysis and the catheter-based measurement of CV for predicting TLVR was determined. Results Accuracy for correctly classifying TLVR with EBV was similar for Chartis System and HRCT fissure analysis (74 vs 77%). The sensitivity and specificity of the Chartis measurement were 86% and 61% and those of HRCT fissure analysis 75% and 79%. Patients with TLVR >= 350 mL had statistically significant improvement in respiratory function, exercise performance and quality of life measures. Conclusions When evaluating patients for likelihood of successful EBV therapy, the Chartis System CV assessment and HRCT fissure analysis appear to have comparable accuracy. Both techniques were found to be beneficial for EBV procedure planning.

Introduction High resolution computed tomography (HRCT) plays an indispensable role in the diagnosis of Idiopathic Pulmonary Fibrosis (IPF). Due to the short median survival of 2 to 5 years, understanding the chance of progression in the next follow-up of a patient is critical. The aim of this study is to test the predictability of progression in IPF of the near follow-up of a patient using the early quantitative changes in HRCT scans. Methods Volumetric anonymized HRCT scans were obtained. Two data sets were collected: a pilot study with 35 subjects in three sequential scans (baseline, 6 and 12 month scans) and validation study involving 157 subjects with clinical information. Quantitative lung fibrosis (QLF) scores were calculated for HRCT scans using an automated system. A log-rank test and Cox regression were used to test a prediction with early changes in QLF. Results In the pilot study mean changes in QLF increased by 4.33% at 6 months and 14.83% at 12 months, when the visual assessments changed from stable to worse. Mean QLF changes were ~1% for the stable cases during 12 months. Using 4% QLF differences at 6 months in the validation cohort, significant differences in progression were found (p Conclusion Early changes in a quantitative score from HRCT can predict progression in lung function. A quantitative texture-based score can play a pivotal role for informed and timely management decision for patients with IPF.

The original version of this article, published on 24 July 2014, unfortunately contained a mistake. In section "Discussion," a sentence was worded incorrectly.

Phytophthora root rot causes major economic losses in woody ornamental nurseries, especially in plants exposed to flooding. Ambrosia beetles, which attack stressed trees, are also important pests of woody plants. In this study, several products were evaluated for control of Phytophthora root rot and ambrosia beetles on containerized flowering dogwoods (Cornus florida L.) exposed to simulated flood events under field conditions. In two trials, preventive (7 days preflooding) or curative (1 day postflooding) sprench (combination spray and drench) treatments were applied to containerized dogwoods artificially inoculated with Phytophthora cinnamomi Rands. The plants were flooded by maintaining standing water for 1, 3, or 7 days. After the trials, plant growth data (dry shoot weight, dry root weight, plant height, plant width, caliper) were recorded, and roots were assessed for disease severity using a scale of 0–100% roots affected, as well as plated on PARPH-V8 medium to determine the percentage of Phytophthora-infected root samples. Ambrosia beetle attacks were recorded throughout the study. Longer flooding duration increased disease pressure from Phytophthora cinnamomi Rands. Preventive treatment of plants with Subdue MAXX consistently reduced disease severity at tested flooding durations of 1, 3, or 7 days. Preventively- (before flooding) or curatively- (after flooding) applied treatments of Empress Intrinsic, Orkestra Intrinsic, and Pageant Intrinsic reduced disease pressure at 1 and 3 days of flooding. Preventive RootShield Plus+ treatment reduced disease pressure on plants flooded 1 day. Longer flooding periods as well as Phytophthora cinnamomi inoculation increased ambrosia beetle attacks on flooded dogwoods, but preventively- and curatively-applied Orkestra Intrinsic as well as preventively-applied Subdue MAXX reduced ambrosia beetle attacks compared to non-treated, inoculated controls. These treatments were effective at reducing both Phytophthora and ambrosia beetle damage. These results can help nursery producers successfully manage Phytophtora root rot and also reduce the risk of ambrosia beetle attacks during flood events.

625 Background: Clear cell renal cell carcinoma (ccRCC) is a highly vascularized tumor and has a heterogeneous molecular profile, but the correlation between its tumor biology and radiomic features have only recently been investigated. As 70% of ccRCCs are detected incidentally at imaging, and tumor phenotypes are only available after surgery, a non-invasive biomarker to predict ccRCC phenotypes and aggressiveness on imaging may be clinically valuable, as low grade lesions may undergo active surveillance. Methods: With IRB approval and HIPAA compliance, our study cohort comprised 92 consecutive patients with 102 ccRCCs (mean age, 62 years (SD ± 14.2) with histopathology and molecular endpoints imaged preoperatively on CT with a four-phase renal protocol [unenhanced (U), corticomedullary (C), nephrographic (N), excretory (E)]. Radiomic data was obtained by contouring the entire ccRCC in each phase to obtain a 3D tumor volume of interest (VOI). The mean enhancement in each phase the and wash-in and wash-out of enhancement was calculated for each ccRCC. Molecular data was obtained through immunohistochemistry of resected ccRCCs to assess carbonic anhydrase-IX (CAIX), microvessel density (MVD), phosphatase and tension homolog (PTEN), and tumor grade (TG). Categorical variables were analyzed with logistic regression and odds ratio (OR) was reported. Continuous variables were analyzed with linear regression and Pearson correlation coefficient (r2) was reported. P-values < .05 were considered significant. Results: Significant radiomic associations included TG and 3D tumor enhancement in the C (OR = 4.72, p = .030), N (OR = 17.71, p < .0001), and E phases (OR = 17.10, p < .0001), and wash-in from U to C (OR = 8.27, p = .004). MVD had a significant positive association with 3D tumor enhancement in the C phase (r2= 0.410, p < .0001), wash-in from U to C (r2= 0.435, p < .0001) and wash-out from the C to N (r2= 0.435, p = .001). There were no significant radiomic correlates with CAIX or PTEN expression. Conclusions: A 3D ccRCC VOI on multiphasic CT had significant correlations with TG and MVD, independent of clinical features, potentially becoming a predictive imaging biomarker of ccRCC aggressiveness and clinical outcome.

Purpose To assess the intra- and interscan reproducibility of LV twist using FAST. Assessing the reproducibility of the measurement of new MRI biomarkers is an important part of validation. Fourier Analysis of STimulated Echoes (FAST) is a new MRI tissue tagging method that has recently been shown to compare favorably with conventional estimates of left ventricular (LV) twist from cardiac tagged images, but with significantly reduced user interaction time. Materials and Methods Healthy volunteers (N = 10) were scanned twice using FAST over 1 week. On day 1, two measurements of LV twist were collected for intrascan comparisons. Measurements for LV twist were again collected on day 8 for interscan assessment. LV short-axis tagged images were acquired on a 3 Tesla (T) scanner to ensure detectability of tags during early and mid-diastole. Peak LV twist is reported as mean ± SD. Reproducibility was assessed using the concordance correlation coefficient (CCC) and the repeatability coefficient (RC) (95% confidence interval [CI] range). Results Mean peak twist measurements were 13.4 ± 4.3° (day 1, scan 1), 13.6 ± 3.7° (day 1, scan 2), and 13.0 ± 2.7° (day 8). Bland-Altman analysis resulted in intra- and interscan bias and 95% CI of −0.6° [−1.0°, 1.6°] and 1.4° (−1.0°, 3.0°), respectively. The Bland-Altman RC for peak LV twist was 2.6° and 4.0° for intra- and interscan, respectively. The CCC was 0.9 and 0.6 for peak LV twist for intra- and interscan, respectively. Conclusion FAST is a semi-automated method that provides a quick and quantitative assessment of LV systolic and diastolic twist that demonstrates high intrascan and moderate interscan reproducibility in preliminary studies. J. Magn. Reson. Imaging 2014;39:463–468. © 2013 Wiley Periodicals, Inc.

Lung cancer screening CT is already performed at low dose. There are many techniques to reduce the dose even further, but it is not clear how such techniques will affect nodule detectability. In this work, we used an in-house CAD algorithm to evaluate detectability. 90348 patients and their raw CT data files were drawn from the National Lung Screening Trial (NLST) database. All scans were acquired at ~2 mGy CTDIvol with fixed tube current, 1 mm slice thickness, and B50 reconstruction kernel on a Sensation 64 scanner (Siemens Healthcare). We used the raw CT data to simulate two additional reduced-dose scans for each patient corresponding to 1 mGy (50%) and 0.5 mGy (25%). Radiologists’ findings on the NLST reader forms indicated 65 nodules in the cohort, which we subdivided based on LungRADS criteria. For larger category 4 nodules, median sensitivities were 100% at all three dose levels, and mean sensitivity decreased with dose. For smaller nodules meeting the category 2 or 3 criteria, the dose dependence was less obvious. Overall, mean patient-level sensitivity varied from 38.5% at 100% dose to 40.4% at 50% dose, a difference of only 1.9%. However, the false-positive rate quadrupled from 1 per case at 100% dose to 4 per case at 25% dose. Dose reduction affected lung-nodule detectability differently depending on the LungRADS category, and the false-positive rate was very sensitive at sub-screening dose levels. Thus, care should be taken to adapt CAD for the very challenging noise characteristics of screening.

Blister-actuated laser-induced forward transfer (BA-LIFT) is a versatile printing technique in which fine jets of ink are ejected from a thin donor film onto an acceptor substrate, enabling high-resolution patterns to be formed. Fluid ejections are initiated by the rapid expansion of micrometre-sized blisters that form on a polymer film underneath the ink layer. Recent work has demonstrated that these ejections exhibit novel flow phenomena due to the unique dimensions and geometry of the BA-LIFT configuration. In this work, we study the dynamics of BA-LIFT printing using a computational model in which fluid is forced by a boundary that deforms according to experimental time-resolved measurements of an expanding blister profile. This allows the model’s predictions to be unambiguously correlated with experimental blister-actuated ejections without any fitting parameters. First, we validate the model’s predictive capabilities against experimental results, including the ability to accurately reproduce the size, shape and temporal evolution of the jet as well as the total volume of ink released. The validated model is then used to interrogate the flow dynamics in order to better understand the mechanisms for fluid ejection. Finally, parametric studies are conducted to investigate the influence of ink density, surface tension, viscosity and film thickness as well as the size of the blister used. These results provide key insights into avenues for optimization and better control of the BA-LIFT process for improved resolution and repeatability of the printed features.

To investigate volumetric and density changes in the ipsilateral and contralateral lobes following volume reduction of an emphysematous target lobe.The study included 289 subjects with heterogeneous emphysema, who underwent bronchoscopic volume reduction of the most diseased lobe with endobronchial valves and 132 untreated controls. Lobar volume and low-attenuation relative area (RA) changes post-procedure were measured from computed tomography images. Regression analysis (Spearman's rho) was performed to test the association between change in the target lobe volume and changes in volume and density variables in the other lobes.The target lobe volume at full inspiration in the treatment group had a mean reduction of -0.45 L (SE = 0.034, P 0.0001), and was associated with volume increases in the ipsilateral lobe (rho = -0.68, P 0.0001) and contralateral lung (rho = -0.16, P = 0.006), and overall reductions in expiratory RA (rho = 0.31, P 0.0001) and residual volume (RV)/total lung capacity (TLC) (rho = 0.13, P = 0.03).When the volume of an emphysematous target lobe is reduced, the volume is redistributed primarily to the ipsilateral lobe, with an overall reduction. Image-based changes in lobar volumes and densities indicate that target lobe volume reduction is associated with statistically significant overall reductions in air trapping, consistent with expansion of the healthier lung.Computed tomography allows assessment of the treatment of emphysema with endobronchial valves. • Endobronchial valves can reduce the volume of an emphysematous lung lobe. • Compensatory expansion is greater in ipsilateral lobes than in the contralateral lung. • Reduced air trapping is measurable by RV/TLC and smaller low attenuation area.

The Scleroderma Lung Study showed the efficacy of cyclophosphamide in modestly improving the forced vital capacity (FVC) compared with placebo over 1 year. Using changes in texture-based scores that quantify lung fibrosis as the percentage involvement of reticulation patterns, the effectiveness of cyclophosphamide was re-assessed by examining its impact on quantitative lung fibrosis (QLF). Axial HRCT images were acquired (1-mm slice thickness, 10-mm increments) in the prone position at inspiration. A validated model for quantifying interstitial disease patterns was applied to images from 83 subjects at baseline and 12 months. Scores were calculated for six zones (upper, mid, lower of the right/left lung) and the whole lung. Average changes were compared. Correlations were performed between QLF and physiological and clinical scores. From the most severe zones identified at baseline, QLF scores decreased by 2.6% in the cyclophosphamide group, whereas they increased by 9.1% in the placebo group, leading to ~12% difference (p = 0.0027). Between-treatment difference in whole lung QLF was ~5% (p = 0.0190). Significant associations were observed between changes in QLF and FVC (r = −0.33), dyspnea score (r = −0.29), and consensus visual score (p = 0.0001). QLF scores provide an objective quantitative tool for assessing treatment efficacy in scleroderma-related interstitial lung disease.

Blister-actuated laser-induced forward transfer (BA-LIFT) is a direct-write technique, which enables high-resolution printing of sensitive inks for electronic or biological applications. During BA-LIFT, a polymer laser-absorbing layer deforms into an enclosed blister and ejects ink from an adjacent donor film. In this work, we develop a finite element model to replicate and predict blister expansion dynamics during BA-LIFT. Model inputs consist of standard mechanical properties, strain-rate-dependent material parameters, and a parameter encapsulating the thermal and optical properties of the film. We present methods to determine these material parameters from experimental measurements. The simulated expansion dynamics are shown to be in good agreement with experimental measurements using two different polymer layer thicknesses. Finally, the ability to model high-fluence blister rupture is demonstrated through a strain-based failure approach.

Laser-induced forward transfer (LIFT) is a high-resolution direct-write technique, which can print a wide range of liquid materials without a nozzle. In this process, a pulsed laser initiates the expulsion of a high-velocity micro-jet of fluid from a thin donor film. LIFT involves a novel regime for impulsively driven free-surface jetting in that viscous forces developed in the thin film become relevant within the jet lifetime. In this work, time-resolved microscopy is used to study the dynamics of the laser-induced ejection process. We consider the influence of thin metal and thick polymer laser-absorbing layers on the flow actuation mechanism and resulting jet dynamics. Both films exhibit a mechanism in which flow is driven by the rapid expansion of a gas bubble within the liquid film. We present high-resolution images of the transient gas cavities, the resulting ejection of high aspect ratio external jets, as well as the first images of re-entrant jets formed during LIFT. These observations are interpreted in the context of similar work on cavitation bubble formation near free surfaces and rigid interfaces. Additionally, by increasing the laser beam size used on the polymer absorbing layer, we observe a transition to an alternate mechanism for jet formation, which is driven by the rapid expansion of a blister on the polymer surface. We compare the dynamics of these blister-actuated jets to those of the gas-actuated mechanism. Finally, we analyze these results in the context of printing sensitive ink materials.

BACKGROUND AND PURPOSE: Currently it is difficult to predict tumor response to anti-angiogenic therapy in individual patients. Our aim was to determine if ADC histogram analysis can stratify progression-free and overall survival in patients with newly diagnosed GBM treated “up-front” (ie, before tumor recurrence) with bevacizumab. MATERIALS AND METHODS: Up-front bevacizumab-treated and control patients (n = 59 and 62, respectively) with newly diagnosed GBM were analyzed by using an ADC histogram approach based on enhancing tumor. Progression-free and overall survival was determined by using Cox proportional HRs and the Kaplan-Meier method with logrank and Wilcoxon tests. RESULTS: For up-front bevacizumab-treated patients, lower ADC(L) was associated with significantly longer progression-free survival (median, 459 days for ADC(L) < 1200 versus 315 days for ADC(L) ≥ 1200 10(−6)mm(2)/s; P = .008, logrank test) and trended with longer overall survival (581 versus 429 days, P = .055). ADC values did not stratify progression-free or overall survival for patients in the control group (P = .92 and P = .22, respectively). Tumors with MGMT promoter methylation had lower ADC(L) values than unmethylated tumors (mean, 1071 versus 1183 10(−6)mm(2)/s; P = .01, 2-group t test). CONCLUSIONS: Pretreatment ADC histogram analysis can stratify progression-free survival in bevacizumab-treated patients with newly diagnosed GBM. Lower ADC is associated with tumor MGMT promoter methylation, which may, in part, account for the favorable outcome associated with low ADC(L) tumors.

Objective: We evaluated patterns of tumor progression in patients with recurrent glioblastoma who were treated with bevacizumab (BEV) alone or in combination with irinotecan (CPT-11) while participating in the BRAIN study. Methods: An independent neuroradiologist reviewed MRI scans at baseline and progression in patients who received BEV (n = 85) or BEV+CPT-11 (n = 82) while on BRAIN. Tumor patterns were scored as local, distant, diffuse, or multifocal. Median progression-free survival (PFS) and overall survival (OS) were estimated using Kaplan-Meier methods. Hazard ratios for PFS and OS were estimated using a Cox regression model. Results: Twenty-eight percent of patients who participated in BRAIN had nonlocal disease at baseline (72% local disease). Sixty-seven (79%) patients treated with single-agent BEV and 57 (70%) patients treated with BEV+CPT-11 experienced disease progression while on BRAIN. Most patients in each treatment group did not have a change in the radiographic pattern of their tumor (i.e., “no shift”) at the time of progression. The proportion of BEV patients with no shift (82%) was greater than that of BEV+CPT-11 patients (53%, χ 2 p = 0.0004), and a greater proportion of BEV+CPT-11 patients (39%) compared with BEV patients (16%) experienced local-to-diffuse tumor pattern at progression (χ 2 p = 0.002). Patients treated with BEV or BEV+CPT-11 who had local-to-local or local-to-diffuse progression patterns had similar efficacy outcomes, including objective response, PFS, and OS. Conclusions: Most patients treated with BEV or BEV+CPT-11 on BRAIN did not experience a change from baseline in radiographic characteristics of disease at the time of progression.

Purpose: The development of computer-aided diagnostic (CAD) methods for lung nodule detection, classification, and quantitative assessment can be facilitated through a well-characterized repository of computed tomography (CT) scans. The Lung Image Database Consortium (LIDC) and Image Database Resource Initiative (IDRI) completed such a database, establishing a publicly available reference for the medical imaging research community. Initiated by the National Cancer Institute (NCI), further advanced by the Foundation for the National Institutes of Health (FNIH), and accompanied by the Food and Drug Administration (FDA) through active participation, this public-private partnership demonstrates the success of a consortium founded on a consensus-based process. Methods: Seven academic centers and eight medical imaging companies collaborated to identify, address, and resolve challenging organizational, technical, and clinical issues to provide a solid foundation for a robust database. The LIDC/IDRI Database contains 1018 cases, each of which includes images from a clinical thoracic CT scan and an associated XML file that records the results of a two-phase image annotation process performed by four experienced thoracic radiologists. In the initial blinded-read phase, each radiologist independently reviewed each CT scan and marked lesions belonging to one of three categories (" nodule�3 mm," " nodule

Background: Manually importing and analyzing image data can be time-consuming, prone to human error, and costly for large clinical trial datasets. This can lead to delays in quality control (QC) feedback to imaging sites and in obtaining data analysis results. Herein we describe the creation and application of a high-throughput review process for import, classification, labeling and QC of large multimodal clinical trial image datasets.Methods: Automated methods were used to remove patient identifying information, extract image header data, and filter image data for usability. A convolutional neural net was applied to estimate anatomy for CT images. Internal scores were assigned for each image series to identify the optimal series for labeling and reading of each anatomical region. Image QC reports were automatically generated for all patients.Results: In combined studies for which 204,492 series were received, 27,841 series were identified as usable and 13,415 series were labeled. Using this high-throughput method, total work-hours required per time point were reduced by an approximate factor of ten when compared to traditional review and labeling methods. Our anatomic classification system identified 95.7% of image series correctly, with the remaining series being manually corrected before labeling and analysis.Conclusions: A high-throughput image analysis pipeline was implemented in a large combined dataset of clinical trial image series. This pipeline can be applied across other studies and modalities for fast image data characterization, labeling and QC.

Emphysema is a disease that causes destruction of lung tissue and an overall reduction in lung compliance as air becomes trapped in the diseased regions. It affects an estimated 60 million people worldwide and up to 2 million Americans. As the disease progresses, the hyperinflated regions continue to expand, limiting the effective volume available for more viable lung tissue. New biomedical valves and stents are being developed that can be implanted in target airways to allow air to escape from the most diseased lung lobe, thereby reducing its volume and providing more space for the adjacent (healthier) lobe to expand. The challenge facing the clinician is to determine which patients are most likely to benefit from the treatment and which lobe/airway should be treated. This article describes imaging biomarkers and an automated computer-aided diagnosis system to perform patient selection and treatment targeting in emphysema.

The goal of this study was to develop a computer-aided therapeutic response (CADrx) system for early prediction of drug treatment response for glioblastoma multiforme (GBM) brain tumors with diffusion weighted (DW) MR images. In conventional Macdonald assessment, tumor response is assessed nine weeks or more post-treatment. However, we will investigate the ability of DW-MRI to assess response earlier, at five weeks post treatment. The apparent diffusion coefficient (ADC) map, calculated from DW images, has been shown to reveal changes in the tumor’s microenvironment preceding morphologic tumor changes. ADC values in treated brain tumors could theoretically both increase due to the cell kill (and thus reduced cell density) and decrease due to inhibition of edema. In this study, we investigated the effectiveness of features that quantify changes from pre- and post-treatment tumor ADC histograms to detect treatment response. There are three parts to this study: first, tumor regions were segmented on T1w contrast enhanced images by Otsu’s thresholding method, and mapped from T1w images onto ADC images by a 3D region of interest (ROI) mapping tool using DICOM header information, second, ADC histograms of the tumor region were extracted from both pre- and five weeks post-treatment scans, and fitted by a two-component Gaussian mixture model (GMM). The GMM features as well as standard histogram-based features were extracted. Finally, supervised machine learning techniques were applied for classification of responders or non-responders. The approach was evaluated with a dataset of 85 patients with GBM under chemotherapy, in which 39 responded and 46 did not, based on tumor volume reduction. We compared adaBoost, random forest and support vector machine classification algorithms, using ten-fold cross validation, resulting in the best accuracy of 69.41% and the corresponding area under the curve (Az) of 0.70.

To retrospectively investigate the prevalence of tracheal collapse in an emphysema cohort. The occurrence of a large degree of tracheal collapse may have important implications for the clinical management of respiratory symptoms and air trapping in patients with emphysema.Paired full-inspiratory and end-expiratory thin-section volumetric computed tomographic scans were available for 1071 long-term smokers with clinically and physiologically confirmed emphysema. The percentage reduction in the cross-sectional tracheal luminal area from full-inspiration to end-expiration was automatically computed at 2.5-mm intervals along the centerline of the trachea using customized software.Maximal tracheal collapse did not follow a normal distribution in the emphysema cohort (P.0001, skewness/kurtosis tests for normality); the median collapse was 18% (intraquartile range, 11%-30%). Statistically significant differences were found in the distribution of maximal collapse by gender (P.005, Wilcoxon rank sum test). Overall, 10.5% of men and 17.1% of women showed evidence of tracheomalacia on the basis of the criterion of a reduction of 50% or greater in cross-sectional tracheal luminal area at end-expiration.This study offers insights into the prevalence of tracheal collapse in a cohort of patients with emphysema; future work is needed to determine the possible relationship between tracheal collapse and air trapping in subjects with emphysema.

In measuring changes of gallbladder wall thickness using CT, robustness to differences in acquisition protocols including slice thickness can be important. We have developed an automated technique based on Laplace's equation to measure the gallbladder wall thickness using computer tomography (CT). The purpose of this work is to investigate the usefulness of the Laplacian technique in obtaining gallbladder wall thickness measurements that are reproducible with variations in CT slice thickness. This study included 2D (2D) and 3D (3D) wall thickness measurements using Laplace's equation. Ten subjects who had 5 mm (thick) and 2.5 mm (thin) reconstruction (from a single set of raw data) through the abdomen were randomly selected from a research database. Their volumetric CT images were acquired using a multidetector GE MEDICAL SYSTEMS---LightSpeed 16 scanner at 120 KVP, ~250 mAs, with standard filter reconstruction algorithm and manually segmented on all CT cross sections by a radiologist. The inner and outer boundaries of the gallbladder wall were obtained from the segmentation. The thickness of the wall was quantified by computing the distance between the boundaries for each scan and over the entire volume using Laplace's equation from mathematical physics. The distance between the surfaces is found by computing normalized gradients that form a vector field. The vector fields represent tangent vectors along field lines connecting both boundaries. The Laplacian technique was compared with the conventional Euclidean distance transformation (EDT) technique using coefficient of variation. EDT results in an Euclidean distance mapping between the two extracted surfaces. Both techniques were compared in 2D and 3D. For the 2D and 3D wall thickness measurements, a mean difference of 0.35 and 0.25 mm between thin and thick reconstruction was found respectively using Laplace's equation. EDT resulted in a higher mean difference for both 2D and 3D. In addition, a significant difference in thickness between the Laplacian technique and EDT techniques (p?0.001) were obtained. The Laplacian measurement of gallbladder wall showed significantly lower variation compared to EDT on different CT slice thickness for both 2D and 3D techniques. Hence, proving to be an important technique for obtaining reproducible wall thickness measurements of the gallbladder using CT.

Computer-aided diagnosis (CAD) technology is becoming an important tool to assess treatment response in clinical trials. However, CAD software alone is not sufficient to conduct an imaging-based clinical trial. There are a number of architectural requirements such as image receive (from multiple field sites), a database for storing quantitative measures, and data mining and reporting capabilities. In this paper we describe the architectural requirements to incorporate CAD into clinical trials and illustrate their functionality in therapeutic trials for emphysema.

Objectives The aim is to investigate whether the 12-month quantitative changes in high-resolution CT (HRCT) measures of interstitial lung disease (ILD) are different, and to understand how they change, in patients with scleroderma-related ILD who receive drug therapy versus placebo. Methods HRCT images were acquired at baseline and at 12 months in 83 participants in Scleroderma Lung Study I, a clinical trial comparing treatment with oral cyclophosphamide versus placebo. A computer-aided model was used to quantify the extent of fibrotic reticulation, ground glass and honeycomb patterns and quantitative ILD (QILD: sum of these patterns) in the whole lung and the lung zone (upper, middle or lower) of maximal disease involvement. Results Mean QILD score decreased by 3.9% in the cyclophosphamide group while increasing by 4.2% in the placebo group in the most severe zone (p=0.01) and decreased by 3.2% in the cyclophosphamide group while increasing by 2.2% in the placebo group in the whole lung (p=0.03). Transitional probabilities demonstrated greater changes from a fibrotic to either a ground glass or normal pattern in the cyclophosphamide group and the reverse in the placebo group. Conclusions Changes in quantitative HRCT measures of ILD provide a sensitive indication of disease progression and response to treatment. Trial registration number NCT00004563; Post-results.

Lack of classifier robustness is a barrier to widespread adoption of computer-aided diagnosis systems for computed tomography (CT). We propose a novel Robustness-Driven Feature Selection (RDFS) algorithm that preferentially selects features robust to variations in CT technical factors. We evaluated RDFS in CT classification of fibrotic interstitial lung disease using 3D texture features. CTs were collected for 99 adult subjects separated into three datasets: training, multi-reconstruction, testing. Two thoracic radiologists provided cubic volumes of interest corresponding to six classes: pulmonary fibrosis, ground-glass opacity, honeycombing, normal lung parenchyma, airway, vessel. The multi-reconstruction dataset consisted of CT raw sinogram data reconstructed by systematically varying slice thickness, reconstruction kernel, and tube current (using a synthetic reduced-tube-current algorithm). Two support vector machine classifiers were created, one using RDFS ("with-RDFS") and one not ("without-RDFS"). Classifier robustness was compared on the multi-reconstruction dataset, using Cohen's kappa to assess classification agreement against a reference reconstruction. Classifier performance was compared on the testing dataset using the extended g-mean (EGM) measure. With-RDFS exhibited superior robustness (kappa 0.899-0.989) compared to without-RDFS (kappa 0.827-0.968). Both classifiers demonstrated similar performance on the testing dataset (EGM 0.778 for with-RDFS; 0.785 for without-RDFS), indicating that RDFS does not compromise classifier performance when discarding nonrobust features. RDFS is highly effective at improving classifier robustness against slice thickness, reconstruction kernel, and tube current without sacrificing performance, a result that has implications for multicenter clinical trials that rely on accurate and reproducible quantitative analysis of CT images collected under varied conditions across multiple sites, scanners, and timepoints.

Symptomatic localized adenomyosis is generally treated conservatively, or with more radical treatments such as hysterectomy. An effective non-invasive therapy is needed, especially for those who wish to preserve their fertility. MR-guided Focused Ultrasound ablation (MRgFUS) has been shown as an effective treatment for symptomatic uterine leiomyomas with a large non-perfused volume (NPV) immediately after treatment. Our specific aim was to compare the change in NPV in subjects with localized adenomyomas treated with MRgFUS, with the change in NPV in subjects treated with MRgFUS for symptomatic uterine leiomyomas to determine if similar results were achieved in the localized adenomyoma cohort.

PURPOSE: Texture features have been investigated as a biomarker of response and malignancy. Because these features reflect local differences in density, they may be influenced by acquisition and reconstruction parameters. The purpose of this study was to investigate the effects of radiation dose level and reconstruction method on features derived from lung lesions. METHODS: With IRB approval, 33 lung tumor cases were identified from clinically indicated thoracic CT scans in which the raw projection (sinogram) data were available. Based on a previously-published technique, noise was added to the raw data to simulate reduced-dose versions of each case at 25%, 10% and 3% of the original dose. Original and simulated reduced dose projection data were reconstructed with conventional and two iterative-reconstruction settings, yielding 12 combinations of dose/recon conditions. One lesion from each case was contoured. At the reference condition (full dose, conventional recon), 17 lesions were randomly selected for repeat contouring (repeatability). For each lesion at each dose/recon condition, 151 texture measures were calculated. A paired differences approach was employed to compare feature variation from repeat contours at the reference condition to the variation observed in other dose/recon conditions (reproducibility). The ratio of standard deviation of the reproducibility to repeatability was used as the variation measure for each feature. RESULTS: The mean variation (standard deviation) across dose levels and kernel was significantly different with a ratio of 2.24 (±5.85) across texture features (p=0.01). The mean variation (standard deviation) across dose levels with conventional recon was also significantly different with 2.30 (7.11) (p=0.025). The mean variation across reconstruction settings of original dose has a trend in showing difference with 1.35 (2.60) among all features (p=0.09). CONCLUSION: Texture features varied considerably with variations in dose and reconstruction condition. Care should be taken to standardize these conditions when using texture as a quantitative feature. This effort supported in part by a grant from the National Cancer Institute's Quantitative Imaging Network (QIN): U01 CA181156; The UCLA Department of Radiology has a Master Research Agreement with Siemens Healthcare; Dr. McNitt-Gray has previously received research support from Siemens Healthcare.

PURPOSE: Quantitative Imaging and CAD tasks performed with CT (e.g. lung nodule detection, assessment of tumor size, etc.) may be sensitive to image acquisition and reconstruction parameters such as dose level, image thickness and reconstruction method (FBP, iterative, etc.). The purpose of this work was to develop a research pipeline for generating CT image series that represent a wide variety of acquisition and reconstruction conditions under which CAD and Quantitative Imaging performance would be evaluated. METHODS: With IRB approval, we have collected the raw CT data from hundreds of patients. These raw sinogram files serve as the input to the research pipeline. To simulate a wide range of dose levels, we developed software which adds noise to the sinogram. Multiple reduced-dose sinograms can be generated for a single patient, and those reduced-dose sinograms are then fed either to the scanner's reconstruction engine or to our in-house reconstruction engine; each has conventional filtered back projection (FBP) and iterative reconstruction methods. After generating image series across a range of dose levels and reconstruction methods, we can evaluate the performance of various quantitative imaging or CAD tools in tasks such as automated and semi-automated lesion segmentation, assessment of lesion size, and measurement of density or texture. RESULTS: We have successfully applied this pipeline across a range of clinical CT applications, including: (1) chest oncology, where the pipeline was used to quantify the effects of dose and reconstruction method on nodule volumetry, and (2) lung cancer screening, where the pipeline is being used to measure the robustness of an automated CAD algorithm with respect to dose. CONCLUSION: The acquisition/reconstruction pipeline shows promise for investigating and quantifying the effects of dose and reconstruction method on various clinical CT applications. NCI grant U01 CA181156 (Quantitative Imaging Network); Tobacco Related Disease Research Project grant 22RT-0131.

Purpose To determine whether conventional nodule densitometry or analysis based on contrast enhancement maps of indeterminate lung nodules imaged with contrast-enhanced CT can distinguish benign from malignant lung nodules.

Rationale and Objectives The objective is to study the incremental effects of using a computer-aided lung nodule detection (CAD) system on the performance of a large pool of observers. Materials and Methods A set of eight thin-section computed tomographic data sets with limited longitudinal coverage, containing a total of 22 lung nodules, was analyzed by using the automated nodule detection system. When applied to all eight cases, the CAD system alone achieved a detection rate of 86.4%, with 2.64 false-positive results per case. This study included 202 observers at a national radiology meeting: 39 thoracic radiologists, 95 nonthoracic radiologists, and 68 nonradiologists. Each participant read from one to eight cases in random order, first without and then with CAD system output available. Observer performance in nodule detection was measured before and after CAD was made available. Differences in performance of groups of observers before and after CAD were tabulated by mean, median, and SD in detection rate and number of false-positive results and tested by using nonparametric methods. Results In an analysis involving only the first randomly selected case read by all 202 participants, there were statistically significant increases in nodule detection rates and numbers of false-positive results for all types of observers. There was a significant difference in detection rates between radiologists and nonradiologists before CAD, but after CAD, there was no significant difference in detection rates between these observer types. In a second analysis involving 13 participants who read all eight cases, mean detection rates were 64.0% before CAD and 81.9% after CAD. Mean numbers of false-positive results were 0.144 per case before CAD and 0.173 after CAD. Conclusion In a large observer study, use of a CAD system for nodule detection resulted in an incremental increase in detection rate, but also led to an increase in number of false-positive results. Also, CAD appears to be an equalizer of detection rates between observers of different levels of experience.

Rationale and objectives To investigate the performance of observers with different levels of experience in distinguishing between benign and malignant solitary pulmonary nodules (SPN) on CT, and to determine the effects on interpretation of three different conditions: image data alone, the addition of clinical data, and the addition of output from a computer-aided diagnosis (CAD) system. Materials and methods 28 thin-section CT datasets of SPNs with proven diagnoses (15 malignant and 13 benign) were used to measure observer performance. Readers were categorized according to their experience and read the cases in random order. For each case readers were asked to assign a level of confidence on a scale from 0.0–1.0 (0.0 benign, 1.0 malignant) for the diagnosis of the nodule. Each reader scored the cases based on review of image data alone (phase 1), then with limited clinical data (phase 2), and finally with CAD output (phase 3). To assess performance, multiple reader multiple case (MRMC) receiver operating characteristic (ROC) analysis was used. Results 2 thoracic radiologists, 1 thoracic radiology fellow, 2 nonthoracic radiologists, and 3 radiology residents read the cases. The average area under the ROC curve for all readers (A z ) at each stage was 0.68, 0.75, and 0.81, for image data alone, with clinical data, and with CAD output respectively. The difference in performance between phases (2 and 3) and (1 and 3) was significantly different ( P = 0.018 and P = 0.020). However, the difference between phases (1 and 2) was not significantly different ( P = 0.155). Conclusion Diagnostic performance increased significantly with the addition of CAD output. With further validation CAD output may play a significant role in SPN management.

Fifty-four solitary pulmonary nodules with diagnosis established with pathological or 2 years of radiographic followup were imaged using CT with a high-resolution protocol without contrast. A thoracic radiologist manually outlined each nodule on a single slice that they chose as most representative of it. Nineteen two-dimensional attenuation, size, and shape features were calculated based on this slice and were input into a OneR and logistic regression classifier. Classifier performance was measured using resubstitution, leave-one-out, and a split training set (42 cases) and test set (12 cases). When testing the classifier using resubstitution OneR achieved 90.9% sensitivity, 71.4% specificity, and 83.3% accuracy while the logistic regression classifier achieved 81.8% sensitivity, 85.7% specificity, and 83.3% accuracy. The OneR classifier in a leave-one-out testing methodology was able to achieve 90.9% sensitivity, 66.7% specificity, and 81.4% accuracy compared to 63.6% sensitivity, 52.4% specificity, and 59.3% accuracy of the logistic regression classifier. When separate training/testing sets were made OneR resulted in 85.7% sensitivity, 60% specificity, and 75% accuracy while logistic regression resulted in 71.4% sensitivity, 40% specificity, and 58.3% accuracy. Thus a simple classifier like OneR can perform better than a more complex classifier.

Purpose: A method to discriminate different types of renal cell carcinoma (RCC) was developed using attenuation values observed in multiphasic contrast-enhanced CT. This work evaluates the sensitivity of this RCC discrimination task at different CT radiation dose levels. Methods: We selected 5 cases of kidney lesion patients who had undergone four-phase CT scans covering the abdomen to the lilac crest. Through an IRB-approved study, the scans were conducted on 64-slice CT scanners (Definition AS/Definition Flash, Siemens Healthcare) using automatic tube-current modulation (TCM). The protocol included an initial baseline unenhanced scan, followed by three post-contrast injection phases. CTDIvol (32 cm phantom) measured between 9 to 35 mGy for any given phase. As a preliminary study, we limited the scope to the cortico-medullary phase—shown previously to be the most discriminative phase. A previously validated method was used to simulate a reduced dose acquisition via adding noise to raw CT sinogram data, emulating corresponding images at simulated doses of 50%, 25%, and 10%. To discriminate the lesion subtype, ROIs were placed in the most enhancing region of the lesion. The mean HU value of an ROI was extracted and used to discriminate to the worst-case RCC subtype, ranked in the order of clear cell, papillary, chromophobe and the benign oncocytoma. Results: Two patients exhibited a change of worst case RCC subtype between original and simulated scans, at 25% and 10% doses. In one case, the worst-case RCC subtype changed from oncocytoma to chromophobe at 10% and 25% doses, while the other case changed from oncocytoma to clear cell at 10% dose. Conclusion: Based on preliminary results from an initial cohort of 5 patients, worst-case RCC subtypes remained constant at all simulated dose levels except for 2 patients. Further study conducted on more patients will be needed to confirm our findings. Institutional research agreement, Siemens Healthcare; Past recipient, research grant support, Siemens Healthcare; Consultant, Toshiba America Medical Systems; Consultant, Samsung Electronics; NIH Grant Support from: U01 CA181156

An automated system was developed for detecting lung micronodules on thin-section computed tomographic images and was applied to data from 15 subjects with 77 lung nodules. The automated system, without user interaction, achieved a sensitivity of 100% for nodules (>3 mm in diameter) and 70% for micronodules (

Medical image segmentation techniques typically require some form of expert human supervision to provide accurate and consistent identification of anatomic structures of interest. A novel segmentation technique was developed that combines a knowledge-based segmentation system with a sophisticated active contour model. This approach exploits the guidance of a higher-level process to robustly perform the segmentation of various anatomic structures. The user need not provide initial contour placement, and the high-level process carries out the required parameter optimization automatically. Knowledge about the anatomic structures to be segmented is defined statistically in terms of probability density functions of parameters such as location, size, and image intensity (eg, computed tomographic [CT] attenuation value). Preliminary results suggest that the performance of the algorithm at chest and abdominal CT is comparable to that of more traditional segmentation techniques like region growing and morphologic operators. In some cases, the active contour-based technique may outperform standard segmentation methods due to its capacity to fully enforce the available a priori knowledge concerning the anatomic structure of interest. The active contour algorithm is particularly suitable for integration with high-level image understanding frameworks, providing a robust and easily controlled low-level segmentation tool. Further study is required to determine whether the proposed algorithm is indeed capable of providing consistently superior segmentation.

e16049 Background: Clear cell renal cell carcinoma (ccRCC) is the most common tumor of the kidney. Up to 70% are incidentally detected on multiphasic CT. The prognosis for patients with ccRCC is related to Fuhrman grade (FG) and is diagnosed by biopsy or excision. There is a great need for a non-invasive method to assess tumor grade which may help inform clinical decision-making. The purpose of our study is to determine if contrast enhancement on CT predicts FG and microvessel density (MVD) of ccRCC lesions and to assess which combination of quantitative and qualitative radiological features and clinical features predict high FG ccRCC lesions. Methods: With IRB approval for this HIPAA-compliant retrospective study, our pathology and imaging databases were queried to obtain a cohort of ccRCC with a preoperative multiphasic (unenhanced (U), corticomedullary (C), nephrographic (N), and excretory (E)) CT scan. Tumors were stained with CD4 to quantify % MVD. Spearman’s rank correlation was calculated to test the strength of the association between CT enhancement, %MVD and FG. Stepwise logistic regression analysis was performed to determine the quantitative and/or qualitative feature with the highest performance in predicting high FG tumor. The multivariate logistic regression analysis was evaluated using ROC curves and AUCs. Results: Our cohort had 127 patients with 89 low-grade tumors and 43 high-grade tumors. The %wash-in of enhancement from the U to the C phase showed a significant correlation with %MVD of the tumor (R2= 0.181,p < . 001) as did enhancement of the tumor in the early C phase with %MVD (R2= 0.159,p < . 001). There was a significant inverse correlation with %MVD and FG (R2= 0.137,p < . 001). T-stage, tumor size, %MVD and presence of renal vein invasion were determined to be significant independent predictors of high grade lesions with an AUC of .838 (95% CI .748-.927). Conclusions: Lower grade ccRCC tumors have a higher MVD. Therefore, contrast material washes in at a faster rate from the UN to the CM phase, enabling low grade to be discriminated from high-grade tumors on CT.

To study the between-scanner variation and the between-visit reproducibility of brain apparent diffusion coefficient (ADC) measurements in the setting of a multi-center chemotherapy clinical trial for glioblastoma multiforme.ADC maps of 52 patients at six sites were calculated in-house from diffusion-weighted images obtained by seven individual scanner models of two vendors. The median and coefficient of variation (CV) of normal brain white matter ADC values from a defined region of interest were used to evaluate the differences among scanner models, vendors, magnetic fields, as well as successive visits. All patients participating in this study signed institutional review board approved informed consent. Data acquisition was performed in compliance with all applicable Health Insurance Portability and Accountability Act regulations. The study spanned from August 1, 2006, to January 29, 2008.For baseline median ADC, no difference was observed between the different scanner models, different vendors, and different magnetic field strength. For baseline ADC CV, a significant difference was found between different scanner models (p = 0.0002). No between-scanner difference was observed in ADC changes between two visits. For between-visit reproducibility, significant difference was seen between the ADC values measured at two successive visits for the whole patient group.The CVs varied significantly between scanners, presumably due to image noise. Consistent scanner parameter setup can improve reproducibility of the ADC measurements between visits.

Many medical image analysis systems require segmentation of the structures of interest as a first step. For scans with gross pathology, automatic segmentation methods may fail. The authors' aim is to develop a versatile, fast, and reliable interactive system to segment anatomical structures. In this study, this system was used for segmenting lungs in challenging thoracic computed tomography (CT) scans.In volumetric thoracic CT scans, the chest is segmented and divided into 3D volumes of interest (VOIs), containing voxels with similar densities. These VOIs are automatically labeled as either lung tissue or nonlung tissue. The automatic labeling results can be corrected using an interactive or a supervised interactive approach. When using the supervised interactive system, the user is shown the classification results per slice, whereupon he/she can adjust incorrect labels. The system is retrained continuously, taking the corrections and approvals of the user into account. In this way, the system learns to make a better distinction between lung tissue and nonlung tissue. When using the interactive framework without supervised learning, the user corrects all incorrectly labeled VOIs manually. Both interactive segmentation tools were tested on 32 volumetric CT scans of pigs, mice and humans, containing pulmonary abnormalities.On average, supervised interactive lung segmentation took under 9 min of user interaction. Algorithm computing time was 2 min on average, but can easily be reduced. On average, 2.0% of all VOIs in a scan had to be relabeled. Lung segmentation using the interactive segmentation method took on average 13 min and involved relabeling 3.0% of all VOIs on average. The resulting segmentations correspond well to manual delineations of eight axial slices per scan, with an average Dice similarity coefficient of 0.933.The authors have developed two fast and reliable methods for interactive lung segmentation in challenging chest CT images. Both systems do not require prior knowledge of the scans under consideration and work on a variety of scans.

Bone tumor segmentation on bone scans has recently been adopted as a basis for objective tumor assessment in several phase II and III clinical drug trials. Interpretation can be difficult due to the highly sensitive but non-specific nature of bone tumor appearance on bone scans. In this paper we present a machine learning approach to segmenting tumors on bone scans, using intensity and context features aimed at addressing areas prone to false positives. We computed the context features using landmark points, identified by a modified active shape model. We trained a random forest classifier on 100 and evaluated on 73 prostate cancer subjects from a multi-center clinical trial. A reference segmentation was provided by a board certified radiologist. We evaluated our learning based method using the Jaccard index and compared against the state of the art, rule based method. Results showed an improvement from 0.50 ±0.31 to 0.57 ±0.27. We found that the context features played a significant role in the random forest classifier, helping to correctly classify regions prone to false positives.