Noriyuki Moriyama

Quantitative classification based on CT histogram analysis of non-small cell lung cancer: correlation with histopathological characteristics and recurrence-free survival.

PURPOSEnQuantification of the CT appearance of non-small cell lung cancer (NSCLC) is of interest in a number of clinical and investigational applications. The purpose of this work is to present a quantitative five-category (α, β, γ, δ, and ɛ) classification method based on CT histogram analysis of NSCLC and to determine the prognostic value of this quantitative classification.nnnMETHODSnInstitutional review board approval and informed consent were obtained at the National Cancer Center Hospital. A total of 454 patients with NSCLC (maximum lesion size of 3 cm) were enrolled. Each lesion was measured using multidetector CT at the same tube voltage, reconstruction interval, beam collimation, and reconstructed slice thickness. Two observers segmented NSCLC nodules from the CT images by using a semi-automated three-dimensional technique. The two observers classified NSCLCs into one of five categories from the visual assessment of CT histograms obtained from each nodule segmentation result. Interobserver variability in the classification was computed with Cohens κ statistic. Any disagreements were resolved by consensus between the two observers to define the gold standard of the classification. Using a classification and regression tree (CART), the authors obtained a decision tree for a quantitative five-category classification. To assess the impact of the nodule segmentation on the classification, the variability in classifications obtained by two decision trees for the nodule segmentation results was also calculated with the Cohens κ statistic. The authors calculated the association of recurrence with prognostic factors including classification, sex, age, tumor diameter, smoking status, disease stage, histological type, lymphatic permeation, and vascular invasion using both univariate and multivariate Cox regression analyses.nnnRESULTSnThe κ values for interobserver agreement of the classification using two nodule segmentation results were 0.921 (Pu2009<u20090.001) and 0.903 (Pu2009<u20090.001), respectively. The κ values for the variability in the classification task using two decision trees were 0.981 (Pu2009<u20090.001) and 0.981 (Pu2009<u20090.001), respectively. All the NSCLCs were classified into one of five categories (type α, nu2009=u20098; type β, nu2009=u200938; type γ, nu2009=u2009103; type δ, nu2009=u2009112; type ɛ, nu2009=u2009193) by using a decision tree. Using a multivariate Cox regression analysis, the classification (hazard ratio 5.64; Pu2009=u20090.008) and disease stage (hazard ratio 8.33; Pu2009<u20090.001) were identified as being associated with an increased recurrence risk.nnnCONCLUSIONSnThe quantitative five-category classifier presented here has the potential to provide an objective classification of NSCLC nodules that is strongly correlated with prognostic factors.

Staging performance of carbon-11 choline positron emission tomography/computed tomography in patients with bone and soft tissue sarcoma: comparison with conventional imaging.

The present study was conducted to compare the diagnostic accuracy between carbon‐11 choline (11C‐choline) positron emission tomography (PET)/computed tomography (CT) and conventional imaging for the staging of bone and soft tissue sarcomas. Sixteen patients who underwent 11C‐choline PET/CT prior to treatment were evaluated retrospectively for staging accuracy. Conventional imaging methods consisted of 99,mTc‐hydroxymethylene diphosphonate bone scintigraphy, chest CT and magnetic resonance imaging of the primary site. The images were reviewed and a consensus was reached by two board‐certified radiologists who were unaware of any clinical or radiological information using hard‐copy films and multimodality computer platform. Tumor stage was confirmed by histological examination and/or by an obvious progression in number and/or size of the lesions on follow‐up examinations. Reviewers examining both 11C‐choline PET/CT and conventional imaging classified T stage in all patients. Interpretation based on 11C‐choline PET/CT, the Node (N) stage was correctly diagnosed in all patients, whereas the accuracy of conventional imaging in N stage was 63%. Tumor Node Metastasis (TNM) stage was assessed correctly with 11C‐choline PET/CT in 15 of 16 patients (94%) and with conventional imaging in eight of 16 patients (50%). The overall TNM staging and N staging accuracy of 11C‐choline PET/CT were significantly higher than that of conventional imaging (P < 0.05). 11C‐choline PET/CT is more accurate than conventional imaging regarding clinical staging of patients with bone and soft tissue sarcomas. A whole body 11C‐choline PET/CT might be acceptable for imaging studies of tumor staging prior to treatment. (Cancer Sci 2006; 97: 1125–1128)

Extraskeletal osteosarcoma: Extensive tumor thrombus on fused PET-CT images

A 30-year-old woman developed extraskeletal osteosarcoma in the right buttock and thigh. Radiographs and unenhanced computed tomography (CT) showed a large, multilobulated mass accompanied by mineralized matrix. Contrast-enhanced CT and magnetic resonance (MR) images showed extensive tumor thrombus in the right internal- and external iliac veins. Co-registered positron emission tomography (PET) and CT images showed abnormal F-18 2-fluoro-2-deoxy-D-glucose (FDG) uptake in the tumor thrombus. PET study in our patient provided information concerning disease extent and viability of tumor thrombus.

Improvement of the diagnostic accuracy of lymph node metastases of colorectal cancer in 18F-FDG-PET/CT by optimizing the iteration number for the image reconstruction

ObjectiveImprovement in the diagnostic accuracy of lymph node (LN) metastases of colorectal cancer by optimizing the reconstruction parameters was studied, including a phantom study and clinical studies.MethodsIn the experimental study, the contrast ratio was evaluated using a standard image quality phantom, changing the iteration number of ordered subsets expectation maximization algorithm from 2 to 6. In the clinical study, 89 patients with preoperative colorectal cancer who received 18F-2-deoxy-2-fluoro-d-glucose positron emission tomography/computed tomography (PET/CT) were studied. Their LN metastases were visually assessed when systematically changing the iteration number, and the optimal iteration number was determined. After the appropriate cut-off value of maximum standardized uptake value was determined, the improvement of the diagnostic accuracy of LN metastases was evaluated for the proximal nodes and the distal ones using the reconstructed images with the optimal iteration number. This was compared with the conventional method that had an iteration number of 2.ResultsIn the phantom study, it was confirmed that the contrast ratio improved when the iteration number increased. In clinical evaluation, the optimal iteration number was determined to be 5 by visual assessment. When the cut-off value of 1.5 was used, which happened to be the best number, the sensitivity/specificity/accuracy improved from 52%/91%/72% to 77%/89%/83% for the proximal nodes, and from 63%/90%/84% to 91%85%/87% for the distal nodes, respectively.ConclusionsWhen the iteration number of the reconstruction algorithm was optimized, the sensitivity of LN metastasis improved by more than 20%, and the accuracy exceeded 80%. Optimization of the image reconstruction parameters in the diagnosis of LN metastases using PET/CT is clinically important.

Preliminary Study of Percutaneous Alcohol Injection into the Lung

Although percutaneous ethanol injection is widely used to treat hepatic tumors, this technique has not been applied to lung tumors. We performed a preliminary experiment with percutaneous ethanol injection into the rabbit lung, and evaluated the local and systemic effects of absolute ethanol injection on pulmonary structures in order to assess the feasibility and safety of this technique as a local treatment for human lung tumors. Percutaneous injection of absolute ethanol into the rabbit lung was performed under CT guidance. The volume of ethanol injected ranged from 0.6 to 1.0 ml (approximately 0.2–0.5 ml/kg). Follow‐up CT scans were performed 1, 2, 7 and 30 days after the injection. The animals were killed at intervals (range: 3 h–30 days), and the lung was examined histologically. The ethanol was well tolerated and did not induce significant systemic side‐effects. All doses induced necrosis in the injected lung, but none was lethal. Although ethanol spilling into the thoracic cavity induced effusion and pleuritis, these reactions were manageable. Alcohol injection produced an area of necrosis surrounded by pulmonary edema associated with polymorphonuclear cells invasion within 24 h; moreover, granulation change, epithelial regeneration, and alveolar septal fibrosis had appeared by one week. The necrosis was sometimes multifocal, probably due to transbronchial spread of the injected ethanol. In conclusion, the feasibility and safety of absolute ethanol injection were confirmed. Neither severe systemic side effects nor lethal extensive necrosis were observed with injected ethanol; however, an unexpected side effect, multifocal necrosis, was seen. The latter reaction suggests that careful observation and care would be essential after alcohol injection into the lung.

Blood vessel-based liver segmentation through the portal phase of a CT dataset

Blood vessels are dispersed throughout the human body organs and carry unique information for each person. This information can be used to delineate organ boundaries. The proposed method relies on abdominal blood vessels (ABV) to segment the liver considering the potential presence of tumors through the portal phase of a CT dataset. ABV are extracted and classified into hepatic (HBV) and nonhepatic (non-HBV) with a small number of interactions. HBV and non-HBV are used to guide an automatic segmentation of the liver. HBV are used to individually segment the core region of the liver. This region and non-HBV are used to construct a boundary surface between the liver and other organs to separate them. The core region is classified based on extracted posterior distributions of its histogram into low intensity tumor (LIT) and non-LIT core regions. Non-LIT case includes normal part of liver, HBV, and high intensity tumors if exist. Each core region is extended based on its corresponding posterior distribution. Extension is completed when it reaches either a variation in intensity or the constructed boundary surface. The method was applied to 80 datasets (30 Medical Image Computing and Computer Assisted Intervention (MICCAI) and 50 non-MICCAI data) including 60 datasets with tumors. Our results for the MICCAI-test data were evaluated by sliver07 [1] with an overall score of 79.7, which ranks seventh best on the site (December 2013). This approach seems a promising method for extraction of liver volumetry of various shapes and sizes and low intensity hepatic tumors.

A method for determining the modulation transfer function from thick microwire profiles measured with x-ray microcomputed tomography.

PURPOSEnThis study describes a model-dependent method to determine the modulation transfer function (MTF) in the transversal plane, obtained by a microcomputed tomography (micro-CT) system from profiles of a thick wire phantom instead of a thin wire phantom, and the study evaluates the feasibility of the proposed method in the MTF determination of micro-CT systems.nnnMETHODSnThe MTF is generally calculated as the absolute value of the normalized Fourier transform from the point spread function obtained by scanning a thin wire phantom. Since the wire is not a point source, the raw MTF is corrected for the finite size of the wire phantom; a wire with too large a diameter introduces inaccuracies in the MTF values. Therefore, we solved the MTF determination from profiles of a thick wire phantom via MTF modeling on the basis of the symmetric Lévy function that generalizes Gaussian and Lorentzian functions. We then applied the method to profiles of wire phantoms (1 mm, 2 mm, and 3 mm in diameter) measured by a clinical CT system to evaluate the applicable diameter range of the thick wire phantom. Two types of reconstruction kernels (standard and sharp) were used in the clinical CT. The performance of the method was evaluated using microwire phantoms (10 and 30μm in diameter) measured by a synchrotron radiation micro-CT (SRμCT) system, in which the Shepp-Logan filter and Ramachandran-Lakshminarayanan filter were used as the reconstruction kernel. The MTFs obtained using thin wire phantoms of 0.1 mm and 3 μm in diameter were regarded as the gold standard MTFs for the clinical CT and SRμCT, respectively. The root-mean-square error (RMSE) and relative error (RE) of the 10% value of the MTF were used to measure the difference between the MTF determined by the method and the gold standard.nnnRESULTSnThe mean RMSEs for two types of reconstruction kernels of three wire phantoms (1, 2, and 3 mm in diameter) were 0.0085, 0.012, and 0.021, respectively. The mean REs for the 1-, 2-, and 3-mm wire phantoms gave the same values of 2.0%, 3.5%, and 3.5%, respectively, for two types of reconstruction kernel. The MTFs determined from thick wire phantoms reveal the spatial resolution for the two kernels. The mean RMSEs for two types of reconstruction kernels of the microwire phantoms of 10 and 30μm in diameter were 0.0045 and 0.0035, respectively. The mean REs of the two wire phantoms of 10 and 30 μm diameter had 4.0% and 3.1%, respectively, for two types of reconstruction kernel.nnnCONCLUSIONSnExperimental data presented in this paper support the effectiveness of the model-dependent method based on the symmetric Lévy function. We conclude that the method is a useful approach for measuring the spatial resolution in the x/y-scan plane (transversal orientation) of micro-CT systems by substituting a thick wire phantom for a thin wire phantom.PURPOSEnThis study describes a model-dependent method to determine the modulation transfer function (MTF) in the transversal plane, obtained by a microcomputed tomography (micro-CT) system from profiles of a thick wire phantom instead of a thin wire phantom, and the study evaluates the feasibility of the proposed method in the MTF determination of micro-CT systems.nnnMETHODSnThe MTF is generally calculated as the absolute value of the normalized Fourier transform from the point spread function obtained by scanning a thin wire phantom. Since the wire is not a point source, the raw MTF is corrected for the finite size of the wire phantom; a wire with too large a diameter introduces inaccuracies in the MTF values. Therefore, we solved the MTF determination from profiles of a thick wire phantom via MTF modeling on the basis of the symmetric Lévy function that generalizes Gaussian and Lorentzian functions. We then applied the method to profiles of wire phantoms (1 mm, 2 mm, and 3 mm in diameter) measured by a clinical CT system to evaluate the applicable diameter range of the thick wire phantom. Two types of reconstruction kernels (standard and sharp) were used in the clinical CT. The performance of the method was evaluated using microwire phantoms (10 and 30 μm in diameter) measured by a synchrotron radiation micro-CT (SRμCT) system, in which the Shepp-Logan filter and Ramachandran-Lakshminarayanan filter were used as the reconstruction kernel. The MTFs obtained using thin wire phantoms of 0.1 mm and 3 μm in diameter were regarded as the gold standard MTFs for the clinical CT and SRμCT, respectively. The root-mean-square error (RMSE) and relative error (RE) of the 10% value of the MTF were used to measure the difference between the MTF determined by the method and the gold standard.nnnRESULTSnThe mean RMSEs for two types of reconstruction kernels of three wire phantoms (1, 2, and 3 mm in diameter) were 0.0085, 0.012, and 0.021, respectively. The mean REs for the 1-, 2-, and 3-mm wire phantoms gave the same values of 2.0%, 3.5%, and 3.5%, respectively, for two types of reconstruction kernel. The MTFs determined from thick wire phantoms reveal the spatial resolution for the two kernels. The mean RMSEs for two types of reconstruction kernels of the microwire phantoms of 10 and 30 μm in diameter were 0.0045 and 0.0035, respectively. The mean REs of the two wire phantoms of 10 and 30 μm diameter had 4.0% and 3.1%, respectively, for two types of reconstruction kernel.nnnCONCLUSIONSnExperimental data presented in this paper support the effectiveness of the model-dependent method based on the symmetric Lévy function. We conclude that the method is a useful approach for measuring the spatial resolution in the x∕y-scan plane (transversal orientation) of micro-CT systems by substituting a thick wire phantom for a thin wire phantom.

Extraction method of interlobar fissure based on multi-slice CT images

Extraction of inter lobar fissure is an active study for diagnosis and treatment. However, lung diseased cases have problems. The proposed method covers diseased cases, and contains three phases of coarse extract, fine extract and correction, using behavior of membrane. We applied this method to normal and lung diseased cases. Rate (average±standard eviation) of gold standard within 2 mm of extraction result were 91.2±3.6% for normal, and 89.7±4.9% for lung diseased cases. Rate of extraction result within 2 mm of gold standard for normal cases were 95.5±3.7%, and 93.6±4.8% for lung diseased.

Diagnosis supporting algorithm for lymph node metastases from colorectal carcinoma on 18F-FDG PET/CT.

ObjectiveWe studied the improvement of the detect-ability of lymph node (LN) metastases from colorectal cancer in 18F-fluorodeoxyglucose positron emission tomography (FDG PET)/computed tomography (CT) by analyzing the acquired counts with a statistical method.MethodsThirty-nine metastatic LNs from 32 cases with colorectal cancer were included in this study. “Uptake region” was defined as the site where counts were higher than the average plus 3 standard deviations (SDs) on each transaxial image of FDG PET. After the initial uptake regions were selected, these high accumulation areas were automatically excluded from consideration thereafter. This method was repeated and new uptake regions were identified. This method was repeated up to five times. After that, the stacked-up uptake regions were compared with computed tomography (CT) images, and the high accumulation areas that were superimposed on the normal structures, such as intestine, vessels, and ureters, were excluded from the consideration. The remaining uptake regions were diagnosed as metastatic LNs, and the detectability of LN metastases was calculated. We then compared these statistical results with the results obtained on the basis of visual assessments by radiologists.ResultsOur proposed method showed the best results when the procedures were repeated three times in the light of detectability. After being repeated three times, this method detected 15/23 (65.2%) metastatic LNs in the first LN group, 16/16 (100%) in the second-third LN groups and 31/39 (79.4%) in the total LNs, whereas the radiologists diagnosed 8/23 (34.8%) of metastatic first LNs, 12/16 (75.0%) in the second-third LNs and 20/39 (51.3%) in the total LNs. A statistically significant difference was observed between the result of iteration number 3 and that by radiologists as for the second—third LNs and the total LNs.ConclusionsThis study suggests that our proposed statistical method could improve the detectability of LN metastases from colorectal cancer. Our method will help radiologists to detect small metastatic lesions such as LN metastases.

Stochastic tracking of small pulmonary vessels in human lung alveolar walls using synchrotron radiation micro CT images

Small pulmonary vessel networks (arteriole and venule) provide a significant insight into understanding the alveolated structure in the human acinus. However, automatic extraction of small pulmonary vessels is a challenge due to the presence of abundant complexities in the networks. We thereby introduce a stochastic framework, a particle filter, to track small vessels running inside alveolar walls in human acinus using synchrotron radiation micro CT (SRμCT) images. We formulated vessel tracking using a non-linear sate space which captures both smoothness of the trajectories and intensity coherence along vessel orientations. In the particle filter scheme, we computed the proposal distribution by using the orientation distribution function (ODF), which is estimated as the combination of three different profiles; appearance, directional, and medialness profiles. To model the posterior distribution, we obtained voxels inside cylindrical tube which encapsulated a local vessel part. We constructed the prior distribution using the von Mises-Fisher (vMF) distribution on a unit sphere. At the same time, we detected branches of a vessel by analyzing the dominance of local vessel orientations through the vMF mean shift algorithm. Given a seed point, the method is able to locate the optimal vessel networks inside alveolar walls. Applying the method to the SRμCT images of the human lung acini, we demonstrate its potential usefulness to extract the trajectories of small pulmonary vessels running inside the alveolar walls.