Three-dimensional lung tumor segmentation from x-ray computed tomography using sparse field active models.

PURPOSE: Manual segmentation of lung tumors is observer dependent and time-consuming but an important component of radiology and radiation oncology workflow. The objective of this study was to generate an automated lung tumor measurement tool for segmentation of pulmonary metastatic tumors from x-ray computed tomography (CT) images to improve reproducibility and decrease the time required to segment tumor boundaries. METHODS: The authors developed an automated lung tumor segmentation algorithm for volumetric image analysis of chest CT images using shape constrained Otsu multithresholding (SCOMT) and sparse field active surface (SFAS) algorithms. The observer was required to select the tumor center and the SCOMT algorithm subsequently created an initial surface that was deformed using level set SFAS to minimize the total energy consisting of mean separation, edge, partial volume, rolling, distribution, background, shape, volume, smoothness, and curvature energies. RESULTS: The proposed segmentation algorithm was compared to manual segmentation whereby 21 tumors were evaluated using one-dimensional (1D) response evaluation criteria in solid tumors (RECIST), two-dimensional (2D) World Health Organization (WHO), and 3D volume measurements. Linear regression goodness-of-fit measures (r(2) = 0.63, p < 0.0001; r(2) = 0.87, p < 0.0001; and r(2) = 0.96, p < 0.0001), and Pearson correlation coefficients (r = 0.79, p < 0.0001; r = 0.93, p < 0.0001; and r = 0.98, p < 0.0001) for 1D, 2D, and 3D measurements, respectively, showed significant correlations between manual and algorithm results. Intra-observer intraclass correlation coefficients (ICC) demonstrated high reproducibility for algorithm (0.989-0.995, 0.996-0.997, and 0.999-0.999) and manual measurements (0.975-0.993, 0.985-0.993, and 0.980-0.992) for 1D, 2D, and 3D measurements, respectively. The intra-observer coefficient of variation (CV%) was low for algorithm (3.09%-4.67%, 4.85%-5.84%, and 5.65%-5.88%) and manual observers (4.20%-6.61%, 8.14%-9.57%, and 14.57%-21.61%) for 1D, 2D, and 3D measurements, respectively. CONCLUSIONS: The authors developed an automated segmentation algorithm requiring only that the operator select the tumor to measure pulmonary metastatic tumors in 1D, 2D, and 3D. Algorithm and manual measurements were significantly correlated. Since the algorithm segmentation involves selection of a single seed point, it resulted in reduced intra-observer variability and decreased time, for making the measurements.

Pulmonary tumor measurements from x-ray computed tomography in one, two, and three dimensions.

RATIONALE AND OBJECTIVES: We evaluated the accuracy and reproducibility of three-dimensional (3D) measurements of lung phantoms and patient tumors from x-ray computed tomography (CT) and compared these to one-dimensional (1D) and two-dimensional (2D) measurements. MATERIALS AND METHODS: CT images of three spherical and three irregularly shaped tumor phantoms were evaluated by three observers who performed five repeated measurements. Additionally, three observers manually segmented 29 patient lung tumors five times each. Follow-up imaging was performed for 23 tumors and response criteria were compared. For a single subject, imaging was performed on nine occasions over 2 years to evaluate multidimensional tumor response. To evaluate measurement accuracy, we compared imaging measurements to ground truth using analysis of variance. For estimates of precision, intraobserver and interobserver coefficients of variation and intraclass correlations (ICC) were used. Linear regression and Pearson correlations were used to evaluate agreement and tumor response was descriptively compared. RESULTS: For spherical shaped phantoms, all measurements were highly accurate, but for irregularly shaped phantoms, only 3D measurements were in high agreement with ground truth measurements. All phantom and patient measurements showed high intra- and interobserver reproducibility (ICC >0.900). Over a 2-year period for a single patient, there was disagreement between tumor response classifications based on 3D measurements and those generated using 1D and 2D measurements. CONCLUSION: Tumor volume measurements were highly reproducible and accurate for irregular, spherical phantoms and patient tumors with nonuniform dimensions. Response classifications obtained from multidimensional measurements suggest that 3D measurements provide higher sensitivity to tumor response.

MR angiography of aortoiliac occlusive disease: a phase III study of the safety and effectiveness of the blood-pool contrast agent MS-325.

PURPOSE: To evaluate prospectively the safety and effectiveness of aortoiliac magnetic resonance (MR) angiography enhanced with MS-325 (gadofosveset trisodium) at a dose of 0.03 mmol/kg; effectiveness was defined as accuracy relative to the reference standard, conventional angiography. MATERIALS AND METHODS: Study was approved by institutional review boards of participating institutions, and required national approvals were obtained. Study protocol conformed to Good Clinical Practice guidelines, and informed patient consent was obtained. Patients with known or suspected peripheral vascular disease received 0.03 mmol/kg MS-325 for aortoiliac MR angiography. They were also examined with conventional angiography. MS-325-enhanced MR was evaluated for safety and effectiveness. Along with unenhanced two-dimensional time-of-flight MR angiography, it was compared with conventional angiography for presence of vascular stenosis. Student t tests were used to identify significant improvement in diagnostic sensitivity, specificity, and accuracy, as well as quantitative characterization of stenoses by three blinded readers. Correlations between readers of conventional angiograms were calculated and compared with MR results. RESULTS: In 174 patients, MS-325-enhanced MR angiography showed significant improvement (P < or = .001) in sensitivity, specificity, and accuracy for diagnosis of clinically significant (> or =50%) stenosis, compared with unenhanced MR. For all readers, areas under the receiver operating characteristic curve for both quantitative and qualitative measures of significant disease increased (P < .001) for MS-325-enhanced MR compared with time-of-flight MR. All readers also expressed higher confidence in diagnosis (P < .001) and found fewer images uninterpretable with MS-325 enhancement. All measures of interpretation accuracy approached corresponding measures of correlation between readers of conventional angiograms. Incidence of severe and serious adverse events with MS-325 was low. No patients were withdrawn from study due to adverse events or abnormalities in laboratory results. There were no clinically important trends in findings at hematology, blood chemistry, urinalysis, electrocardiography, or physical examination. CONCLUSION: MR angiography with MS-325 provides significant improvement in effectiveness over unenhanced MR (and minimal and transient side effects) at a dose of 0.03 mmol/kg and was safe and effective for MR evaluation of patients with aortoiliac occlusive disease.

Manually respiratory-triggered single-shot fast spin-echo: a non-breath-hold T2-weighted method for liver lesion detection.

OBJECTIVES: To determine whether manual respiratory triggering of a T2-weighted single-shot fast spin-echo sequence (M-SSFSE) improves detection and characterization of liver lesions compared with conventional breath-held single-shot fast spin-echo (C-SSFSE) technique. METHODS: M-SSFSE is performed by manually triggering a series of single-slice acquisitions through the liver at end of expiration. There were 171 hepatic lesions in 49 patients. Images were randomized and reviewed by 3 radiologists. Lesions were characterized as hemangiomas, cystic or solid. Dynamic gadolinium-enhanced sequences were used as the reference standard. Contrast-to-noise ratios (CNR) were calculated for all lesions that measured 1 cm or more. RESULTS: M-SSFSE was more sensitive than C-SSFSE in the detection of liver lesions (48%-58% v. 37%-46%, p < 0.001). There were no significant differences in the specificity of lesion detection between the 2 sequences. Artifacts were significantly less severe for M-SSFSE compared with C-SSFSE (p = 0.001). The CNR was significantly higher for all liver lesions on M-SSFSE compared with C-SSFSE (p < 0.001). CONCLUSION: M-SSFSE significantly improved hepatic lesion detection and, in particular, improved characterization of solid liver lesions and hemangiomas compared with C-SSFSE imaging.