健山 智子

This study elucidated the effects of a three-dimensional k-space trajectory incorporating the partial Fourier (PF) technique on a time-intensity curve (TIC) in a dynamic contrast-enhanced magnetic resonance imaging of a typical malignant breast tumor using a digital phantom. Images were obtained from the Cancer Imaging Archive Open Data for Breast Cancer, and 1-min scans with high temporal resolution were analyzed. The order of the k-space trajectory was set as Linear (sequential), Low-High (centric), PF (62.5%; Z-, Y-, and both directions), and Low-High Radial. k0 (center of the k-space) timing and TIC shape were affected by the chosen k-space trajectory and implementation of the PF technique. A small TIC gradient was obtained using a Low-High Radial order. If the k-space filling method (particularly the radial method) produces a gentle TIC gradient, misinterpretation could arise during the assessment of tumor malignancy status.

This initial study aimed at testing whether fat-containing agents can be used for the fat mass estimation methods using magnetic resonance imaging (MRI). As an example for clinical application, fat-containing agents (based on soybean oil, 10% and 20%), 100% soybean oil, and saline as reference substances were placed outside the proximal femurs obtained from 14 participants and analyzed by 0.3 T MRI. Fat content was the estimated fat fraction (FF) based on signal intensity (SIeFF, %). The SIeFF values of the femoral bone marrow, including the femoral head, neck, shaft, and trochanter area, were measured. MRI data were compared in terms of bone mineral content (BMC) and bone mineral density (BMD) by dual-energy X-ray absorptiometry (DXA) in the proximal femur. Twelve pig femurs were also used to confirm the correlation between FF by the DIXON method and SIeFF. According to Pearson's correlation coefficient, the SIeFF and total BMC and BMD data revealed strong and moderate negative correlations in the femoral head (r <  - 0.74) and other sites (r =  - 0.66 to - 0.45). FF and SIeFF showed a strong correlation (r = 0.96). This study was an initial investigation of a method for estimating fat mass with fat-containing agents and showed the potential for use in MRI. SIeFF and FF showed a strong correlation, and SIeFF and BMD and BMC showed correlation; however, further studies are needed to use SIeFF as a substitute for DXA.

Robust hand gesture recognition has been playing a significant role in the field of human-computer interaction for a long time, but it is still full of challenges due to many accept such as cluttered backgrounds and hand self-occlusion. With the help of depth information, depth-based methods have better performance, but the depth cameras are not as widely used and affordable as color cameras. Therefore, in this paper, we propose a two-stage deep convolutional neural network (CNN) architecture for accurate color-based hand gesture recognition. The first stage performs generation of pseudo-depth hand images from color images and the second stage recognizes hand gesture classes using both the color image and its pseudo-depth hand image. The generation stage architecture is based on an image-to-image translation network. In the recognition stage, a two-stream CNN architecture with color image and its pseudo depth image is proposed to improve the color image-based recognition performance. We also propose two strategies in two-stream fusion: feature fusion and committee fusion. To validate our approach, we construct a new dataset called MaHG-RGBD dataset. Experiments demonstrate that our approach significantly improves the performance in RGB-only recognition for hand gestures.

Here, we present a new dataset, named MaHGRGBD, including 25 hand gestures performed by 15 participants as viewed from multiple angles. This dataset is intended to train models for deep -learning -based hand -gesture recognition. Unlike existing datasets, MaHG-RGBD includes not only front views (tilt angle = 0) but also the titled views (tilt angle = 45 degrees), which are often needed when there are space constraints. In addition, the dataset includes pairs of synchronized color and depth images of the hand region that are well segmented. Users can utilize just one of the image modalities or both depending on the application. This dataset includes a wide variety of different gestures classes: a total of 25 hand gestures. We evaluate the recognition accuracy of 25 different hand gestures using deep learning as a benchmark with this dataset. The MaHG-RGBD dataset is available at http: /AN ritsumei.ac.jp/MaHG-RGBD.

肝臓疾患の一つである肝硬変症は、肝臓表面の凹凸化に加えて右葉の萎縮・左葉の腫大、脾臓の腫大化という臨床的特徴が観測される。先行研究では、主成分分析(PCA)で得られた形態変化の成分に基づいて肝硬変症の特徴を示す有効成分を選択することで、正常症例と肝硬変症例を分類する肝硬変症支援診断法を提案し、その有効性を示した。しかし、肝硬変症はその重症度に応じて3つのステージが存在することから、支援診断には正常症例と肝硬変症例の識別に加え、さらにその進行度合いを示すステージ予測が必要である。本研究では、ステージ予測にも適応するため、部分最小二乗(Partial Least Square:PLS)回帰を用いた新たな肝硬変支援診断への応用を提案し、支援診断の精度向上を目指す。(著者抄録)

Probabilistic atlas based on human anatomical structure has been widely used for organ segmentation. The challenge is how to register the probabilistic atlas to the patient volume. Additionally, there is the disadvantage that the conventional probabilistic atlas may cause a bias toward the specific patient study due to a single reference. Hence, we propose a template matching framework based on an iterative probabilistic atlas for organ segmentation. Firstly, we find a bounding box for the organ based on human anatomical localization. Then, the probabilistic atlas is used as a template to find the organ in this bounding box by using template matching technology. Comparing our method with conventional and recently developed atlas-based methods, our results show an improvement in the segmentation accuracy for multiple organs (p &lt 0:00001).

Probabilistic atlas based on human anatomical structure has been widely used for organ segmentation. The challenge is how to register the probabilistic atlas to the patient volume. Additionally, there is the disadvantage that the conventional probabilistic atlas may cause a bias toward the specific patient study due to a single reference. Hence, we propose a template matching framework based on an iterative probabilistic atlas for organ segmentation. Firstly, we find a bounding box for the organ based on human anatomical localization. Then, the probabilistic atlas is used as a template to find the organ in this bounding box by using template matching technology. Comparing our method with conventional and recently developed atlas-based methods, our results show an improvement in the segmentation accuracy for multiple organs (p < 0.00001).

In medical, scientific, and other fields, transparent surface visualization is useful for investigating inner three-dimensional (3D) structures. Such visualization typically involves the use of polygon graphics in which the polygons must be sorted along the line of sight. However, sorting involves considerable computation time for large-scale data. Furthermore, the order of polygons in the sorting can often become indefinite, especially for intersecting surfaces. Therefore, particle-based volume rendering that does not require sorting is proposed as a transparent-rendering method. The proposed method obtains slice images with non-uniform opacity using color and opacity maps similar to volume rendering. The method initially generates the particles, a process it performs only once. It additionally employs particle shuffling, which requires considerably less computation time than particle sorting. To demonstrate the efficacy of the proposed method, we rendered 3D-fused images, including slice-slice and volume-slice images, for medical volumetric data. The results show that the performance of the proposed method is satisfactory in cases in which the area of the particle is greater than that of the cell.

The purpose of this study is to construct a system for surgical assistance by touchless interactions. In the clinical site, surgeons usually need to use some contacting devices to display or visualize medical images and check the anatomic information of the patient. However, after operating the visualization device, re-sterilization is necessary in order to maintain hygiene. Though some touchless surgery support systems using Kinect have been developed, their functions are not enough for surgical support. In this paper, we develop a new system, which can visualize 3D medical image by simple touchless single-handed interactions.

In recent years, there has been increasing interest in statistical shape modeling of human anatomy. The statistical shape model can capture the morphological variations of human anatomy. Since liver cirrhosis will cause significant morphological changes, the authors propose a computer-aided diagnosis method for liver cirrhosis based on statistical shape models. In the proposed method, the authors first construct a statistical shape model of the liver using 50 clinical CT datasets (25 sets of normal data and 25 sets of abnormal data). The authors apply the marching cubes algorithm to convert the segmented liver volume to a triangulated mesh surface containing 1000 vertex points. The coordinates of these vertex points are used to represent the 3D liver shape as a shape vector. After normalization and identification of correspondences between all datasets, principal component analysis (PCA) is employed to find the principal variation modes of the shape vectors. Then the authors propose a mode selection method based on class variations between the normal class and abnormal class. The authors found that the top two modes of class variations are most effective for the classification of normal and abnormal livers. The classification rate of abnormal livers and normal livers by the use of a simple linear discriminant function were 84% and 80%, respectively. (C) 2014 by Wiley Periodicals.

Probabilistic atlases based on human anatomy structure have been widely used for organ segmentation, which is used as a prior probability in a Bayes framework. The challenge is how to register the probabilistic atlas to the patient volume. Taking these into consideration, we propose a template matching framework based on the probabilistic atlas for spleen segmentation. Firstly, we find a bounding box of the spleen based on human anatomical localization, which is the statistical geometric location of spleens. Then, the probabilistic atlas is used as a template to find the spleen in this bounding box by using template matching technology. We apply our method into 60 datasets including normal and pathological cases. The Dice/Tanimoto volume overlaps are 0.922/0.857, the root-mean-squared error (RMSE) is 1.992 mm. The algorithm is robust to segment normal and abnormal spleens, such as the presence of tumors and large morphological changes. Meanwhile, our proposed method was compared with conventional atlas-based methods. Results demonstrate that segmentation accuracy improved using our method.

In this paper, we propose an automated method for assessment of small bowel contraction movement based on the simple linear iterative clustering (SLIC) with cine-MRI. In our proposed method, the small bowel in each frame is considered as a super-pixel and is first segmented by the use of SLIC. SLIC performs a local clustering of pixels in a 3-D space defined by intensity and (x, y) coordinates. Then the dynamic area change of the segmented small bowel is used for quantitative analysis of the small bowel motility function. Compared with existing methods, our proposed method can achieve better results.

A random walks-based (RW) segmentation method has been gaining popularity in recent years with its ability to interactively segment the objects with minimal guidance. It has potential applications in segmenting the 3D image. However, due to the large computational burden of the classical RW algorithm, it is a challenge to use this algorithm to segment 3D medical images interactively. Hence, a knowledge-based segmentation framework for the liver is proposed based on random walks and narrow band threshold (RWNBT). Our strategy is to employ the previous segmented slice to achieve a prior knowledge (the shape and intensity constraints) of liver for automatic segmentation of the adjacent slice. With a small number of user-defined seeds, we can obtain the segmentation results of the start slice in the volume which would be used as the prior knowledge of the segmented organ. According to this intensity constraints, the "Candidate Pixels" image can be generated by thresholding the organ models with Gaussian Mixture Model (GMM), which can remove the noise and non-liver parts. Furthermore, the object/background seeds can be dynamically updated for the adjacent slice by combining a narrow band threshold (NBT) method and the shape constrains. Finally, a combinational random walker algorithm is applied to automatically segment the whole volume in a slice-by-slice manner. Comparing our method with conventional RW and the state-of-the-art interactive segmentation methods, our results show an improvement in the accuracy for liver segmentation.

Ubiquitous computing is an important technology in medicine that is predicted to support doctors anywhere and anytime. To help achieve it, this paper develops the Interactive Segmentation and Visualization System for Medical images on Mobile devices (ISVS_M-2), which originally was designed to work on workstations, but also works on a wide range of mobile devices via a mobile client-server platform. The developed ISVS_M-2 basically consists of three modules: asegmentation module that is implemented on a server; commutation modules on both the server and mobile device; and interactive and visualization modules on the mobile device that not only give visualization of internal information of an organ model, but also interactively refine organ segmentation according to user experience. With interaction via a computer graphic interface on the mobile device, and communication via the mobile client-server platform, ISVS_M-2 offers users a novel and efficient approach to computer-aided medicine.

VR技術を用いた臨床読影支援は、コントローラやマーカ等の接触デバイス操作による患者体内の臨床情報提示が一般的である。そのため、手術現場ではデバイス装着の制限や接触時の滅菌処理、執刀医の術場離脱が要求される。よって、衛生状態を維持したまま執刀医が術野離脱することなく、自ら操作により必要な患者体内情報を直感的に提示する手術支援システム確立は、臨床現場からも要求が高い。Kinectは内蔵の深度センサーによって人物の姿勢や動作をコントローラやマーカなしで認識できる非接触性デバイスであることから、臨床現場の要望に応えるデバイスとして十分に期待が高い。Kinectを用いたハンズフリーかつ直感的なインタラクション操作で、清潔を維持したまま執刀医自ら患者の必要な臨床情報へ直感的にアクセスすることが可能である。本研究では、Kinectを用いたハンズフリー対話操作による医用画像三次元可視化システムを検討し、新たな手術支援の構築を目指す。(著者抄録)

A framework is proposed for automatic liver segmentation from CT volumes using probabilistic atlases and template matching techniques. Probabilistic atlases of human anatomy have been widely used for organ segmentation, which is used as a prior probability in a Bayes framework. The challenge is how to register the atlas to the patient volume. In this paper, we propose a template matching based technique for probabilistic atlas based organ segmentation. In our proposed method, we first find a Region of Interest (ROI) of the organ, which is based on human anatomic structure, and then the probabilistic atlas is used as a template to find the organ in the ROI by the use of template matching.