研究者詳細 - 神谷　亨

論文 - 神谷　亨

分割表示757 件中 21 - 40 件目／全件表示 >>

色彩の分離に基づいた3次元点群顔データからの口唇領域の抽出

木原，木村（野元），大河内，李, 中村, 神谷

第43回日本医用画像工学会大会 278 - 279 2024年08月

Ensemble Multiscale Attention UNetによる胸部CT 画像からの肺腫瘍領域の自動抽出

高橋神谷，寺澤，青木

第43回日本医用画像工学会大会 216 - 217 2024年08月

Multi-robot navigation based on velocity obstacle prediction in dynamic crowded environments 査読有り国際誌

Chen Y., Wang Y., Li B., Kamiya T.

Industrial Robot 51 ( 4 ) 607 - 616 2024年07月

記述言語：英語掲載種別：研究論文（学術雑誌）

Purpose: The purpose of this paper is to propose a new velocity prediction navigation algorithm to develop a conflict-free path for robots in dynamic crowded environments. The algorithm BP-prediction and reciprocal velocity obstacle (PRVO) combines the BP neural network for velocity PRVO to accomplish dynamic collision avoidance. Design/methodology/approach: This presented method exhibits innovation by anticipating ahead velocities using BP neural networks to reconstruct the velocity obstacle region; determining the optimized velocity corresponding to the robot’s scalable radius range from the error generated by the non-holonomic robot tracking the desired trajectory; and considering acceleration constraints, determining the set of multi-step reachable velocities of non-holonomic robot in the space of velocity variations. Findings: The method is validated using three commonly used metrics of collision rate, travel time and average distance in a comparison between simulation experiments including multiple differential drive robots and physical experiments using the Turtkebot3 robot. The experimental results show that our method outperforms other RVO extension methods on the three metrics. Originality/value: In this paper, the authors propose navigation algorithms capable of adaptively selecting the optimal speed for a multi-robot system to avoid robot collisions during dynamic crowded interactions.

DOI： 10.1108/IR-12-2023-0337

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85189666801&origin=inward

A Temporal Subtraction Technique for Phalange CR Image Using CNN 査読有り国際誌

Ono H., Kamiya T., Aoki T.

Journal of Advances in Information Technology 15 ( 7 ) 873 - 878 2024年07月

担当区分：責任著者記述言語：英語掲載種別：研究論文（学術雑誌）

X-ray examinations are widely used in the diagnosis of Rheumatoid Arthritis (RA). However, the condition of many phalanges and joints must be evaluated visually, which causes a lack of objectivity due to subjective evaluation by the physician and an increased workload for the physician in reading the images. In this paper, we propose an image analysis method for hand Computed Radiography (CR) images based on the temporal subtraction method to support the diagnosis of rheumatoid arthritis. The proposal method consists of three steps. First, a Convolutional Neural Network (CNN) model for semantic segmentation, which is efficient in terms of computational complexity and accuracy, is proposed to extract phalangeal regions. Second, a geometric-matching CNN with instance-specific optimization is used to align the phalangeal regions. Finally, the current image and the aligned past image are subtracted to visualize the temporal changes. We applied the proposed method to hand CR images and confirmed its effectiveness.

DOI： 10.12720/jait.15.7.873-878

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85212714373&origin=inward

Detection of mutation of driver gene from the CT images based on ensemble learning 査読有り国際誌

Kamiya, Y. Yoshifuku, T. Terasawa, T. Aoki

Computer Assisted Radiology and Surgery Proceedings of the 38th International Congress and Exhibition ( S97-S98 ) 2024年06月

Optimization of Corneal Parameters in Refractive Surgery 査読有り国際誌

Wang L., Li G., Liu Z., Huang Y., Shi B., Kamiya T.

ACM International Conference Proceeding Series 144 - 148 2024年04月

担当区分：最終著者記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

This study establishes a personalized human eye model for application in refractive surgery and proposes a method to optimize corneal parameters during surgery. The input data includes preoperative spherical equivalent, corneal curvature, and corneal thickness of the patient. Linear regression analysis and neural network methods are employed to predict corneal curvature and thickness, respectively, during the surgery based on the preoperative examination results. The results demonstrate that the neural network method performs accurately in predicting corneal ablation curvature, while the linear regression analysis method is effective in predicting ablation thickness. Utilizing these methods to predict corneal parameters in refractive surgery can assist surgeons in achieving precise corneal ablation and significantly reduce surgical risks.

DOI： 10.1145/3669754.3669776

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85203828242&origin=inward

Deep Learning Based Prediction of Heat Transfer Coefficient Using Spectrogram Images from Boiling Sound 査読有り国際誌

Mitsuyama, Umeno, Yabuki, Kamiya

29th International Conference on Artificial Life and Robotics 914 - 917 2024年02月

Unsepervised image registration based on Residual-connected DRMINE for diagnostic metastatic bone tumors 査読有り国際誌

Baba S., T. kamiya, T. Terasawa, T. Aoki

29th International Conference on Artificial Life and Robotics 1016 - 1020 2024年02月

Classification of circulating tumor cell clusters by morphological characteristics using convolutional neural network-support vector machine 査読有り国際誌

Park J., Ha S.M., Kim J., Song J.W., Hyun K.A., Kamiya T., Jung H.I.

Sensors and Actuators B: Chemical 401 2024年02月

記述言語：英語掲載種別：研究論文（学術雑誌）

Metastasis is the leading cause of cancer-associated deaths, and the circulating tumor cell (CTC) cluster plays a significant role as a precursor to metastasis. Thus, there is a great demand for high-throughput identification of rare CTC clusters for prognostic diagnosis. Immunofluorescence staining is considered the gold standard for identifying CTCs. However, as CTC clusters are extremely heterogeneous cells, multiple staining markers are required for accurate discrimination. Additionally, the staining procedure is tedious and the analysis of large amounts of stained images is labor-intensive and error-prone. Recently, machine learning-based identification has been introduced to achieve accurate discrimination, but they still rely on immunofluorescence staining for dataset preparation. In this study, we developed a hybrid algorithm, a convolutional neural network support vector machine (CNN-SVM), for the accurate classification of CTC clusters without immunofluorescence staining. In dataset preparation, the Wright–Giemsa staining was used to highlight the morphological features of the cells. Four morphological characteristics that display the unique traits of cells were drawn with each eigenvector, as a result of learning, the algorithm classified CTC clusters of various configurations with a sensitivity and specificity of > 90%. Therefore, our algorithm is expected to be a powerful tool for cancer diagnosis and prognosis.

DOI： 10.1016/j.snb.2023.134896

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85176212805&origin=inward

ランドマーク点の時系列変化と3次元構造に基づいた顔の非対称度評価法査読有り

木原，野元，大河内，李（天津工大），中村，神谷

バイオメディカル・ファジィ・システム学会誌 ( バイオメディカル・ファジィ・システム学会誌 ) 25 ( 2 ) 57 - 64 2024年02月

Three-dimensional quantification of postoperative facial asymmetry in patients with unilateral cleft lip and palate using facial symmetry plane 査読有り国際誌

Shiigi, Okawachi, Kamiya, Hosoki, Nomoto, Ratman, Amir, Ishihata, Nakamura

The Cleft Palate Craniofacial Journal 61 ( 2 ) 209 - 218 2024年02月

Unsupervised image registration based on Residual-connected DRMINE for diagnostic metastatic bone tumors 査読有り

Baba S., Kamiya T., Terasawa T., Aoki T.

Proceedings of International Conference on Artificial Life and Robotics 1016 - 1020 2024年01月

担当区分：責任著者記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

Metastatic tumors are frequently identified through follow-up surveillance using computed tomography (CT) scans. However, CT scans produce more than 100 images in an examination, which imposes a significant burden on radiologists and entails a potential risk of misdiagnosis. Temporal subtraction is utilized in Computer-Aided Diagnosis (CAD) and proves to be an effective technique in aiding image interpretation process for the radiologists. In this study, we focus on the preliminary stage of CAD development specialized in bone metastasis extraction, with a particular emphasis on rigid registration. We propose a novel rigid registration technique by augmenting DRMINE, which estimates mutual information using neural networks, with skip connections and normalization. From the three datasets, ten images were selected randomly from the cervical, thoracic, and lumbar regions. These images were then augmented with rotation as well as horizontal and vertical translations to create modified versions. The registration accuracy was assessed based on the Full Width at Half Maximum (FWHM) of the difference images. In the proposed method, FWHM values for the thoracic and lumbar regions of the spine exhibited a maximum reduction rate of 2.8% and a minimum reduction rate of 0.533%. However, the cervical spine region exhibited superior FWHM results with DRMINE compared to the proposed methodology. The proposed method was influenced by the capture area, but it indicated the potential to provide stable registration as the standard deviation decreased for all FWHM values.

Kyutacar

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85190241162&origin=inward

Physics-based Optical Coherence Tomography Angiography (OCTA) Image Correction for Shadow Compensation 査読有り国際誌

Li G., Wang K., Dai Y., Zheng D., Wang K., Zhang L., Kamiya T.

IEEE Transactions on Biomedical Engineering 2024年01月

担当区分：最終著者記述言語：英語掲載種別：研究論文（学術雑誌）

Optical coherence tomography (OCT) is being widely applied in clinical studies to investigate insight into the retina under the retinal pigment epithelium. Optical coherence tomography angiography (OCTA) is one of the functional extensions of OCT, for visualizing retinal circulation. Due to obstruction of light propagation, such as vitreous floaters or pupil boundaries, OCTA remains challenged by shadow artifacts that can disrupt volumetric data. Detecting and removing these shadow artifacts are crucial when quantifying indicators of retinal disease progression. We simplified an optical attenuation model of shadow formation in OCTA to a linear illumination transformation. And learn its parameters using an adversarial neural network. Our framework also consists of a sub-network for shadows automatic detection. We experimented our method on 28 OCTA images of normal eyes and compared the non-perfusion area (NPA), an index to measure retinal vascularity. The results showed that the NPA adjusted to a reasonable range after image processing using our method. Furthermore, we tested 150 OCTA images of synthesis artifacts, and the mean absolute error(MAE) values reached 0.83 after shadow removal.

DOI： 10.1109/TBME.2024.3478384

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85206991320&origin=inward

Super Resolution of Remote Sensing Image Using Improved SwinIR 査読有り国際誌

Matsumoto S., Kamiya T.

2024 Joint 13th International Conference on Soft Computing and Intelligent Systems and 25th International Symposium on Advanced Intelligent Systems, SCIS and ISIS 2024 2024年01月

担当区分：最終著者,　責任著者記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

In recent years, the number of satellite launches has been increasing. One of the reasons for this is the growing market for space development, which has led to the release of a large amount of satellite data. Among them are satellite images, which capture a wide range of ground conditions and are used in various fields, such as the extraction of building areas. However, human analysis of these images is time-consuming and inefficient, so deep learning is expected to be used for image analysis. The images that can be used to recognize objects in satellite images are high-resolution images taken by high-performance satellites, but such images are expensive, and the cost of acquiring a large number of images is high. In this paper, we propose a method to reconstruct a high-resolution satellite image by super-resolution from a low-resolution satellite image using the Vision Transformer model. Image Restoration Using Transformer (SwinIR) is adopted as the base model. To improve the accuracy, Knowledge Distillation and Multi-Layer Perceptron with Gating Unit are incorporated into the base model, and good results are obtained.

DOI： 10.1109/SCISISIS61014.2024.10759940

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85214686521&origin=inward

Registration of Spinal Region on CT Images Based on SuperPoint with Convolutional Block Attention Module 査読有り国際誌

Nabeshima K., Terasawa T., Kamiya T., Aoki T.

2024 Joint 13th International Conference on Soft Computing and Intelligent Systems and 25th International Symposium on Advanced Intelligent Systems, SCIS and ISIS 2024 2024年01月

担当区分：責任著者記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

Cancer is currently the leading cause of death in Japan, and the number of cancer patients is increasing every year. The number of patients with bone metastases from cancer is also increasing. The spine is a common site of bone metastasis, and cancer metastasis to the spine can cause serious symptoms such as hypercalcaemia and nerve paralysis due to spinal cord compression. As a result, it can affect patients' quality of life. Early detection and treatment of bone metastases is therefore important. To analyse the image information, CAD (Computer Aided Diagnosis) is introduced in image registration, which is obtained by temporal subtraction technique. One of them is CT (Computed Tomography) image-based registration technique by taken at different times of the same subject and performs subtraction to enhance changes in the lesion over time. In this approach, the registration of the same slice is important for the generation of temporal subtraction image. In this paper, we propose a method to register both images obtained different time series based on SuperPoint algorithm incorporated CBAM (Convolutional Block Attention Module) as a method to select the same slice and register the spinal region on the CT image.

DOI： 10.1109/SCISISIS61014.2024.10759972

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85214674575&origin=inward

Registration Method for Head CT and MR Images Based on TransMorph Incorporated ASPP 査読有り国際誌

Murakami T., Yamamoto A., Kamiya T.

2024 Joint 13th International Conference on Soft Computing and Intelligent Systems and 25th International Symposium on Advanced Intelligent Systems, SCIS and ISIS 2024 2024年01月

担当区分：責任著者記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

In recent years, with the development of diagnostic imaging equipment, image fusion, which fuses medical images from multiple modalities, has been proposed as a diagnostic imaging application. This has the advantage of integrating images with different characteristics and resolutions into a single composite image, which compensates for the missing information of each and facilitates the reading of lesions. Examples of the use of fusion images include surgical simulation and CyberKnife-based surgery. To create a fusion image, the two images must be aligned. Currently, this is often done manually, which increases the burden on the surgeon and introduces problems such as variations in accuracy depending on the surgeon's skill. Therefore, this paper describes the development of an automatic registration method for creating a fusion image between head CT and MRI, which is necessary for CyberKnife-based surgery of brain tumors. The method uses TransMorph, a registration model, as the base model, and a new model is constructed by adding ECAM and ASPP. As a result, the proposed method with the introduction of ASPP to the base model shows an improvement in the mutual information value and succeeds in improving the accuracy.

DOI： 10.1109/SCISISIS61014.2024.10760204

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85214692089&origin=inward

Detection of Temporal Changes from Satellite Images Based on Improved SNUNet-CD 査読有り国際誌

Watanabe Y., Kamiya T.

2024 Joint 13th International Conference on Soft Computing and Intelligent Systems and 25th International Symposium on Advanced Intelligent Systems, SCIS and ISIS 2024 2024年01月

担当区分：責任著者記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

Change Detection (CD) technology is an integral part of Earth observation in remote sensing (RS) image analysis. It is widely used in areas such as natural disasters, urban development monitoring, land resource management, and surveillance. As a pixel-to-pixel prediction task, the CD technique is location sensitive and requires more attention to feature extraction. Recently, deep learning (DL)-based CD methods have made various efforts to extract deep semantic features, but they ignore the importance of shallow layer information, including high-resolution and fine-grained features, which leads to uncertainty in the change-ground entities. To address these problems, SNUNet-CD is a model that mitigates the loss of localization information in the deeper layers and improves detection accuracy by using dense connections. Although it shows higher detection accuracy than other methods and provides a good trade-off between accuracy and computational complexity, it has difficulty in exploiting the different semantic and detailed information contained in deep and shallow features, which often blurs the target boundaries of the identified changes. Therefore, in this paper, we propose a model that incorporates Attention Gates, an attention mechanism that selectively highlights and updates salient regions of the feature map, into the skip-connection part to effectively link features of different levels. The performance evaluation results showed that Precision=0.9073, Recall=0.9127, and Fl-score=0.9099, confirming the effectiveness of introducing Attention Gates.

DOI： 10.1109/SCISISIS61014.2024.10760239

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85214693174&origin=inward

A Denoising Method from Low-dose CT Images Based on Improved U-Net with Attention Mechanism 査読有り国際誌

Sawada S., Li G., Murakami S., Kamiya T.

2024 Joint 13th International Conference on Soft Computing and Intelligent Systems and 25th International Symposium on Advanced Intelligent Systems, SCIS and ISIS 2024 2024年01月

担当区分：最終著者,　責任著者記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

Japan has the highest number of CT scanners per million population among developed countries, and the annual medical radiation dose is about four times the world average. Screening of high-risk populations with low-dose CT has been shown to reduce lung cancer mortality. However, in order to obtain image quality comparable to that of normal-dose CT, it is necessary to accurately estimate the image noise of low-dose CT. In this paper, we apply a noise reduction method using convolutional neural networks: the CNN model is based on U-Net, a fully convolutional network with encoder-decoder structure, with two convolutional layer modifications and three attention mechanisms. Experimental results using whole body slice CT images of piglets show that the proposed method is useful by comparing with CT images of normal dose and by evaluating image quality using peak signal-to-noise ratio (PSNR).

DOI： 10.1109/SCISISIS61014.2024.10759999

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85214672500&origin=inward

呼吸音から生成した複数画像による呼吸器疾患の自動分類

田端, 陸, 神谷, 間普, 木戸

電子情報通信学会技法IE2023-32 2023年12月

4次元データからの口唇領域の抽出に基づいた顔の対称性解析

木原、木村、大河内、李(天津工大)、中村、神谷

信学技法 ( 信学技法 ) MI2023-18 1 - 4 2023年11月

　前のページ - 次のページ　

<神谷 亨>

論文 - 神谷 亨

<神谷　亨>

論文 - 神谷　亨