Search

Human-Object Interaction (HOI) recognition, due to its significance in many computer vision-based applications, requires in-depth and meaningful details from image sequences. Incorporating semantics in scene understanding has led to a deep understanding of human-centric actions. Therefore, in this research work, we propose a semantic HOI recognition system based on multi-vision sensors. In the proposed system, the de-noised RGB and depth images, via Bilateral Filtering (BLF), are segmented into multiple clusters using a Simple Linear Iterative Clustering (SLIC) algorithm. The skeleton is then extracted from segmented RGB and depth images via Euclidean Distance Transform (EDT). Human joints, extracted from the skeleton, provide the annotations for accurate pixel-level labeling. An elliptical human model is then generated via a Gaussian Mixture Model (GMM). A Conditional Random Field (CRF) model is trained to allocate a specific label to each pixel of different human body parts and an interaction object. Two semantic feature types that are extracted from each labeled body part of the human and labelled objects are: Fiducial points and 3D point cloud. Features descriptors are quantized using Fisher’s Linear Discriminant Analysis (FLDA) and classified using K-ary Tree Hashing (KATH). In experimentation phase the recognition accuracy achieved with the Sports dataset is 92.88%, with the Sun Yat-Sen University (SYSU) 3D HOI dataset is 93.5% and with the Nanyang Technological University (NTU) RGB+D dataset it is 94.16%. The proposed system is validated via extensive experimentation and should be applicable to many computer-vision based applications such as healthcare monitoring, security systems and assisted living etc.

Electrocardiogram (ECG)-based biometrics are gaining popularity because of its robustness against falsification. In this letter, we have designed a new long short-term memory (LSTM) based framework for person identification using ECG signal. Our model learns the underlying temporal representation of an ECG signal by considering both intra-beat and inter-beat variations. We have shown that the LSTM model captures the intra-beat variations better for smaller ECG segments. Unlike most of the models in the literature, our model doesn't require the detection of any fiducial points. This is achieved by training the LSTM network with smaller segments of ECG signals, which are extracted by sliding a rectangular window. The effect of different window length on person identification accuracy is studied. The efficiency of the model is extensively tested on four databases; PTB, MIT-BIH arrhythmia database, ECG-ID, and CYBHi. An accuracy of 97.3% is obtained for 290 subjects of the PTB database. Similar results are also obtained for other databases, as well as 79.37% accuracy for CYBHi.

Compared with traditional learning criteria, such as minimum mean square error (MMSE), the minimum error entropy (MEE) criterion has received increasing attention in the domains of nonlinear and non-Gaussian signal processing and machine learning. Since the MEE criterion is shift-invariant, one has to add a bias to achieve zero-mean error over training datasets. Thus, a modification of the MEE called minimization of error entropy with fiducial points (MEEF) was proposed, which controls the bias for MEE in a more elegant and efficient way. In the present paper, we propose a fixed-point minimization of error entropy with fiducial points (MEEF-FP) as an alternative to the gradient based MEEF for training a linear-in-parameters (LIP) model because of its fast convergence speed, robustness and step-size free. Also, we provide a sufficient condition that guarantees the convergence of the MEEF-FP algorithm. Moreover, we develop a recursive MEEF-FP (RMEEF-FP) for online adaptive learning with low-complexity. Finally, illustrative examples are presented to show the excellent performance of the new methods.

In the lung biopsy image-guided surgery systems, the fiducial markers are used for point-based registration of the patient space to the CT image space. Fiducial marker detection and fiducial point localization in CT images have great influence on the accuracy of registration and guidance. This study proposes a fiducial marker detection approach based on the features of marker image slice sequences and a fiducial point localization approach according to marker projection images, without depending on the priori-knowledge of the marker default parameters provided by the manufacturers. The accuracy of our method was validated based on a CT image dataset of 24 patients. The experimental results showed that all 144 markers of 24 patients were correctly detected, and the fiducial points were localized with the average error of 0.35 mm. In addition, the localization accuracy of the proposed method was improved by an average of 12.5% compared with the accuracy of the previous method using the marker default parameters provided by the manufacturers. Thus, the study demonstrated that the proposed detection and localization methods are accurate and robust, which is quite encouraging to meet the requirement of future clinical applications in the image guided lung biopsy and surgery systems.

The characteristics or aspects of important fiducial points (FPs) in the electrocardiogram (ECG) signal are complicated because of various factors, such as non-stationary effects and low signal-to-noise ratio. Due to the various noises caused by the ECG signal measurement environment and by typical ECG signal deformation due to heart diseases, detecting such FPs becomes a challenging task. In this study, we introduce a novel PQRST complex detector using a one-dimensional bilateral filter (1DBF) and the temporal characteristics of FPs. The 1DBF with noise suppression and edge preservation preserves the P- or T-wave whereas it suppresses the QRS-interval. The 1DBF acts as a background predictor for predicting the background corresponding to the P- and T-waves and the remaining flat interval excluding the QRS-interval. The R-peak and QRS-interval are founded by the difference of the original ECG signal and the predicted background signal. Then, the Q- and S-points and the FPs related to the P- and T-wave are sequentially detected using the determined searching range and detection order based on the detected R-peak. The detection performance of the proposed method is analyzed through the MIT-BIH database (MIT-DB) and the QT database (QT-DB).

According to World Health Organisation (WHO), a large population of this age is affected by cardiovascular illnesses. Detection of cardiac arrhythmias at an earlier stage plays an important role in cardiac assessment and treatment. Continuous and accurate measurement of the pulse signal becomes essential as it provides information about the regularity of the heartbeat and pulse strength. A pulse recording system which is low powered, easy to access in the home environment and portable will serve the purpose. This paper describes the design and implementation of a wrist pulse acquisition system. System validation results and detailed analysis of the heart rate and average period of a cardiac cycle calculated using different fiducial points are also presented.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

The process of age estimation is a task of computer vision that represents a challenge to the diverse factors that influence the process of aging in humans such as environment, food, and toxic habits, among others. Such a task has been studied from different approaches. In this work, a geometric approach is developed by using distances and proportions existing in every face. The functionality is increased by the use of fiducial points to locate any characteristic features of the face; its location is based on craniofacial growth studies used in the area of forensic anthropometry and plastic surgery. The location of 15 fiducial points is considered through the ASM algorithm and 8 distances between fiducial points, which statistically verify the difference in facial proportions between adults and minors. The distances obtained are analyzed with the Discriminant Analysis procedure to study the correlations between distances and the corresponding age group. From this, a model that allows classifying a new image in an age group is constructed: adults; those over 18 years of age, and minors; those under 17 years of age, obtaining 90% accuracy.

Surgical navigation system (SNS) has been an important tool in surgery. However, the complicated and tedious manual selection of fiducial points on preoperative images for registration affects operational efficiency to large extent. In this study, an oral and maxillofacial navigation system named BeiDou-SNS with automatic identification of fiducial points was developed and demonstrated. To solve the fiducial selection problem, a novel method of automatic localization for titanium screw markers in preoperative images is proposed on the basis of a sequence of two local mean-shift segmentation including removal of metal artifacts. The operation of the BeiDou-SNS consists of the following key steps: The selection of fiducial points, the calibration of surgical instruments, and the registration of patient space and image space. Eight cases of patients with titanium screws as fiducial markers were carried out to analyze the accuracy of the automatic fiducial point localization algorithm. Finally, a complete phantom experiment of zygomatic implant placement surgery was performed to evaluate the whole performance of BeiDou-SNS. The coverage of Euclidean distances between fiducial marker positions selected automatically and those selected manually by an experienced dentist for all eight cases ranged from 0.373 to 0.847 mm. Four implants were inserted into the 3D-printed model under the guide of BeiDou-SNS. And the maximal deviations between the actual and planned implant were 1.328 mm and 2.326 mm, respectively, for the entry and end point while the angular deviation ranged from 1.094° to 2.395°. The results demonstrate that the oral surgical navigation system with automatic identification of fiducial points can meet the requirements of the clinical surgeries.

Background:The estimation of fiducial points is specially important in the analysis and automatic diagnose of Electrocardiographic (ECG) signals.Objective:A new algorithm which could be easily implemented is presented to accomplish this task.Methods:Its methodology is rather simple, and starts from some ideas available in the literature combined with new approachs provided by the authors. First, aQRScomplex detection algorithm is presented based on the computation of energy maxima in ECG signals which allow the measurement of cardiac frequency (in beats per minute) and the estimation of R peaks temporal positions (in number of samples). From these ones, an estimation of fiducial points Q, S, J, P and T waves onset and offset points are worked out, supported in a simple modified slope method with constraints.The location process of fiducial points is assisted with the help of the so called curvature filters, which allow to improve the accuracy in this task.Results:The procedure is simulated in Matlab and GNU Octave by using test signals from the MIT medical database, Cardiosim II equipment patterns and synthetic signals developed by the authors.Conclusion:One of the novelties of this work is the global strategy. Also, another significant innovation is the introduction of the curvature filters. We think this concept will prove to be a useful tool in signal processing, not only in ECG analysis.

The basis and reliability for timely diagnosis of cardiovascular diseases depend on the robust and accurate detection of QRS complexes along with the fiducial points in the electrocardiogram (ECG) signal. Despite, the several QRS detection algorithms reported in the literature, the development of an efficient QRS detector remains a challenge in the clinical environment. Therefore, this article summarizes the performance analysis of various QRS detection techniques depending upon three assessment factors which include robustness to noise, computational load, and sensitivity validated on the benchmark MIT-BIH arrhythmia database. Moreover, the limitations of these algorithms are discussed and compared with the standard signal processing algorithms, followed by the future suggestions to develop a reliable and efficient QRS methodology. Further, the suggested method can be implemented on suitable hardware platforms to develop smart health monitoring systems for continuous and long-term ECG assessment for real-time applications.

PURPOSE Permanent breast seed implantation (PBSI) is a promising radiotherapy technique for early-stage breast cancer, completed in a single visit by permanently implanting 103 Pd seeds using needles inserted through a template and guided by two-dimensional (2D) ultrasound (US). However, operator dependence has been highlighted as a limitation of this procedure. Consequently, we propose and have developed an intraoperative guidance system using three-dimensional (3D) US and an instrumented mechanical arm to provide intraoperative 3D imaging and needle template tracking. METHODS A mechatronic 3D US scanner reconstructs a 3D image from 150 2D images. A tracked mechanical arm mounted to the scanner locates four fiducial points on the template, registering the template to the 3D image. 3D reconstruction was validated for linear and volumetric measurement accuracy using phantoms of known geometry. In vivo breast US image quality was evaluated in a healthy volunteer. The encoded arm was calibrated and validated using a jig with divots at known locations relative to the scanner and the scanner registered to the 3D US image using intersecting strings in a fluid-filled test jig. Template registration accuracy was assessed using a machined test jig. Tracking accuracy was assessed in a liquid medium by comparing tracked and imaged needle tip positions. Finally, the system was used to guide a mock procedure in a patient-specific phantom and micro-CT imaging used to evaluate its accuracy. RESULTS Geometric validation showed median distances within ±1.1% of expected values and volumetric validation showed differences of ≤4.1%. Tracking arm point measurements showed an average error of 0.43 mm and 3D US volume registration showed target registration error ≤0.9 mm. Mean template registration accuracy in each axis of translation/rotation was ≤1.3 mm/1.0°. Mean needle-targeting error was 2.5 mm and 1.6° for needle tips and trajectories, respectively. Mean needle tip and angular errors of the phantom procedure were 2.1 mm and 2.6°. Modeled seed displacement of the phantom procedure showed mean error of 2.6 mm and a maximum of 3.8 mm. CONCLUSIONS A 3D US guidance system for PBSI has been developed. Benchtop performance and image quality in volunteer scans are satisfactory. A phantom PBSI procedure was successfully delivered using the system with maximum seed error within dosimetric benchmarks (

In this paper, we investigate the estimation of the maximum target registration error (TRE) magnitude of the target location while using point-based rigid registration in the image guided system. Under the uniform restriction of fiducial localization error (FLE) magnitude, we explicitly formulate the estimation as an optimization problem. Through analyzing the approximated problem which assumes the rigidity of the fiducial set holds with the perturbation of FLE, we present a strict lower bound for the maximum TRE magnitude. The simulations show that the lower bound is close to the actual maximum TRE magnitude for the target locations lying far away from the fiducial points. Unlike the expected TRE magnitude in which all fiducial points contribute, the lower bound is only related to the fiducial points serving as the vertices of the convex hull of the fiducial set. Our analysis provides a new perspective of investigating the problem of TRE estimation and is helpful for the surgeons to learn about the worst situation during using the image guided system.

Accurately detected human body fiducial points provide an easy and efficient method for human body posture analysis and the extraction of anthropometric parameters. In the proposed work, an efficient algorithm for automated and accurate detection of fiducial points is developed for both the frontal and the lateral images. An algorithm for automatic human body segmentation of the frontal image is also developed using automatically detected set of primary fiducial points. Additional fiducial points are obtained by applying peak and valley algorithm on the silhouettes of each segment. The detection accuracy of the automatically detected fiducial points is calculated by comparing their locations with the manually marked fiducial points. The proposed algorithm is tested on 45 subjects including both male and female genders and variable Body Mass Indexes. In most cases, the algorithm successfully detects seventy fiducial points for each subject in the testing set. A quantitative analysis of the error in the position of the detected fiducial points shows that the algorithm performs better than the state-of-the-art algorithms found in the existing literature. In the evaluation of the algorithm, the percentage accuracy of the detected fiducial points is calculated and it is observed that the proposed algorithm performs better for the majority of the fiducial points.

Accurate automatic identification of fiducial points within an ECG is required for the automatic interpretation of this signal. Several methods exist in the literature for automatic ECG segmentation. These algorithms are based on different methodologies and often evaluated with different datasets and protocols, which makes their performance challenging to compare. For this study, nine segmentation algorithms were selected from the literature and evaluated with the same protocol in order to study their performance. One hundred signals from the PhysioNet’s QT database were used for this evaluation. Results showed that one of the algorithms based in the discrete wavelet transform achieved sensitivity of 100% when detecting the onset and offset of the QRS complex. An algorithm using the Multi-scale Morphological Derivate achieved sensitivities of 99.81%, 98.17% and 99.56% when detecting the peak, onset and offset respectively of the P-wave. When segmenting the T-wave, an algorithm based on the Phasor transform had a good performance with sensitivities of 97.78%, 97.81% and 95.43% when detecting the peak, onset and offset, respectively. Additionally, probabilistic methods such as Hidden Markov Models had good results due to the fact that they can learn from real signals and adapt to specific conditions. However, these techniques are often computationally more complex and require training. This study could help in selecting optimal algorithms for ECG segmentation when implementing a system for automatic ECG interpretation.

Precise estimation of fiducial points of a seismocardiogram (SCG) signal is a challenging problem for its clinical usage. Delineation techniques proposed in the existing literature do not estimate all the clinically significant points of an SCG signal, simultaneously. The aim of this research work is to propose a delineation framework to identify IM, AO, IC, AC, pAC and MO fiducial points with the help of a PPG signal. The proposed delineation method processes a wavelet-based scalographic PPG and an envelope construction scheme is proposed to estimate the prominent peaks of the PPG signal. A set of amplitude histogram based decision rules is developed for estimation of SCG diastole phases, namely AC, pAC and MO. Subsequently, the systolic phases, IM, AO and IC are detected by applying diastole masking on SCG and decision rules. Experimental results on real-time SCG signals acquired from our designed data acquisition-circuitry and their analysis show the effectiveness of the proposed scheme. Additionally, these estimated parameters are analyzed to show the discrimination between normal breathing and breathlessness conditions., Comment: IEEE Transactions on Instrumentation and Measurement, 2020

Image landmark detection aims to automatically identify the locations of predefined fiducial points. Despite recent success in this field, higher-ordered structural modeling to capture implicit or explicit relationships among anatomical landmarks has not been adequately exploited. In this work, we present a new topology-adapting deep graph learning approach for accurate anatomical facial and medical (e.g., hand, pelvis) landmark detection. The proposed method constructs graph signals leveraging both local image features and global shape features. The adaptive graph topology naturally explores and lands on task-specific structures which are learned end-to-end with two Graph Convolutional Networks (GCNs). Extensive experiments are conducted on three public facial image datasets (WFLW, 300W, and COFW-68) as well as three real-world X-ray medical datasets (Cephalometric (public), Hand and Pelvis). Quantitative results comparing with the previous state-of-the-art approaches across all studied datasets indicating the superior performance in both robustness and accuracy. Qualitative visualizations of the learned graph topologies demonstrate a physically plausible connectivity laying behind the landmarks.

Facial expression recognition is a very challenging problem and has attracted more and more researchers’ attention. In this paper, considering that facial expression recognition is closely related to the features of key facial regions, we propose a facial expression recognition network that explicitly utilizes the landmark heatmap information to precisely capture the most discriminative features. In addition to directly adding the information of facial fiducial points in the form of landmark heatmaps, we also propose an end-to-end network structure--heatmap aiding emotion network (HAE-Net) by embedding the landmark detection module based on stack-based hourglass network into the facial expression recognition network. Experiments on CK+, RAF and AffectNet databases show that our method achieves better results compared with the state-of-the-art methods, which demonstrates that adding additional landmark information, as well as joint training of landmark detection and expression recognition, are beneficial to improve recognition performance.

Electrocardiography (ECG or EKG) is a medical test that is heavily used to assess human heart condition and investigate a large set of cardiac diseases. Automated ECG analysis has become a task of increased clinical importance since it can aid physicians in improved diagnosis. Most of the automated ECG analysis techniques require first identifying the onset and offset locations of its fiducial points and characteristic waves. Two of the important characteristic waves are P and T waves. They mark the beginning and end of an ECG cycle, respectively. In this paper, a fast technique is proposed that can segment ECG signals by accurately identifying the P and T waves. In this work, we evaluate the performance of our model on standard QT database (Laguna et al. Comput Cardiol 24:673–676, 1997 [1]). We achieved high accuracies above 99% and 97% while detecting P waves and T waves respectively.

Biometry is becoming increasingly important in order to identify or authenticate individuals. Since the seminar work of Biel et at. in 1999 and 2001, the feasibility of using the electrocardiogram (ECG) for biometric recognition has been considered by several authors. Both fiducial methods, which are based on using fiducial points related to the detected QRS complexes, and non-fiducial methods, which do not require the extraction of the QRS complexes from the signals, have been considered. However, the feasibility of ECG-based biometry is still unclear, as the results from different studies are difficult to compare. In this paper, we concentrate on fiducial methods, comparing the performance of several classifiers and dimensionality reduction techniques on a publicly available dataset. Our results show that ECG-based biometry is indeed a feasible alternative to other widely used biometric traits, since an accuracy above 99.95% can be attained with the appropriate choice of the dimensionality reduction method and classifier.

The paper is devoted to the creation of an algorithm for determining the parameters of the electrocardiosignal (ECS) QRS-complex. The ECS model is created, all elements of which have average values of parameters of real ECS. An algorithm is developed that determines the temporal position of the peaks of the QRS-complex, as well as its beginning and end. The work of the algorithm when applying white Gaussian noise on the ECS is investigated. According to the research results, an algorithm of noise smoothing was developed and applied. This algorithm is based on averaging the signal in a specially selected time interval.

Automatic detection of cardiac abnormalities in early stage is a popular area of research for decades. In this work a novel algorithm for detection of cardiac arrhythmia is proposed using variational mode decomposition (VMD). Arrhythmia is a crucial abnormality of heart in which the rhythmic disorder may lead to sudden cardiac arrest. Existing algorithms for arrhythmia detection are based on accuracy of detection of fiducial points, parameter selection and extraction, quality of classifier and other factors. Unlike other works, proposed method tries to characterize both atrial and ventricular arrhythmias simultaneously and independently from the segmented sections of the signal. VMD, being able to separate closely spaced frequencies, has a good potential to be useful to provide significant features in transformed domain. Unique feature combinations are also proposed to characterize different arrhythmic events.

International audience; Face landmarking, defined as the detection of fiducial points on faces, has received a lot of attention over the last two decades within the computer vision community. While research literature documents major advances using state-of-art deep convolutional neural networks, earlier cascaded regression tree-based approaches remain a relevant alternative for low-cost, low-power embedded systems. Yet, from a practical point of view, their implementation and parametrization can be a difficult and tedious process. In this paper, we provide the readers with insights and advice on how to design a successful face landmarking system using a cascade of regression trees.

With the development of AI, more and more deep learning methods are adopted on medical data for computer-aided diagnosis. In this paper, a 50-layer convolutional neural network (CNN) is trained for normal and abnormal short-duration electrocardiogram (ECG) classification. We do this using a forward neural network with one or more layers of quick connections. This network is deeper than previously used plain network, and it resolves the notorious problem of network degradation of training accuracy and can significantly increase depth to improve accuracy. Detecting fiducial points and combining features are not required, and the classification model can effectively replace the traditional predefined and time-wasting user’s manual selection features. The method was tested on over 150,000 recorded short-duration ECG clinical datasets and achieves 89.43% accuracy, the sensitivity was 87.73%, and the specificity was 91.63%. The experiments demonstrate that our method is efficient and powerful in clinical applications.

In this work, we have proposed enhancements that improve the performance of state-of-the-art facial emotion recognition (FER) systems. We believe that the changes in the positions of the fiducial points and the intensities capture the crucial information regarding the emotion of a face image. We propose the inputting of the gradient and the Laplacian of the input image together with the original into a convolutional neural network (CNN). These modifications help the network learn additional information from the gradient and Laplacian of the images. However, as shown by our results, the CNN in the existing state-of-the-art models is not able to extract this information from the raw images. In addition, we employ spatial transformer network to add robustness to the system against rotation and scaling. We have performed a number of experiments on two well known datasets, namely KDEF and FERplus. Our approach enhances the already high performance of the state-of-the-art FER systems by 3 to 5%.

The detection of fiducial points on faces has significantly been favored by the rapid progress in the field of machine learning, in particular in the convolution networks. However, the accuracy of most of the detectors strongly depends on an enormous amount of annotated data. In this work, we present a domain adaptation approach based on a two-step learning to detect fiducial points on human and animal faces. We evaluate our method on three different datasets composed of different animal faces (cats, dogs, and horses). The experiments show that our method performs better than state of the art and can use few annotated data to leverage the detection of landmarks reducing the demand for large volume of annotated data., https://ieeexplore.ieee.org/document/8451026/

Cardiovascular diseases (CVD) are the leading cause of death in the world, representing almost 31\% of global deaths. Therefore, continuous research on early diagnosis methods has been made to achieve more effective treatments of CVDs. Electro-cardiogram (ECG) stands as the most common technique used to monitor cardiac variations. In ECG, five key points (a.k.a. fiducial points) correspond to the wave peaks P, Q, R, S, and T, whereby anomalies in the relative positions of these fiducial points indicate a potential CVD. Thus, a real-time decision support system capable of detecting ECG fiducial points is of paramount importance to early treatment. This paper proposes a real-time method to identify fiducial points in ECGs leveraging for this moving averages and heuristics. The performance comparison between the proposed method applied to the MIT-BIH Arrhythmia Database over other relevant methods found in the literature has been carried out. Results show that the proposed method outperforms the other techniques concerning both accuracy and computational cost impact.

The 100-MeV proton linac and magnets for the KOMAC (Korea Multi-purpose Accelerator Complex) were installed in the tunnel and the beamlines. The fiducialization process was accomplished with the measurement of mechanical shape and the transfer of the coordinates to the fiducial points that are used in two laser-trackers based alignments. The reference points called the alignment network were set up on the wall inside tunnel. The linac and the beam transport magnets were aligned based on the survey results of the alignment networks. In this paper, the alignment procedure and the alignment results are presented, and an algorithm that was developed to manipulate the adjusters of the magnetsis introduced.

A smartphone mobile medical application is presented, that provides analysis of the health of skin on the face using a smartphone image and cloud-based image processing techniques. The mobile application employs the use of the camera to capture a front face image of a subject, after which the captured image is spatially calibrated based on fiducial points such as position of the iris of the eye. A facial recognition algorithm is used to identify features of the human face image, to normalize the image, and to define facial regions of interest (ROI) for acne assessment. We identify acne lesions and classify them into two categories: those that are papules and those that are pustules. Automated facial acne assessment was validated by performing tests on images of 60 digital human models and 10 real human face images. The application was able to identify 92% of acne lesions within five facial ROIs. The classification accuracy for separating papules from pustules was 98%. Combined with in-app documentation of treatment, lifestyle factors, and automated facial acne assessment, the app can be used in both cosmetic and clinical dermatology. It allows users to quantitatively self-measure acne severity and treatment efficacy on an ongoing basis to help them manage their chronic facial acne.

A CT-guided biopsy is a specialised surgical procedure where a needle is used to withdraw a tissue or fluid specimen from a lesion of interest. The needle is guided while being viewed by the surgeon on a computed tomography (CT) scan. CT guided biopsies expose patients to a high dosage of radiation. They are lengthy procedures and the lack of spatial reference while guiding the needle down the predicted path are some of the difficulties currently encountered. To explore possible approaches to this problem, we investigate the use of two infrared cameras capable of imaging the biopsy needle area. These are then mapped into scaled 3D co-ordinate space using an extension of a previously reported method. The system is able to read, in real-time, infrared data from two cameras and import the data. The result is a scaled 3D estimate of the needle endpoints.

In this paper, a scheme was presented to identify the locations of key features of a human face such as eyes, nose, chin known as the fiducial points and form a face graph. The relative distances between these features are calculated. These distance measures are considered to be unique identifying attributes of a person. The distance measures are used to train a Support Vector Machine (SVM). The identification takes place by matching the features of the presented person with the features that were used to train the SVM. The closest match results in identification. The Minimum Distance Classifier has been used to recognize a person uniquely using this SVM.

Recently, holistic face recognition technology has been increasingly mature. However, as for many unconstrained environments, the captured face image is more likely not holistic, but partial discriminative face area. To address this, we propose a fast and general method for partial face recognition. There, our method needn’t alignment by fiducial points or cropping to the same size, for all facial images in the gallery set and probe set. In other words, we use the initial captured faces as input. Besides, our method can deal with single sample face recognition problem. For a pair of gallery image and probe image, firstly we detect key-points as well as extracting their local descriptors. Specially, in order to improve robustness of descriptors, we exploit gradient orientation modification and L2 normalization. Then, we use sparse representation based on multi-descriptors to recognize probe image. Experimental results on public face datasets demonstrate the effectiveness of the proposed method.

Heart rate variability analysis (HRVA) gives valuable insight to the cardiovascular system. Electrocardiogram (ECG) based HRVA has been assessment gold standard but eavesdropping of wearable technology requires the comparison of its surrogacy to an accepted standard. In this study, optical and mechanical measures at distal artery waveform are compared to the electrical signal of the heart. The sensor data of the six healthy volunteers are collated and compared at fiducial points in various time, frequency and non-linear domains for HRVA. We have found that during early systole fiducial location on waveforms can be surrogate to ECG standard and mechanical sensor 2nd derivative proved to be the best among them. Also, the comparative technology shows enormous potential for cardiovascular diagnostic.

Seismocardiogram, SCG, is the measure of precordial vibrations produced by the beating heart, from which cardiac mechanics may be explored on a beat-to-beat basis. We recently collected a large amount of SCG data (>69 recording hours) from an astronaut to investigate cardiac mechanics during sleep aboard the International Space Station and on Earth. SCG sleep recordings are characterized by a prolonged duration and wide heart rate swings, thus a specific algorithm was developed for their analysis. In this article we describe the new algorithm and its performance. The algorithm is composed of three parts: 1) artifacts removal, 2) identification in each SCG waveform of four fiducial points associated with the opening and closure of the aortic and mitral valves, 3) beat-to-beat computation of indexes of cardiac mechanics from the SCG fiducial points. The algorithm was tested on two sleep recordings and yielded the identification of the fiducial points in more than 36,000 beats with a precision, quantified by the Positive Predictive Value, ≥99.2%. These positive findings provide the first evidence that cardiac mechanics may be explored by the automatic analysis of SCG long-lasting recordings, taken out of the laboratory setting, and in presence of significant heart rate modulations.

Registering a 3D facial model to a 2D image under occlusion is difficult. First, not all of the detected facial landmarks are accurate under occlusions. Second, the number of reliable landmarks may not be enough to constrain the problem. We propose a method to synthesize additional points (SensiblePoints) to create pose hypotheses. The visual clues extracted from the fiducial points, non-fiducial points, and facial contour are jointly employed to verify the hypotheses. We define a reward function to measure whether the projected dense 3D model is well-aligned with the confidence maps generated by two fully convolutional networks, and use the function to train recurrent policy networks to move the SensiblePoints. The same reward function is employed in testing to select the best hypothesis from a candidate pool of hypotheses. Experimentation demonstrates that the proposed approach is very promising in solving the facial model registration problem under occlusion., Comment: Accepted in International Joint Conference on Biometrics (IJCB) 2017

In this paper an analysis on the influence that the selection of fiducial points has on the Moving Least Square registration of medical images has been presented. MLS is a point based method which needs selection of fiducial (control) points. Here the mapping is weighted by the distance of current pixel from the selected point. Hence it is deemed significance to investigate on the effect that the position and number of the selected control points have on registered image. The analysis is done by manually selecting the points from rigid and non-rigid regions, near and far off regions from the two images and by also varying the number of points. To assess the results comparison has been done with the TPS registration by computing the TRE.

In this paper, we demonstrate an automatic image analysis for Micro Total Analysis Systems (μTAS) or Lab-on-Paper (LoP). By placing fiducial markers on μTAS, application can crop image and remove bending distortions, also it can perform colorimetric analysis. With previously taken photo of μTAS device with applied reagents, user can determine ie. illness presence by uploading photo into the program. In the paper was prepared μTAS sheet, algorithm for finding fiducial points and software for colorimetric analysis written in Matlab.

The rigid registration is a key step of Image Guided Surgery (IGS), and the point-pair method is the main way used for registration. However the configuration of fiducial points has a great influence on the registration accuracy at the target point. Now almost all the optimization method of fiducial points configuration relies on the empirical simulation-based Fitzpatrick's target registration error (TRE). In this paper, a phantom and some markers were designed and some experiments were conducted to measure and compare the affecting factors on the registration. By the markers repeated selections, the fiducial location error (FLE) has a small deviation of maximum 0.4 mm, and the average of the Fitzpatrick's TRE (F-TRE) has almost 86% proportion to the average of the actual TRE (A-TRE), but the standard deviation (STD) just has 7% proportion. Also, the experiment result showed that six fiducial markers already had the 86% accuracy, and spreading the fiducial markers led to 30% reduction in mean of A-TRE and 40% reduction in STD of A-TRE comparing with the centralized. Overall, to find a strategy of optimization, reducing the TRE has the great meaning to support safer and more accurate minimally IGS procedures.

Intraoperative guidance is often used during procedures to provide surgeons with information linking the current surgical scene to some preoperative plan or model. This allows surgeons to visualize structures or targets of interest that are not visible in the intraoperative imaging modality. A common method to enable this type of technology is the use of fiducials that can be located both preoperatively and intraoperatively. In this work, we focus on the registration of preoperative computed tomography with intraoperative three-dimensional ultrasound. We examine the use of titanium fiducials activated intraoperatively by the photoacoustic effect. The photoacoustic effect is generated when the metal fiducials are illuminated by laser light, resulting in an acoustic signal at each fiducial that can be observed by a conventional ultrasound transducer. A transrectal ultrasound transducer is translated to generate a three-dimensional ultrasound volume. The set of fiducial points are segmented from the three-dimensional ultrasound volume and registered with the corresponding set of fiducial points segmented from the computed tomography volume. The target registration error metric is used to validate the registration between these two coordinate systems. The resulting target registration error was 1.55 ± 0.62mm. It was observed in this experiment that there were certain acoustic reverberation and refraction artifacts that occurred at distances coinciding with the size of the fiducial. Further work with different shapes and sizes of fiducials may be necessary to quantify and analyze how they affect the generation of a photoacoustic signal.

O bjective: The purpose of this research was to design a delineation algorithm that could detect specific fiducial points of the seismocardiogram (SCG) signal with or without using the electrocardiogram (ECG) R-wave as the reference point. The detected fiducial points were used to estimate cardiac time intervals. Due to complexity and sensitivity of the SCG signal, the algorithm was designed to robustly discard the low-quality cardiac cycles, which are the ones that contain unrecognizable fiducial points. Method: The algorithm was trained on a dataset containing 48 318 manually annotated cardiac cycles. It was then applied to three test datasets: 65 young healthy individuals (dataset 1), 15 individuals above 44 years old (dataset 2), and 25 patients with previous heart conditions (dataset 3). Results: The algorithm accomplished high prediction accuracy with the root-mean-square error of less than 5 ms for all the test datasets. The algorithm overall mean detection rate per individual recordings (DRI) were 74%, 68%, and 42% for the three test datasets when concurrent ECG and SCG were used. For the standalone SCG case, the mean DRI was 32%, 14%, and 21%. Conclusion: When the proposed algorithm was applied to concurrent ECG and SCG signals, the desired fiducial points of the SCG signal were successfully estimated with a high detection rate. For the standalone case, however, the algorithm achieved high prediction accuracy and detection rate for only the young individual dataset. Significance: The presented algorithm could be used for accurate and noninvasive estimation of cardiac time intervals.

The accuracy of a real facial recognition system can be varied according to the accuracy of the eye detection algorithm when we design and implement a semi-automatic facial recognition algorithm depending on the eye position of a database. In this paper, a fully automatic facial recognition algorithm is proposed such that Gabor and LBP features are extracted from fiducial points of a face graph which was created by using fiducial points based on the eyes, nose, mouth and border lines of a face, fitted on the face image. In this algorithm, the recognition performance could be increased because a face graph can be fitted on a face image automatically and fiducial points based LPB features are implemented with the basic Gabor features. The simulation results show that the proposed algorithm can be used in real-time recognition for more than 1,000 faces and produce good recognition performance for each data set.

This paper presents an efficient and powerful method of electrocardiogram (ECG) identification. Specially, a set of discriminative feature representations can be learned from one-dimensional ECG signals with an arbitrary origin through deep learning, referred to as deep fusion features. Non-linear classifier is adopted to classify test ECG signals. A final simple voting step can further enhance performance. Based on the above steps, our method can reduce the dependence of algorithm accuracy on the origin and length of the ECG signal. Unlike traditional methods, detecting fiducial points and combining features are not required. Moreover, test process can use parallel processing to improve efficiency. The method achieves 99.33% accuracy for a publicly available database. The experiments demonstrate that our method is efficient and powerful in real applications.

Face alignment is very crucial to the task of face attributes recognition. The performance of face attributes recognition would notably degrade if the fiducial points of the original face images are not precisely detected due to large lighting, pose and occlusion variations. In order to alleviate this problem, we propose a spatial transform based deep CNNs to improve the performance of face attributes recognition. In this approach, we first learn appropriate transformation parameters by a carefully designed spatial transformer network called LoNet to align the original face images, and then recognize the face attributes based on the aligned face images using a deep network called ClNet. To the best of our knowledge, this is the first attempt to use spatial transformer network in face attributes recognition task. Extensive experiments on two large and challenging databases (CelebA and LFWA) clearly demonstrate the effectiveness of the proposed approach over the current state-of-the-art.

Arrhythmia detection algorithms require the exact and instantaneous detection of fiducial points in the ECG signal. These fiducial points (QRS-complex, P- and T-wave) correspond to distinct cardiac contraction phases. The performance evaluation of different fiducial points detection algorithms require the existence of large databases (DBs) encompassing reference annotations. Up to last year, P- and T-wave annotations were only available for the QT DB. This was addressed by Elgendi et al. who provided P- and T-wave annotations to the MIT-BIH arrhythmia DB. A variety of ECG fiducial points detection algorithms exists in literature, whereas, to the best knowledge of the authors, we could not identify any single-lead algorithm ready for instantaneous P- and T-wave detection. In this work, we present three P- and T-wave detection algorithms: a revised version for QRS detection using line fitting capable to detect P- and T-wave, an expeditious version of a wavelet based ECG delineation algorithm, and a fast naive fiducial points detection algorithm. The fast naive fiducial points detection algorithm performed best on both DBs with sensitivities ranging from 73.0% (P-wave detection, error interval of ± 40 ms) to 89.4% (T-wave detection, error interval of ± 80 ms). As this algorithm detects a wave event in every search window, it has to be investigated how this affects arrhythmia detection algorithms. The reference Matlab implementations are available for download to encourage the development of high-accurate and automated ECG processing algorithms for the integration in daily life using mobile computers.

Image-guided neurosurgery systems (IGNS) play an important role in intracranial surgery. Adhesive markers on the skin are widely used for patient-to-image registration, with the centers of these markers serving as fiducial points in point-pair registration. In this paper, we propose a novel algorithm to automatically locate the center of these markers. The algorithm compares the marker model and the surface patches from Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) images of the patient's head. Automatic localization of these centers will help to reduce the human error in registration and speed up the registration process. Experiments with clinical 3D CT and MRI data confirm that this algorithm can accurately locate the centers of these markers.

Background The point-pair method is widely used to register the patient space and the image space in image-guided neuro surgery. In this kind of registration, the distribution of fiducial points has a great influence on the registration accuracy at the target point. Methods In this paper we propose the idea of target registration accuracy contribution (TRAC) of current fiducial points and of candidate fiducial points. Based on these two values, we give an interactive method to help the surgeon optimize the distribution of fiducial points so as to improve the registration accuracy at the target point. Results The experiments on real patients' data showed that this approach reduced approximately half of the registration error at the target point introduced by unsuitable distribution of fiducial points. When transferring the optimized distribution to real patients, the target registration accuracy obtained from optimization had no serious deterioration. Conclusion This study provides a practical approach for neurosurgeons to arrange the fiducial points properly on a patient's head. Copyright © 2008 John Wiley & Sons, Ltd.

In this paper we propose a real-time method for discrimination of ventricular ectopic and normal beats in the electrocardiogram (ECG). The heartbeat waveforms were evaluated within a fixed-length window around the fiducial points (100 ms before, 100 ms after) after being normalized. Our algorithm was designed to operate with no expert assistance; the operator is not required to initially select any known beat templates which makes it applicable in real-time. It is based on the dominancy of normal beats during the first several seconds for most of records which mostly match the real cases. In addition to R-R intervals, we extract other features as beat width, P wave existence for each beat. Also we apply template matching between the learned template and the unknown beat where template beats are created for each beat shape on runtime. Template matching not only classifies normal dominant beat but also multi-form ventricular ectopic beats where each form is classified separately so as doctors and medical stuff could take the better decision. Our proposed algorithm was tested on MIT-BIH ECG database records with normal and ventricular ectopic beat classes defined in AAMI standard while other records are excluded. Our results show 97.24% overall accuracy with 98.93% and 94.54% sensitivities for normal and ventricular ectopic beats respectively.

This paper concerns the authentication of individuals through analysis of electrocardiogram (ECG) signals. Because the human heart differs physiologically from one person to the next, ECG signals represent a rich source of information that offers strong potential for authentication or identification. We describe a novel approach to ECG-based biometrics in which a dynamical-systems model is employed, resulting in improved registration of pulses as compared to previous techniques. Parameters at the fiducial points are detected using a sum-of-Gaussians representation, resulting in an 18-component feature vector that can be used for classification. Using a publicly available dataset of ECG signals from 47 participants, a classifier was formulated using quadratic discriminant analysis (QDA). The observed mean authentication accuracies were 90% and 97% using 100 beats and 300 beats, respectively. Although tested with standard ECG signals only, we believe that the approach can be extended to other sensor types, such as fingertip-ECG devices.