In the study transfer learning was employed to adapt the previously developed deep networks, 1D_BNCNN and 2D_BNCNN, to handle elliptical phantoms in DOI. The network was fine-tuned using the newly acquired elliptical phantom dataset by leveraging the knowledge and pre-trained weights obtained from the circular phantom dataset. This approach can potentially enhance the realism and accuracy of DOT imaging, enabling more precise characterization of biological tissues and structures.
This paper aims to demonstrate a novel deep-learning network that addresses the prediction of breast tumors for diffuse optical imaging. Two learning schemes, signal encoder and image encoder, in the proposed network are designed for reconstructing optical-property images. The former processing method takes boundary data directly to deep networks, and predicts the optical-coefficient distribution, while the latter feeds images obtained by inverse image reconstruction with artifacts and sometimes hard-to-localized tumors. All 10,000 samples of synthesized homogeneous and heterogeneous phantoms were randomly selected for training, validation, and testing of performance. Twelve phantom samples were employed to justify its effectiveness in real applications.
To handle the existing problems of a wearable inertial-sensors-based gait analysis system, the study aims to propose and implement computation schemes for correcting misaligned coordinates assigned to inertial sensors, and compensating inclined mounting on the foot and lower limbs, which results in incorrect computed gait parameters and errors in the obtained ankle flexion. A rotating device and designated joint movement were employed to justify the proposed correcting and compensating schemes.
PurposeVarious laboratory sources have recently achieved progress in implementing deep learning models on biomedical optical imaging of soft biological tissues. The highly scattered nature of tissues at specific optical wavelengths results in poor spatial resolution. This opens up opportunities for diffuse optical imaging to improve the spatial resolution of obtained optical properties suffering from artifacts. This study aims to investigate a dual-encoder deep learning model for successfully detecting tumors in different phantoms w.r.t tumor size on diffuse optical imaging.ApproachOur proposed dual-encoder network extends U-net by adding a parallel branch of signal data to get information directly from the base source. This allows the trained network to localize the inclusions without degrading or merging with the background. The signals from the forward model and the images from the inverse problem are combined in a single decoder, filling the gap between existing direct processing and post-processing.ResultsAbsorption and reduced scattering coefficients are well reconstructed in both simulation and phantom test datasets. The proposed and implemented dual-encoder networks characterize better optical-property images than the signal-encoder and image-encoder networks, and the contrast-and-size detail resolution of the dual-encoder networks outperforms the other two approaches. From the measures of performance evaluation, the structural similarity and peak signal-to-noise ratio of the reconstructed images obtained by the dual-encoder networks remain the highest values.ConclusionsIn this study, we synthesized the advantages of boundary data direct reconstruction, namely the extracted signals and iterative methods, from the obtained images into a unified network architecture.
SignificanceThe machine learning (ML) approach plays a critical role in assessing biomedical imaging processes especially optical imaging (OI) including segmentation, classification, and reconstruction, intending to achieve higher accuracy efficiently.AimThis research aims to develop an end-to-end deep learning framework for diffuse optical imaging (DOI) with multiple datasets to detect breast cancer and reconstruct its optical properties in the early stages.ApproachThe proposed Periodic-net is a nondestructive deep learning (DL) algorithm for the reconstruction and evaluation of inhomogeneities in an inverse model with high accuracy, while boundary measurements are calculated by solving a forward problem with sources/detectors arranged uniformly around a circular domain in various combinations, including 16 × 15, 20 × 19, and 36 × 35 boundary measurement setups.ResultsThe results of image reconstruction on numerical and phantom datasets demonstrate that the proposed network provides higher-quality images with a greater amount of small details, superior immunity to noise, and sharper edges with a reduction in image artifacts than other state-of-the-art competitors.ConclusionsThe network is highly effective at the simultaneous reconstruction of optical properties, i.e., absorption and reduced scattering coefficients, by optimizing the imaging time without degrading inclusions localization and image quality.
We proposed and implemented a deep learning scheme using convolution neural networks (CNNs) with batch normalization (BNCNN) to construct a sensor-image DOI computation model with the aim of reconstructing tissue optical-property images as well as identifying and localizing breast tumors. A non-iterative learning reconstruction method was developed to recover optical properties, focusing on one-dimensional convolution layers followed by dense layers. Besides simulated data for model training, validation and testing, for the comparison of model performance, measurement data sets were employed to test on the same trained network which results outperform Tikhonov regularization method and other artificial neural networks as well.
We implemented a ring-scanning mechanism in a prone diffuse optical imaging (DOI) system for the application of breast tumor detection before. In the study, simultaneous multiple-sinusoids driving light source and flexible opto-measurement channels were considered to update the DOI system. The designed flexible channels in the scanning module of imaging system prevent optical information loss measured on a noncircular phantom/object if using fixed optical channels. Further, simultaneous multiple-sinusoids driving light sources speed up the acquisition of optical information for the frequency domain DOI. Examination phantoms were designated to justify the proposed measurement schemes.
We implemented a ring-scanning mechanism in a prostrate type for breast tumor detection. Reconstructed μa and μs′
images of multi-layers scanning are presented in good outcomes, showing it’s promising for the 3D scanning of breast.
Diffuse optical tomography (DOT) is showing promise for breast tumor detection by
estimating optical property coefficients of breast tissue. In our previous study, we have
successfully reconstructed the synthetic data into three-dimensional (3-D) images with a
cylindrical model. Thus, the work within this study develops a 3-D image reconstruction algorithm
of DOT with a breast-like model for screening breast tumor. Reconstruction results show that the
quality of reconstructed images can be effective for tumor screening.
We propose and implement three-dimensional (3-D) ring-scanning equipment for near-infrared (NIR) diffuse optical imaging to screen breast tumors under prostrating examination. This equipment has the function of the radial, circular, and vertical motion without compression of breast tissue, thereby achieving 3-D scanning; furthermore, a flexible combination of illumination and detection can be configured for the required resolution. Especially, a rotation-sliding-and-moving mechanism was designed for the guidance of source- and detection-channel motion. Prior to machining and construction of the system, a synthesized image reconstruction was simulated to show the feasibility of this 3-D NIR ring-scanning equipment; finally, this equipment is verified by performing phantom experiments. Rather than the fixed configuration, this addressed screening/diagnosing equipment has the flexibilities of optical-channel expansion for spatial resolution and the dimensional freedom for scanning in reconstructing optical-property images.
In this presentation, we demonstrate a working prototype of an optical breast imaging system using parallel-paddle architecture with dual-direction scanning, of which the designed module can be incorporated with a mammographic system for the acquisition of optical transmission and reflection information in both directions of up-down and down-up. Additionally, the scanning module enables to move with a designated pitch to accommodate varied breast size for acquiring adequate data to reconstruct the images. Currently, continuous-wave near infrared illumination modules are used for experimentation. The feasibility will be presented by phantom test.
Diffuse optical imaging (DOI) providing functional information of tissues has drawn great attention for the last two decades. Near infrared (NIR) DOI systems composed of scanning bench, opt-electrical measurement module, system control, and data processing and image reconstruction schemes are developed for the screening and diagnosis of breast tumors. Mostly, the scanning bench belonging to fixed source-and-detector configuration limits computed image resolution to an extent. To cope with the issue, we propose, design and implement a 3D prostrate ring-scanning equipment for NIR DOI with flexible combinations of illumination and detection, and with the function of radial, circular and vertical movement without hard compression of breast tissue like the imaging system using or incorporating with X-ray mammographic bench. Especially, a rotation-sliding-and-moving mechanism was designed for the guidance of source- and detection-channel movement. Following the previous justification for synthesized image reconstruction, in the paper the validation using varied phantoms is further conducted and 3D image reconstruction for their absorption and scattering coefficients is illustrated through the computation of our in-house coded schemes. The source and detection NIR data are acquired to reconstruct the 3D images through the operation of scanning bench in the movement of vertical, radial and circular directions. Rather than the fixed configuration, the addressed screening/diagnosing equipment has the flexibility for optical-channel expansion with a compromise among construction cost, operation time, and spatial resolution of reconstructed μa and μs’ images.
KEYWORDS: Modulation, Fourier transforms, Bandpass filters, Signal analysis, Signal processing, Signal analyzers, Linear filtering, Diagnostics, Biological research, Amplitude modulation
The traditional envelope analysis is an effective method for the fault detection of rolling bearings. However, all the resonant frequency bands must be examined during the bearing-fault detection process. To handle the above deficiency, this paper proposes using the empirical mode decomposition (EMD) to select a proper intrinsic mode function (IMF) for the subsequent detection tools; here both envelope analysis and cepstrum analysis are employed and compared. By virtue of the band-pass filtering nature of EMD, the resonant frequency bands of structure to be measured are captured in the IMFs. As impulses arising from rolling elements striking bearing faults modulate with structure resonance, proper IMFs potentially enable to characterize fault signatures. In the study, faulty ball bearings are used to justify the proposed method, and comparisons with the traditional envelope analysis are made. Post the use of IMFs highlighting faultybearing features, the performance of using envelope analysis and cepstrum analysis to single out bearing faults is objectively compared and addressed; it is noted that generally envelope analysis offers better performance.
KEYWORDS: Digital signal processing, Control systems, Process control, Fourier transforms, Transducers, Signal processing, Matrices, Dynamical systems, Adaptive control, Krypton
The phenomenon of fluid-induced instability existing in fluid-film bearing systems has been coped with for long time. The study aims to soothe and even eliminate the occurrence of whirl in rotary machinery by increasing the threshold of instability through the anti-swirl injection using an optimal control based linear quadratic regulator. An acceptance region was established in order to decide starting up the control process. Some case studies were carried out to illustrate the effectiveness of the control scheme. The research results present that a simple control method incorporating with an acceptance region enables to avoid the fluid induced instability flexibly in rotary machinery. Moreover, the developed techniques can also be applied in other fluid-induced instability problems such as whip and rub, etc.
The design scheme of the source-and-detector arrangement of a ring-scanning-based near-infrared optical imaging system prior to the mechanical and optical construction is demonstrated. In terms of the effectiveness and efficiency of design, through the computation of image reconstruction for varied imaging configurations, the influences of the source-and-detector arrangement on the resulting images are evaluated and a formula to estimate the scanning time is provided. The basic idea of our design is to divide circular scanning into several zones, each of which includes n sources and l detectors; i.e., m zones and n sources along with l detectors per zone are defined in the design. Comparison is made among different imaging configurations where their contrast-to-noise ratio measures are evaluated and contrast-and-size detail resolution curves are depicted. Results show that the 2Z3S or 3Z3S configuration is the optimal design in terms of the time consumption of a complete scanning and the resolution of reconstructed optical-property images.
The study aims at developing an optical measurement module incorporated with an X-ray mammographic
system to obtain diffuse optical images (DOI) for the detection of breast tumors. Two goals steer the study: (1) to
enhance sensitivity and specificity of tumor detection through the use of functional DOI; and (2) to reduce radiation
exposure by using only one mammogram, instead of two, as structure information to compute optical-coefficient
images. A dual-direction (downward and upward) scanning device to project illuminated near infrared light with
multiple-channel switching for both sources and detectors was designed and constructed to obtain double information.
The designed and constructed NIR scanning module incorporates with GE Senographe 2000D to assist breast tumor
detection.
A promising method to achieve rapid convergence for image reconstruction is introduced for the continuous-wave near-infrared (NIR) diffuse optical tomography (DOT). Tomographic techniques are usually implemented off line and are time consuming to realize image reconstruction, especially for NIR DOT. Therefore, it is essential to both speed up reconstruction and achieve stable and convergent solutions. We propose an approach using a constraint based on a Lorentzian distributed function incorporated into Tikhonov regularization, thereby rapidly converging a stable solution. It is found in the study that using the proposed method with around five or six iterations leads to a stable solution. The result is compared to the primary method usually converging in ~25 iterations. Our algorithm rapidly converges to stable solution in the case of noisy (>20 dB) detected intensities.
A promising method achieving rapid convergence for image reconstruction is introduced for the
continuous-wave NIR-DOT. An approach employs a constraint based on Lorentzian distributed
function incorporated into Tikhonov regularization, thereby rapidly converging a stable solution.
Using extracted spectral features is proposed to reconstruct video-rate optical-properties images. Compared
with reconstruction through time-sequence data, the results through spectral features are exempt from noise
affection, and are able to differentiate hemodynamic conditions in a single heart-beat cycle.
For various size, location and contrast of imitated tumors, both numerical computation and experimental
validation were conducted to investigate and conclude diagnosis limitation of an NIR-DOI system.
We attempt to develop a systematic scheme through adopting high-pass filtering (HPF) to well resolve value-preserved images such as medical images. Our approach is derived from the Poisson maximum a posteriori superresolution algorithm employing the HP filters, where four filters are considered such as two low-pass-filter-combination based filters, wavelet filter, and negative-oriented Laplacian HP filter. The proposed approach is incorporated into the procedure of finite-element-method (FEM)-based image reconstruction for diffuse optical tomography in the direct current domain, posterior to each iteration without altering the original FEM modeling. This approach is justified with various HPF for different cases that breast-like phantoms embedded with two or three inclusions that imitate tumors are employed to examine the resolution performances under certain extreme conditions. The proposed approach to enhancing image resolution is evaluated for all tested cases. A qualitative investigation of reconstruction performance for each case is presented. Following this, we define a set of measures on the quantitative evaluation for a range of resolutions including separation, size, contrast, and location, thereby providing a comparable evaluation to the visual quality. The most satisfactory result is obtained by using the wavelet HP filter, and it successfully justifies our proposed scheme.
Near-infrared diffuse optical tomography (NIR DOT) for noninvasive tissue monitoring have been developed for nearly
two decades. The NIR imaging, however, suffers from low resolution due to the diffusive nature of the scattered light;
there are compelling reasons for merging high-resolution structural information from other imaging modalities with the
functional information attainable with NIR DOT. In this article, slight variation of the inclusion (tumor) in low contrast
of optical properties is estimated and investigated. We present that an initial study of using a structural a prioriknowledge in NIR tomography where absorption image reconstruction of the tested phantom is well defined with the aid
of a structural a priori knowledge obtained from other imaging modalities. This is advantageous compared to either
modality alone. As well, the reconstructed optical absorption coefficient is achieved more accurate near to be exact
value with incorporating the empirical updating information being proportional to the off-boundary distance but not size
of inclusion against the background. Numerical simulation is demonstrated on varied sizes, locations and contrast of the
inclusion. With the comparison between with or without a priori and empirical updating information, it is found that the
reconstructed optical properties are more accurate than the near-infrared imaging alone.
Diffuse optical tomography (DOT) is in an attempt to image the interior of human tissues. However, the NIR imaging
suffers from low resolution due to the diffusive nature of the scattered light, which results into poor reconstructed image
quality. Thus, the effort to improve the image quality remains in progress. The numerical simulation using high-pass
filtering incorporated into the finite-element-based diffusion equation to reconstruct tomographic images of optical
properties was performed where results reveal that several inclusions (tumors) can be well defined separately, thereby
demonstrating the ability to highly resolve the image of interest with the optimal high-pass filtering process.
The study aims at developing a near infrared (NIR) tomography imaging system using a single rotating source/detector scanning device, which will be working for diffuse optical tomography (DOT) on medical applications. Some influential factors in terms of the design of this scanning device and test phantoms are investigated such as the temporal stability, air-absorption, container influence, and radiance normalization, etc. Then, a heterogeneous microsphere phantom with an off-center inclusion is investigated. It is observed that the radiance deviation between the heterogeneous and the homogenous is shifting with the inclusion position accordingly. Through the previously-mentioned system calibration, the back projection method as widely applied in the computed tomography (CT) technique is used to reconstruct optical images which indicate the distribution of optical property of the test phantom.
In the research, the pseudo-model technique is proposed and implemented to simulate biological tissues under the framework of investigating near infrared (NIR) light propagating in diffusive media. The same optical characteristics inhere in the corresponding pseudo-models as in real tissues, where pseudo-models are constructed by using various volume densities of Intralipid. The pseudo-model technique proposed in the study has following advantages:
1.For the NIR tomography imaging system, the output signal from the real tissue may be too weak to be detected beyond the ability of current technologies. Thus, the pseudo-model is a viable alternative to cope with the limitation of the system in the measurement of real tissues.
2.Once the pseudo-model of a real tissue is decided, its optical properties can be investigated thoroughly. In addition, the initial estimates for the reconstruction of NIR optical property images can be selected adequately, and the image reconstruction algorithm can be modified accordingly with the information acquired from the pseudo-model.
In the experiment, a pseudo-model of the background with an inclusion is performed for real tissues of pork inserted by a bone. It is observed that 1 % v.d. and 3 % v.d. of Intralipids can replace the pork and the bone, respectively, and the characteristics of the pseudo-model proposed here are consistent with those of the real tissues. As part of conclusion, the use of the pseudo-model technique is a promising approach to mimicking real tissues, especially for some parts of human body unable to be detected effectively.
Diffuse optical tomography (DOT) using diffuse light, red or near-infrared (NIR) light, is in an attempt to image the interior of human tissues such as breasts, arms, etc. In the current design of our NIR tomography imaging system, the system uses a single rotating source/detector scanning device associated with an image reconstruction scheme. The device can dramatically save source- and detection-fiber-bundles, and offer promising measured radiance reflecting the optical properties of the test phantoms. Both source and detector can rotate on command with any pre-defined angles controlled by the computer. Additionally, an image reconstruction algorithm applied to the tomography scanning device in the DC domain is also implemented. The ability and performance of this image reconstruction algorithm are discussed and presented. Results reveal that inclusion (tumor) positions can be well defined and the spatial resolution is beyond 1:16, inclusion to background.
The research aims at developing an NIR tomography system using a single rotating source/detector scanning device associated with an image reconstruction scheme. Several simulation results concerning the phantom with one, two or three targets are presented. Additionally, the developed system is validated by a hemoglobin phantom inserted by another different volume density one.
This study conducts an investigation on flaw cogged V-belts, galling roller-chains, and imbalancing rotors through a constructed transmission-component test bench. Nine channels of noise and vibration data are acquired and processed to extract features that exhibit the faulty condition of components in specific states. Two artificial neural network schemes, i.e., the backward propagation and self-organization mapping algorithms, are employed as pattern recognition tools. Additionally, the classification of condition patterns of machine components is further illustrated using a discrimination-space technique. Thus, the mechanism of pattern recognition of artificial neural networks can be clearly realized, but not only considered as an inaccessible processing black box.
The study proposes an improved Gabor order tracking (GOT) technique to cope with crossing orders that cannot be effectively separated using the original GOT scheme. The improvement aids both the reconstruction and interpretation of two crossing orders such as a transmission-element-regarding order component and a structural resonant component. In the paper, the influence of the dual function to Gabor expansion coefficients is investigated, which can affect the precision of the tracked order component. Additionally, using the GOT scheme in noise conditions is demonstrated as well. For applying the improved GOT in real tasks, separation and extraction of close-order components of vibration signals measured from a transmission-element test bench is illustrated using both the GOT and Vold-Kalman filtering (VKF) OT schemes. Finally, comprehensive comparisons between the improved GOT and VKF_OT schemes are made from processing results.
The study aims at implementing a remote online machine fault diagnostic system built up in the architecture of both the BCB software-developing environment and Internet transmission communication. Variant signal-processing computation schemes for signal analysis and pattern recognition purposes are implemented in the BCB graphical user interface. Hence, machine fault diagnostic capability can be extended by using the socket application program interface as the TCP/IP protocol. In the study, the effectiveness of the developed remote diagnostic system is validated by monitoring a transmission-element test rig. A complete monitoring cycle includes data acquisition, signal processing, feature extraction, pattern recognition through the ANNs, and online video monitoring, is demonstrated.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.