Noboru Niki

Computer-aided diagnosis system for lung cancer based on retrospective helical CT image

In this paper, we present a computer-aided diagnosis (CAD) system for lung cancer to detect nodule candidates at an early stage from the present and the early helical CT screening of the thorax. We developed an algorithm that can compare automatically the slice images of present and early CT scans for the assistance of comparative reading in retrospect. The algorithm consists of the ROI detection and shape analysis based on comparison of each slice image in the present and the early CT scans. The slice images of present and early CT scans are both displayed in parallel and analyzed quantitatively in order to detect the changes in size and intensity affection. We validated the efficiency of this algorithm by application to image data for mass screening of 50 subjects (total: 150 CT scans). The algorithm could compare the slice images correctly in most combinations with respect to physicians point of view. We validated the efficiency of the algorithm which automatically detect lung nodule candidates using CAD system. The system was applied to the helical CT images of 450 subjects. Currently, we are carrying out the clinical field test program using the CAD system. The results of our CAD system have indicated good performance when compared with physicians diagnosis. The experimental results of the algorithm indicate that our CAD system is useful to increase the efficiency of the mass screening process. CT screening of thorax will be performed by using the CAD system as a counterpart to the double reading technique actually used in herical CT screening program, not by using the film display.

Development of a Novel Computer-aided Diagnosis System for Automatic Discrimination of Malignant From Benign Solitary Pulmonary Nodules on Thin-section Dynamic Computed Tomography

Objectives: As an application of the computer-aided diagnosis of solitary pulmonary nodules (SPNs), 3-dimensional contrast-enhanced (CE) dynamic helical computed tomography (HCT) was performed to evaluate temporal changes in the internal structure of nodules to differentiate between benign nodules (BNs) and malignant nodules (MNs). Methods: There were 62 SPNs (35 MNs and 27 BNs) included in this study. Scanning (2-mm collimation) was performed before and 2 and 4 minutes after CE dynamic HCT. The CT data were sent to a computer, and the pixels inside the nodule were characterized in terms of 3 parameters (attenuation, shape index, and curvedness value). Results: Based on the CT data at 4 (MN: 1.81-27.1, BN: −42.8 to −3.29) minutes after CE-dynamic HCT, a score of 0 or higher can be assumed to indicate an MN. Conclusions: Three-dimensional computer-aided diagnosis of the internal structure of SPNs using CE dynamic HCT was found to be effective for differentiating between BNs and MNs.

Example-based assisting approach for pulmonary nodule classification in three-dimensional thoracic computed tomography images1

RATIONALE AND OBJECTIVES An example-based assisting approach that supports decision making in classifying pulmonary nodules in 3-dimensional (3D) thoracic computed tomography images has been developed. MATERIALS AND METHODS The example-based assisting approach retrieves and displays nodules that exhibit morphologic and internal profiles consistent to the nodule in question. It uses a 3D computed tomography image database containing 143 pulmonary nodules for which diagnosis is known. The central module makes possible analysis of the query nodule image and extraction of the features of interest: shape, surrounding structure, and internal structure of the nodules. The principal axes and the compactness characterize the nodule shape. The surrounding and internal structures are represented by the distribution pattern of computed tomography density value and 3D curvature indexes. The nodule representation is then used for computing a similarity measure such as a correlation coefficient and a malignant likelihood of the query nodule. The malignant likelihood is estimated by the difference between the representation patterns of the query case and the retrieved lesions. The Mahalanobis distance was adopted as the difference measure. The approach performance was assessed through leave-one-out method by the false-positive rate. RESULTS Given a query nodule image, the proposed method retrieved benign and malignant images similar to the query case and provided its malignant likelihood. The number of cases that obtained enough number of the retrieved cases for estimating the malignant likelihood was 107 cases (malignant, 70; benign, 37) in our database. Sensitivity was 91.4% (64 of 70 malignant nodules), specificity was 51.4% (19 of 37 benign nodules), and accuracy values were 77.6% (83 of 107 nodules). CONCLUSION Preliminary assessment of this approach showed that an example-based assisting approach is an effective tool for making the diagnostic decision in the classification of pulmonary nodules using the nodule image database.

Classification of pulmonary nodules in thin-section CT images based on shape characterization

Shape characterization of small pulmonary nodules plays a significant role in differential diagnosis that discriminates malignant and benign nodules at early stages of pulmonary lesion development. This paper presents a method to characterize small pulmonary nodules based on the morphology of the development of lung lesions in thin-section CT images. The feature extraction algorithms are designed to extract the shape characteristic parameters from three-dimensional (3-D) nodule images using surface curvatures and ridge line. Experiments which show the feasibility of our method to improve the diagnostic accuracy are also demonstrated by applying the method to nodule images.

A comparison of Hopfield neural network and Boltzmann machine in segmenting MR images of the brain

The segmentation of the images obtained from magnetic resonance imaging is an important step in the visualization of soft tissues in the human body. In this preliminary study, we report an application of the Hopfield neural network for the multispectral unsupervised classification of head magnetic resonance images. We formulate the classification problem as a minimization of an energy function constructed with two terms, the cost-term which is the sum of the squares errors, and the second term is a temporary noise added to the cost-term as an excitation to the network to escape from certain local minimums and be more close to the global minimum. We present here the segmentation result with two and three channels data obtained using the Hopfield neural network approach. We compare these results to those corresponding to the same data obtained with the Boltzmann machine approach.

Lung cancer detection based on helical CT images using curved surface morphology analysis

Lung cancer is known as one of the most difficult cancers to cure. The detection of lung cancer in its early stage can be helpful for medical treatment to limit the danger. A conventional technique that assists the detection uses helical CT, which provides information of 3D cross sectional images of the lung. We expect that the proposed technique will increase diagnostic confidence. However, mass screening based on helical CT images leads to a considerable number of images for the diagnosis, this time-consuming fact makes it difficult to be used in the clinic. To increase the efficiency of the mass screening process, we had proposed a computer-aided diagnosis (CAD). In this paper, we describe lung cancer detection based on helical CT Images using curved surface morphology analysis. Firstly, we extract the lung area from the original image. Secondly, we compute shape index value of the lung area. Thirdly, we extract the ROI (Region Of Interest) from the computed shape index value. Finally, we apply the diagnosis rule using neural network and detect the suspicious regions. We show here the result of our algorithm which is applied to helical CT images of 390 patients.

Extraction algorithm of pulmonary fissures from thin-section CT images based on linear feature detector method

Describes a new automatic extraction algorithm of the pulmonary major and minor fissures from three-dimensional (3-D) chest thin-section images of helical computed tomography (CT). These fissures are used for the analysis of pulmonary conformation and the diagnosis of lung cancer. This algorithm consists mainly of the correction and the emphasis of a 2-D linear shadow. The authors applied the proposed algorithm to 25 sets of CT examinations of 12 patients. The results showed that major and minor fissures can be extracted by the proposed algorithm, without reference to streak artifacts on axial CT images by the beam hardening effect, and the motion artifacts by the cardiac beat.

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

PURPOSE Quantification 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. METHODS Institutional 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. RESULTS The κ values for interobserver agreement of the classification using two nodule segmentation results were 0.921 (P < 0.001) and 0.903 (P < 0.001), respectively. The κ values for the variability in the classification task using two decision trees were 0.981 (P < 0.001) and 0.981 (P < 0.001), respectively. All the NSCLCs were classified into one of five categories (type α, n = 8; type β, n = 38; type γ, n = 103; type δ, n = 112; type ɛ, n = 193) by using a decision tree. Using a multivariate Cox regression analysis, the classification (hazard ratio 5.64; P = 0.008) and disease stage (hazard ratio 8.33; P < 0.001) were identified as being associated with an increased recurrence risk. CONCLUSIONS The quantitative five-category classifier presented here has the potential to provide an objective classification of NSCLC nodules that is strongly correlated with prognostic factors.

Shape analysis of pulmonary nodules based on thin-section CT images

Shape characterization of small pulmonary nodules plays a significant role in differential diagnosis that discriminates malignant and benign nodules at early stages of pulmonary lesion development. This paper presents a method to characterize small pulmonary nodules based on the morphology of the development of lung lesions in thin section CT images. The feature extraction process focuses on the difference between the malignant and benign surface characteristics. Experiments to show its feasibility to improve the diagnostic accuracy are also demonstrated by applying the algorithm to eighteen cases including twelve malignant and six benign nodules.

Computer aided diagnosis system for lung cancer based on helical CT images

Lung cancer is known as one of the most difficult cancers to cure. The detection of lung cancer in its early stage can be helpful for medical treatment to limit the danger. A conventional technique that assists the detection uses helical CT, which provides information of 3D cross sectional images of the lung. We expect that the proposed technique will increase diagnostic confidence. However mass screening based on helical CT images leads to a considerable number of images for diagnosis, this time-consuming fact makes it difficult to be used in the clinic. To increase the efficiency of the mass screening process, we developed an algorithm for automatic detection of lung cancer candidates based on the helical CT images. Our algorithm consists of analysis and diagnosis procedures. In the analysis procedure, we extract the lung regions and the pulmonary blood vessel regions and we analyze the features of these regions using image processing techniques In the diagnosis procedure, we define diagnosis rules based on these features, and we detect tumor candidates using these rules. The diagnostic algorithm is applied to the helical CT images of 450 cases which have been diagnosed by three radiologists. Our system detected all tumors which were suspected to be lung cancer by the experts. Currently, we are planning to carry out a field test using our algorithm to evaluate the efficiency for visual diagnosis.