中口 俊哉

従来のICG観察システムでは，蛍光をモニタで観察するため，蛍光部位と観察部位の位置関係把握が困難であり，またシステムが高価であることから広く普及するに至らなかった．そこで本研究では，市販のUSBカメラとレーザープロジェクタを用いて蛍光を取得，投影し，より低価格で直感的な観察システムを構築する．本システムは乳がんのセンチネルリンパ節同定，皮膚がん，リンパ浮腫など，体表の比較的浅い部分の病変部を主な観察対象とする．本システムでは赤外LEDを用いてICGを励起させ，励起画像と非励起画像の差分をとることで蛍光部位を抽出する．取得画像は空間コード化法を用いて歪み補正を行い，レーザープロジェクタで対象部位に投影する．投影距離200mmにおける平均投影誤差は0.5mm以内であった．ファントム実験の結果，最大深さ12mmのICG蛍光を観察可能であった．また動物実験によって，本システムの臨床における実現可能性が示唆された．

これまで室内照明としては蛍光灯や白熱電球が一般的であり,また,近年はLED照明が使用されてきたが,省エネルギー性やデザイン性の観点から有機ELパネルが次世代の照明技術として注目を浴びている.そこで,本研究では有機EL照明が生体に与える影響を主観的,客観的に検証した.実験は卓上での作業を想定したタスクを設定し,長時間のタスクにより起きる生体影響を主観的評価によって検証する単一刺激評価,二種類の照明間において視覚疲労の違いを評価する二重刺激評価を行った.単一刺激評価ではタスク中の自律神経活動状態をモニタリングすることで,有機EL照明とその他の照明が生体に与える影響の違いについて検討した.

3D映像は様々な分野で普及しており,その普及に件い3D映像が生体に与える影響が様々な手法で検証されている.しかしながら両眼に映る画面輝度が生体に与える影響についてはまだ解明されていない.そこで二重刺激,単一刺激による主観評価を行い,画面輝度が生体へ及ぼす影響を検証した.二重刺激では輝度変化による影響が示され,長時間の視聴による単一刺激よりも映像を比較する二重刺激が画面輝度による生体影響を評価するのに適していることが示された.また,同条件の実験を複数回行うことで実験の再現性について検証し,実験精度の確認,検証を行った.

We propose a VR based injection training system using Standardized Patient (SP) with an original haptic needle which can represent a haptic expression. SP is trained to realistically portray a real patient. In the proposed system, trainee can virtually puncture the SP using the haptic needle. In addition, the haptic needle can represent a haptic expression of needle insertion of the virtual anatomical model. By using the proposed system, trainee can feel virtual puncture as well as operating for real patient. © 2012 The authors and IOS Press. All rights reserved.

胃がんは日本で最も罹患率の高いがんである。また、胃がんの他臓器への浸潤の有無は手術の外科的治療方針に大きく影響を与えるため、CT画像などから術前に診断されていることが望ましい。しかし、がんの境界が不明瞭であるためその診断が難しいという問題点があった。そこで、本研究では患者の体勢を上向きと横向きで撮影した2組のCT画像を用いて胃がん膵浸潤の有無を示す指標を提案する。その指標は、胃と膵臓の相対的な動きを計算することで求められると予想される。胃と膵臓の相対的な動きを算出するために患者の体勢を変えて撮影した2組のCT画像に対して3段階のレジストレーションを行った。レジストレーションの後、胃と膵臓付近の変化量マップが求められた。診断補助のためにその変化量マップを可視化しCT画像に重ね合わせた。提案手法を11人の胃がん患者に適用し、提案した指標の定量的な評価実験を行った。実験の結果、提案した指標の有用性が示された。(著者抄録)

In this work, we propose a shape-based liver segmentation approach using a patient specific knowledge. In which, we exploit the relation between consequent slices in multi-slice CT images to update the shape template that initially determined by the user. Then, the updated shape template is integrated with the graph cuts algorithm to segment the liver in each CT slice. The statistical parameters of the liver and non-liver tissues are initially determined according to the initial shape template and it is consequently updated from the nearby slices. The proposed approach does not require any prior training and it uses a single phase CT images; however, it is talented to deal with complex shape and intensity variations. The proposed approach is evaluated on 20 CT images with different kinds of liver abnormalities, tumors and cysts, and it achieves an average volumetric overlap error of 6.4% and average symmetric surface distance (ASD) of 0.8 compared to the manual segmentation.

This paper presents the subjective evaluations of multiple observers as part of an investigation into the relationship between CC parameters of digital mockups and visual sensibility. In our experiments, the specular appearance of CG imagery is reproduced on an actual mockup using a projector-camera system. The specular intensity and position are evaluated in terms of magnitude and inauthenticity. For the specular intensity evaluation, it was found that changes to specular intensity in the CG parameters were equal to the changes in the brightness sensibility of the object viewed. Furthermore, the results of inauthenticity evaluations clarified the limitations of the viewpoint range. The specular appearance from the 600 viewpoint gave observers the impression that the form and position of the specular reflection were inauthentic. Therefore, it was determined that the control of appearance in our digital mockup was only suitable for observations within the range from -45 degrees to 45 degrees.

BioEncoding is a recently proposed template protection scheme, based on the concept of cancelable biometrics, for protecting biometric templates represented as binary strings such as IrisCodes. Unlike existing techniques, BioEncoding does not require user-specific keys and/or tokens during verification. Besides, it satisfies all the requirements of the cancelable biometrics construct without deteriorating the matching accuracy of the original biometric system. However, although the cancelable transformation employed in BioEncoding is non-invertible for a single protected template, it might be possible to recover the original biometric template by correlating several protected templates created from the same biometric signal. In this paper, the vulnerability of BioEncoding to correlation attacks is investigated. First, we show that cancelable templates obtained using BioEncoding are indeed vulnerable to correlation attacks. Then, we propose three different approaches to improve the security of BioEncoding against this type of attacks. The effectiveness of adopting the suggested approaches is validated and their impact on the matching accuracy is investigated empirically using CASIA-IrisV3-Interval dataset. Experimental results confirm the efficacy of the proposed approaches and show that they do not affect the matching accuracy of the recognition system.

Using gaze information in designing tone-mapping operators has many potentials over traditional global tone-mapping operators. In this paper, we evaluate a recently proposed real-time tone mapping operator based on gaze information and show that it is highly dependent on the input scene. We propose an important modification to the evaluated method to relief this dependency and to enhance the appearance of the resultant images using smaller processing area. Experimental results show that our method outperforms the evaluated technique.

Improving the security of biometric template protection techniques is a key prerequisite for the widespread deployment of biometric technologies. BioEncoding is a recently proposed template protection scheme, based on the concept of cancelable biometrics, for protecting biometric templates represented as binary strings such as iris codes. The main advantage of BioEncoding over other template protection schemes is that it does not require user-specific keys and/or tokens during verification. Besides, it satisfies all the requirements of the cancelable biometrics construct without deteriorating the matching accuracy. However, although it has been shown that BioEncoding is secure enough against simple brute-force search attacks, the security of BioEncoded templates against more smart attacks, such as record multiplicity attacks, has not been sufficiently investigated. In this paper, a rigorous security analysis of BioEncoding is presented. Firstly, resistance of BioEncoded templates against brute-force attacks is revisited thoroughly. Secondly, we show that although the cancelable transformation employed in BioEncoding might be non-invertible for a single protected template, the original iris code could be inverted by correlating several templates used in different applications but created from the same iris. Accordingly, we propose an important modification to the BioEncoding transformation process in order to hinder attackers from exploiting this type of attacks. The effectiveness of adopting the suggested modification is validated and its impact on the matching accuracy is investigated empirically using CASIA-IrisV3-Interval dataset. Experimental results confirm the efficacy of the proposed approach and show that it preserves the matching accuracy of the unprotected iris recognition system.

Uneven distribution of skin color is one of the biggest concerns about facial skin appearance. Recently several techniques to analyze skin color have been introduced by separating skin color information into chromophore components, such as melanin and hemoglobin. However, there are not many reports on quantitative analysis of unevenness of skin color by considering type of chromophore, clusters of different sizes and concentration of the each chromophore. We propose a new image analysis and simulation method based on chromophore analysis and spatial frequency analysis. This method is mainly composed of three techniques: independent component analysis (ICA) to extract hemoglobin and melanin chromophores from a single skin color image, an image pyramid technique which decomposes each chromophore into multi-resolution images, which can be used for identifying different sizes of clusters or spatial frequencies, and analysis of the histogram obtained from each multi-resolution image to extract unevenness parameters. As the application of the method, we also introduce an image processing technique to change unevenness of melanin component. As the result, the method showed high capabilities to analyze unevenness of each skin chromophore: 1) Vague unevenness on skin could be discriminated from noticeable pigmentation such as freckles or acne. 2) By analyzing the unevenness parameters obtained from each multi-resolution image for Japanese ladies, age-related changes were observed in the parameters of middle spatial frequency. 3) An image processing system modulating the parameters was proposed to change unevenness of skin images along the axis of the obtained age-related change in real time.

In this paper, we propose optical path-length matrix method for high-speed simulation of photon migration in human skin. The optical path-length matrix is defined as the probability density distribution of optical path-length in the skin. Generally, Monte Carlo simulation is used to simulate a skin reflectance, since it can simulate the reflectance accurately. However, it requires a huge computation time, thus this is not easily applicable in practical imaging system with large number of pixels. On the other hand, the proposed optical path-length matrix method achieves the simulation in shorter time. The skin model was assumed to be two-layered media of the epidermal and dermal layers. For obtaining the path-length matrix, photon migration in the model without any absorption was simulated only once by Monte Carlo simulation for each wavelength, and the probabilistic density histograms of the optical path-length at each layer were acquired and stored in the optical path-length matrix. Skin spectral reflectance for arbitrary absorption can be calculated easily by accumulating all combination of an element in the above pre-recomputed path-length matrix and absorption coefficient based on the Beer-Lambert law. Our proposed method was compared with the conventional Monte Carlo simulation. Computational time of the proposed method was approximately two minutes; while that the conventional method was 15 hours. In addition, error margin of the proposed method was approximately less than 1.6%. This method would applied to skin spectral image analysis for skin chromophore quantification.

In this paper, we have developed a feature-based automatic color calibration by using an area-based detection and adaptive nonlinear regression method. Simple color matching of chartless is achieved by using the characteristic of overlapping image area with each camera. Accurate detection of common object is achieved by the area-based detection that combines MSER with SIFT. Adaptive color calibration by using the color of detected object is calculated by nonlinear regression method. This method can indicate the contribution of object's color for color calibration, and automatic selection notification for user is performed by this function. Experimental result show that the accuracy of the calibration improves gradually. It is clear that this method can endure practical use of multi-camera color calibration if an enough sample is obtained.

In order to satisfy the requirements of the cancelable biometrics construct, cancelable biometrics techniques rely on other authentication factors such as password keys and/or user specific tokens in the transformation process. However, such multi-factor authentication techniques suffer from the same issues associated with traditional knowledge-based and token-based authentication systems. This paper presents a new one-factor cancelable biometrics scheme for protecting IrisCodes. The proposed method is based solely on IrisCodes however, it satisfies the requirements of revocability, diversity and noninvertibility without deteriorating the recognition performance. Moreover, the transformation process is easy to implement and can be integrated simply with current iris matching systems. The impact of the proposed transformation process on the the recognition accuracy is discussed and its noninvertibility is analyzed. The effectiveness of the proposed method is confirmed experimentally using CASIA-IrisV3-Interval dataset. © 2010 IEEE.

A method is proposed to separately model the mechanical dot gain and the optical dot gain. First, an iterative algorithm is proposed to estimate the spatio-spectral transmittance of ink layer from the spatio-spectral reflectance of color halftone print measured with the reflection optical microscope attached with the liquid crystal tunable filter (LCTF). The spatio-spectral transmittance of ink layer is not affected by the optical dot gain and is only affected by the mechanical dot gain. Next, a model is proposed to estimate the effective dot coverage using the estimated spatio-spectral transmittance of ink layer. It corresponds to the analysis of mechanical dot gain. Next, a model is proposed to estimate Yule-Nielsen's n parameter using the effective dot coverage. It corresponds to the analysis of optical dot gain. Finally, the prediction accuracy of the proposed model is evaluated by the ΔE 94 between the measured and predicted spectral reflectance of offset printing images with cyan and magenta inks. The prediction accuracy of the proposed model was significant since the average ΔE94 and the maximum ΔE94 of all samples between the measured spectral reflectance and the predicted spectral reflectance were 0.62 and 1.37, respectively.

In many medical applications, the automatic segmentation of deformable organs from medical images is indispensable and its accuracy is of a special interest. However, the automatic segmentation of these organs is a challenging task according to its complex shape. Moreover, the medical images usually have noise, clutter, or occlusion and considering the image information only often leads to meager image segmentation. In this paper, we propose a fully automated technique for the segmentation of deformable organs from medical images. In this technique, the segmentation is performed by fitting a nonlinear shape model with pre-segmented images. The kernel principle component analysis (KPCA) is utilized to capture the complex organs deformation and to construct the nonlinear shape model. The pre-segmentation is carried out by labeling each pixel according to its high level texture features extracted using the over-complete wavelet packet decomposition. Furthermore, to guarantee an accurate fitting between the nonlinear model and the pre-segmented images, the particle swarm optimization (PSO) algorithm is employed to adapt the model parameters for the novel images. In this paper, we demonstrate the competence of proposed technique by implementing it to the liver segmentation from computed tomography (CT) scans of different patients.