Takuji Kiryu

Neurogenic tumors of the mediastinum and chest wall: MR imaging appearance.

Various histologic types of neurogenic tumors may originate in the mediastinum and chest wall. It is possible to make accurate diagnosis of these tumors by using the multiplanar capability and high contrast resolution of MR imaging because of these characteristic imaging findings. MR and histologic features of these tumors are illustrated and described in this essay.

Automatic recognition of lung lobes and fissures from multislice CT images

Computer-aided diagnosis (CAD) has been expected to help radiologists to improve the accuracy of abnormality detection and reduce the burden during CT image interpretations. In order to realize such functions, automated segmentations of the target organ regions are always required by CAD systems. This paper describes a fully automatic processing procedure, which is designed to identify inter-lobe fissures and divide lung into five lobe regions. The lung fissures are disappeared very fuzzy and indefinite in CT images, so that it is very difficult to extract fissures directly based on its CT values. We propose a method to solve this problem using the anatomy knowledge of human lung. We extract lung region firstly and then recognize the structures of lung vessels and bronchus. Based on anatomy knowledge, we classify the vessels and bronchus on a lobe-by-lobe basis and estimate the boundary of each lobe region as the initial fissure locations. Within those locations, we extract lung fissures precisely based on an edge detection method and divide lung regions into five lung lobes lastly. The performance of the proposed method was evaluated using 9 patient cases of high-resolution multi-slice chest CT images; the improvement has been confirmed with the reliable recognition results.

Preliminary Study for Automated Recognition of Anatomical Structure from Torso CT images

The anatomical human structure recognition is very important and necessary during the development of computer-aided diagnosis (CAD) system. In this paper, we propose an image processing scheme that can recognize the general structure of human torso by identifying the human torso region from CT images automatically and separating it into 7 parts: skin, subcutaneous fat, muscle, bone, diaphragm, thoracic cavity and abdominal cavity based on CT number distribution and spatial relations between different organ and tissue regions. We applied this scheme to 313 patient cases of torso CT images and confirmed its usefulness from the preliminary experiment

Automated Estimation of the Upper Surface of the Diaphragm in 3-D CT Images

This communication describes a fully automated method by which the position of the diaphragm surface can be estimated by deforming a thin-plate model to match the bottom surface of the lung in CT images. This method was applied to 338 X-ray CT scans, and its validity was proved by the experimental results.

Automated segmentations of skin, soft-tissue, and skeleton, from torso CT images

We have been developing a computer-aided diagnosis (CAD) scheme for automatically recognizing human tissue and organ regions from high-resolution torso CT images. We show some initial results for extracting skin, soft-tissue and skeleton regions. 139 patient cases of torso CT images (male 92, female 47; age: 12-88) were used in this study. Each case was imaged with a common protocol (120kV/320mA) and covered the whole torso with isotopic spatial resolution of about 0.63 mm and density resolution of 12 bits. A gray-level thresholding based procedure was applied to separate the human body from background. The density and distance features to body surface were used to determine the skin, and separate soft-tissue from the others. A 3-D region growing based method was used to extract the skeleton. We applied this system to the 139 cases and found that the skin, soft-tissue and skeleton regions were recognized correctly for 93% of the patient cases. The accuracy of segmentation results was acceptable by evaluating the results slice by slice. This scheme will be included in CAD systems for detecting and diagnosing the abnormal lesions in multi-slice torso CT images.

Lung structure recognition: a further study of thoracic organ recognitions based on CT images

Abstract We are developing a computer-aided diagnosis (CAD) system for extracting and recognizing thoracic organ regions from chest CT images. In a previous study, we have shown that this system can automatically recognize nine kinds of human organ and tissue regions from multi-slice CT images and provide some useful applications for visualizing CT images three-dimensionally (3-D) or two-dimensionally (2-D) based on preliminary recognition results. In this paper, we propose some further studies for this system that focus on lung structure recognition. These studies include: (1) identification of thoracic cage region, (2) extraction of major and minor fissures and classification of lung regions, (3) classification of lung surface and identification of hilus pulmonis. The proposed methods have been applied to three chest CT images for recognizing lung structure and have demonstrated promising results.

Multiple chondromatous hamartomas of the lung

Multiple chondromatous hamartomas of the lung, which are very rare, are a feature of Carney syndrome. The relation between the two entities is not clear.

Nonfunctional mediastinal parathyroid cyst: imaging findings in two cases

The authors describe the computed tomography (CT) and magnetic resonance (MR) imaging findings of a 69-year-old woman and a 69-year-old man with a nonfunctional mediastinal parathyroid cyst. In the described cases, unenhanced CT showed homogeneous areas of water density, and unenhanced MRI showed homogeneous areas that were isointense to cerebrospinal fluid, reflecting their serous fluid contents. Both cysts were located posterior to the left lower pole of the thyroid gland with an extension to the superior mediastinum, either anterior or posterior to the left brachiocephalic vein. CT and MR imaging findings of parathyroid cysts are nonspecific, and they are often difficult to differentiate from other cystic lesions located in the lower neck or in the superior mediastinum. However, a parathyroid cyst should be considered when radiologic images demonstrate its characteristic location, posterior to the thyroid gland, with an extension to the superior mediastinum.

Rounded atelectasis: delineation of enfolded visceral pleura by MRI.

Rounded atelectasis is an uncommon but increasingly recognized benign form of peripheral lung collapse (1). A variety of terms have been suggested for the entity, including contracted pleurisy, folded lung, pleuroma, atelectatic pseudotumor, Blesovsky syndrome, and shrinking pleuritis with atelectasis (2). Rounded atelectasis appears as a mass-like lesion that often mimics a pulmonary neoplasm (3). We report a case of rounded atelectasis shown as a clear delineation of enfolded visceral pleura by contrast-enhanced dynamic T1-weighted imaging.

Nodule detection in 3D chest CT images using 2nd order autocorrelation features

We have developed a new recognition approach using 2nd order autocorrelation and multi-regression analysis to detect a small (<7mm in diameter) lung nodules in chest 3D CT images. By combining our previous detection method of the template matching based on genetic algorithm, the detection performance was 94% true-positive rate at 2.05 false-positive marks per case using leave-one-out study