Automatic Boundary Detection Research Articles

Deep learning for automated boundary detection and segmentation in organ donation photography

Abstract Objectives Medical photography is ubiquitous and plays an increasingly important role in the fields of medicine and surgery. Any assessment of these photographs by computer vision algorithms requires first that the area of interest can accurately be delineated from the background. We aimed to develop deep learning segmentation models for kidney and liver organ donation photographs where accurate automated segmentation has not yet been described. Methods Two novel deep learning models (Detectron2 and YoloV8) were developed using transfer learning and compared against existing tools for background removal (macBGRemoval, remBGisnet, remBGu2net). Anonymised photograph datasets comprised training/internal validation sets (821 kidney and 400 liver images) and external validation sets (203 kidney and 208 liver images). Each image had two segmentation labels: whole organ and clear view (parenchyma only). Intersection over Union (IoU) was the primary outcome, as the recommended metric for assessing segmentation performance. Results In whole kidney segmentation, Detectron2 and YoloV8 outperformed other models with internal validation IoU of 0.93 and 0.94, and external validation IoU of 0.92 and 0.94, respectively. Other methods – macBGRemoval, remBGisnet and remBGu2net – scored lower, with highest internal validation IoU at 0.54 and external validation at 0.59. Similar results were observed in liver segmentation, where Detectron2 and YoloV8 both showed internal validation IoU of 0.97 and external validation of 0.92 and 0.91, respectively. The other models showed a maximum internal validation and external validation IoU of 0.89 and 0.59 respectively. All image segmentation tasks with Detectron2 and YoloV8 completed within 0.13–1.5 s per image. Conclusions Accurate, rapid and automated image segmentation in the context of surgical photography is possible with open-source deep-learning software. These outperform existing methods and could impact the field of surgery, enabling similar advancements seen in other areas of medical computer vision.

Strain Localization in Sandstone‐Derived Fault Gouges Under Conditions Relevant to Earthquake Nucleation

AbstractConstraining strain localization and the growth of shear fabrics within brittle fault zones at sub‐seismic slip rates is important for understanding fault strength and frictional stability. We conducted direct shear experiments on simulated sandstone‐derived fault gouges at an effective normal stress of 40 MPa, a pore pressure of 15 MPa, and a temperature of 100°C. Using a passive strain marker and X‐ray Computed Tomography, we analyzed the spatial distribution of deformation in gouges deformed in the strain‐hardening, subsequent strain‐softening, and then steady‐state regimes at displacement rates of 1, 30, and 1,000 µm/s. We developed a machine‐learning‐based automatic boundary detection method to recognize the shear fabrics and quantify displacement partitioning between each fabric element. Our results show fabrics oriented along R1 and Y (including boundary) shears are the two major fabric elements. At rates of 1 and 30 µm/s, the relative amount of displacement on R1 shears is displacement dependent, increasing to ∼20% of the total displacement up to the strain‐softening stage, then decreasing to ∼10%–18% at the steady state. This trend is absent at the high rate where ∼18% of the displacement occurs on R1 shears throughout all investigated stages. At all rates, the relative amount of displacement on Y shears increases linearly with displacement to a total of larger than 50% at the steady state. Our study provides constraints on the development of the active slip zone, which is an important factor controlling heating and weakening associated with small‐magnitude earthquakes with limited displacement (mm‐dm), such as induced seismicity.

Towards Automatic Boundary Detection for Human-AI Collaborative Hybrid Essay in Education

The recent large language models (LLMs), e.g., ChatGPT, have been able to generate human-like and fluent responses when provided with specific instructions. While admitting the convenience brought by technological advancement, educators also have concerns that students might leverage LLMs to complete their writing assignments and pass them off as their original work. Although many AI content detection studies have been conducted as a result of such concerns, most of these prior studies modeled AI content detection as a classification problem, assuming that a text is either entirely human-written or entirely AI-generated. In this study, we investigated AI content detection in a rarely explored yet realistic setting where the text to be detected is collaboratively written by human and generative LLMs (termed as hybrid text for simplicity). We first formalized the detection task as identifying the transition points between human-written content and AI-generated content from a given hybrid text (boundary detection). We constructed a hybrid essay dataset by partially and randomly removing sentences from the original student-written essays and then instructing ChatGPT to fill in for the incomplete essays. Then we proposed a two-step detection approach where we (1) separated AI-generated content from human-written content during the encoder training process; and (2) calculated the distances between every two adjacent prototypes (a prototype is the mean of a set of consecutive sentences from the hybrid text in the embedding space) and assumed that the boundaries exist between the two adjacent prototypes that have the furthest distance from each other. Through extensive experiments, we observed the following main findings: (1) the proposed approach consistently outperformed the baseline methods across different experiment settings; (2) the encoder training process (i.e., step 1 of the above two-step approach) can significantly boost the performance of the proposed approach; (3) when detecting boundaries for single-boundary hybrid essays, the proposed approach could be enhanced by adopting a relatively large prototype size (i.e., the number of sentences needed to calculate a prototype), leading to a 22% improvement (against the best baseline method) in the In-Domain evaluation and an 18% improvement in the Out-of-Domain evaluation.

Unsupervised topological analysis of polarized light microscopy: application to quantitative birefringence imaging.

The determination of birefringence (magnitude and axis orientation) of optical materials is of significant interest in various fields. In the case of composite samples, this task becomes complicated and time-consuming; therefore, a partially automated procedure for reconstructing birefringence spatial distribution becomes valuable. Herein, we propose a procedure to reconstruct the spatial distributions of the retardance and optical axis orientation in a geological thin section from sparse quantitative birefringence measurements, using automatic boundary detection on cross-polarized light microscopy images. We examine two particular areas on the selected geological thin section: one that presents a uniaxial crystal with a circular cross-section of its refractive index ellipsoid and the other with grains of varying orientations. The measurement gives the orientation of the grain's optical axis both in and out of the plane of the thin section, which explains the qualitative observations with the cross-polarized light microscope. Future work will connect the measured orientation of the rock thin section with its 3D geological orientation in the field.

Automated Variational Nonlinear Chirp Mode Decomposition for Bearing Fault Diagnosis

The variational non-linear chirp mode decomposition (VNCMD) requires initialization of number of modes (NMs) and instantaneous frequency (IF). This paper proposes an automated method for NM selection and IF initialization which works on the scale-space representation based automated boundary detection in magnitude spectrum (MS). The proposed automated VNCMD (AVNCMD) method is applied for bearing fault detection in which the kurtosis based dominant mode selection method is recommended. The instantaneous amplitude (IA) and IF with spectral entropy are computed from the dominant mode. Features are given to feed forward neural network classifier. Methodology is investigated on two datasets for inner race, outer race, and ball race faults detection. The proposed method classifies inner race, outer race, and ball race bearing faults with 97.52% accuracy and classifies inner race and outer race bearing fault with 100% accuracy. Efficacy of the proposed method is compared with the existing methods to justify the superiority.

Method for automatic detection of movement-related EEG pattern time boundaries

Method for automatic detection of movement-related EEG pattern time boundaries

Automatic boundary detection and severity assessment of mitral regurgitation

The most prevalent disease, cardiovascular disease (CVD), has a high mortality rate among people. The prognosis of CVD depends on early diagnosis. In the event of any abnormality, such as valvular regurgitation and stenosis, the blood flow patterns, direction, and velocity are detected using the color Doppler echocardiography. Today’s clinicians determine the regurgitant jet region manually in the echocardiographic images with the aid of a cursor to assess the severity of mitral regurgitation (MR). Manual tracing takes a lot of time and involves both intra and inter-personal variances. This research offers an automatic delineation technique to quantify the regurgitant jet area value to address this problem. The level of MR severity, specifically mild, moderate, and severe, is then connected to the analyzed jet area. The suggested approach combines various segmentation methods, such as color space model, thresholding, and morphological operators, for segmenting echocardiographic pictures. A total of 75 MR-related images from 25 individuals have been evaluated. In light of the regurgitant jet area identified by the suggested technique and contrasted with the manual way of MR assessment, the severity of MR has been assessed. The segmentation procedure can be segmented to remove the imitation areas using the established method. The regurgitant jet region collected using the suggested technique satisfies the clinical objectives in the quick evaluation of MR severity.

Automatic detection of rock boundaries using a hybrid recurrence quantification analysis and machine learning techniques

Automatic detection of rock boundaries using a hybrid recurrence quantification analysis and machine learning techniques

Automatic esophagus Z‐line delineation in endoscopic images using a new boundary linking method

Due to the American cancer society, many people with esophageal adenocarcinoma are not survived. The treatment rate can be significant in the early detection of Barrett's esophagus (BE) as a premalignant stage for adenocarcinoma. An important landmark to detect BE is the Z-line. BE segmentation is already highly dependent upon the operator's knowledge and skill. The main aim of this study is automatic Z-line extraction using endoscopic images leading to segmentation of the early BE stage. To this end, a computer-aided detection method exploiting k-means clustering, image segmentation using the edge detector, and a novel boundary linking algorithm is proposed. For the evaluation, the gold standard is considered the average contours of Z-lines extracted by the three experts. The proposed method annotated the Z-line with the accuracy and precision of 0.92 and 0.87, respectively, and the value of the average boundary distance is 5.9 pixels. To the results and visual inspection, the presented method can be used for efficient and robust extraction of the Z-line at the early BE stage. Furthermore, it can be used in other medical imaging applications with complex boundaries and low contrast in the images, limiting the common automatic boundary detection methods.

Automatic detection of prosodic boundaries in spontaneous speech.

Automatic speech recognition (ASR) and natural language processing (NLP) are expected to benefit from an effective, simple, and reliable method to automatically parse conversational speech. The ability to parse conversational speech depends crucially on the ability to identify boundaries between prosodic phrases. This is done naturally by the human ear, yet has proved surprisingly difficult to achieve reliably and simply in an automatic manner. Efforts to date have focused on detecting phrase boundaries using a variety of linguistic and acoustic cues. We propose a method which does not require model training and utilizes two prosodic cues that are based on ASR output. Boundaries are identified using discontinuities in speech rate (pre-boundary lengthening and phrase-initial acceleration) and silent pauses. The resulting phrases preserve syntactic validity, exhibit pitch reset, and compare well with manual tagging of prosodic boundaries. Collectively, our findings support the notion of prosodic phrases that represent coherent patterns across textual and acoustic parameters.

Machine Learning Driven Contouring of High-Frequency Four-Dimensional Cardiac Ultrasound Data.

Automatic boundary detection of 4D ultrasound (4DUS) cardiac data is a promising yet challenging application at the intersection of machine learning and medicine. Using recently developed murine 4DUS cardiac imaging data, we demonstrate here a set of three machine learning models that predict left ventricular wall kinematics along both the endo- and epi-cardial boundaries. Each model is fundamentally built on three key features: (1) the projection of raw US data to a lower dimensional subspace, (2) a smoothing spline basis across time, and (3) a strategic parameterization of the left ventricular boundaries. Model 1 is constructed such that boundary predictions are based on individual short-axis images, regardless of their relative position in the ventricle. Model 2 simultaneously incorporates parallel short-axis image data into their predictions. Model 3 builds on the multi-slice approach of model 2, but assists predictions with a single ground-truth position at end-diastole. To assess the performance of each model, Monte Carlo cross validation was used to assess the performance of each model on unseen data. For predicting the radial distance of the endocardium, models 1, 2, and 3 yielded average R2 values of 0.41, 0.49, and 0.71, respectively. Monte Carlo simulations of the endocardial wall showed significantly closer predictions when using model 2 versus model 1 at a rate of 48.67%, and using model 3 versus model 2 at a rate of 83.50%. These finding suggest that a machine learning approach where multi-slice data are simultaneously used as input and predictions are aided by a single user input yields the most robust performance. Subsequently, we explore the how metrics of cardiac kinematics compare between ground-truth contours and predicted boundaries. We observed negligible deviations from ground-truth when using predicted boundaries alone, except in the case of early diastolic strain rate, providing confidence for the use of such machine learning models for rapid and reliable assessments of murine cardiac function. To our knowledge, this is the first application of machine learning to murine left ventricular 4DUS data. Future work will be needed to strengthen both model performance and applicability to different cardiac disease models.

Structural Segmentation of Alap in Dhrupad Vocal Concerts

Dhrupad vocal concerts exhibit a temporal evolution through a sequence of homogeneous sections marked by shared rhythmic characteristics. In this work, we address the segmentation of a concert audio’s unmetered improvisatory section into musically meaningful segments at the highest time scale. Motivated by the distinct musical properties of the sections and their corresponding acoustic correlates, we compute a number of features for the segment boundary detection task. Both supervised and unsupervised approaches are tested using a dataset of commercial performance recordings that is manually annotated. The dataset is augmented suitably for training and testing of the models to obtain new insights about the relevance of the different rhythmic, melodic and timbral cues in the automatic boundary detection task. We also explore the use of a convolutional neural network trained on mel-scale magnitude spectrograms for the boundary detection task to observe that while the implicit musical cues are largely learned by the network, it is less robust to deviations from training data characteristics. We conclude that it can be rewarding to investigate knowledge driven features on new genres and tasks, both to achieve reasonable performance outcomes given limited datasets and for drawing a deeper understanding of genre characteristics from the acoustical analyses.

Modelling automatic detection of prosodic boundaries for brazilian portuguese spontaneous speech

Speech is segmented into intonational units marked by prosodic boundaries. This segmentation is claimed to have important consequences on syntax, information structure and cognition. This work aims both to investigate the phonetic-acoustic parameters that guide the production and perception of prosodic boundaries, and to develop models for automatic detection of prosodic boundaries in male monological spontaneous speech of Brazilian Portuguese. Two samples were segmented into intonational units by two groups of trained annotators. The boundaries perceived by the annotators were tagged as either terminal or non-terminal. A script was used to extract 111 phonetic-acoustic parameters along speech signal in a right and left windows around the boundary of each phonological word. The extracted parameters comprise measures of (1) Speech rate and rhythm; (2) Standardized segment duration; (3) Fundamental frequency; (4) Intensity; (5) Silent pause. The script considers as prosodic boundary positions at which at least 50% of the annotators indicated a boundary of the same type. A training of models composed by the parameters extracted by the script was developed; these models, were then improved heuristically. The models were developed from the two samples and from the whole data, both using non-balanced and balanced data. Linear Discriminant Analysis algorithm was adopted to produce the models. The models for terminal boundaries show a much higher performance than those for non-terminal ones. In this paper we: (i) show the methodological procedures; (ii) analyze the different models; (iii) discuss some strategies that could lead to an improvement of our results.

Detecção automática de fronteiras prosódicas entre unidades entonacionais

A fala é segmentada em unidades entonacionais marcadas por fronteiras prosódicas. Este trabalho tem como objetivo investigar os parâmetros fonético-acústicos que orientam a produção e a percepção de fronteiras prosódicas e desenvolver modelos para detecção automática de fronteiras prosódicas em fala espontânea. Duas amostras de trechos de fala espontânea masculina foram segmentadas em unidades entonacionais por dois grupos de segmentadores treinados. As fronteiras percebidas pelos segmentadores foram anotadas como terminais (TB) ou não-terminais (NTB). Um script foi utilizado para extrair parâmetros fonético-acústicos ao longo do sinal de fala. Os parâmetros extraídos compreendem medidas de: 1) Velocidade e ritmo da fala; 2) Duração normalizada; 3) Frequência fundamental; 4) Intensidade; 5) Pausa silenciosa. Foi desenvolvido um treinamento de modelos compostos por múltiplos parâmetros projetados para a identificação automática das fronteiras marcadas pelos segmentadores. Utilizou-se o algoritmo Linear Discriminant Analysis e considerou-se como fronteira posições em que pelo menos 50% dos segmentadores indicaram uma fronteira do mesmo tipo. O modelo de detecção automática de fronteiras terminais mostra uma convergência de 80% em relação às fronteiras terminais observadas pelos segmentadores na amostra I. Para fronteiras não-terminais, foram obtidos três modelos de classificação estatística. Juntos, os três modelos mostram uma convergência de 98% em relação às fronteiras não-terminais observadas pelos segmentadores na amostra I. Os modelos foram validados posteriormente na amostra II. Os resultados da validação indicam que o desempenho do modelo TB é de 74% e o dos modelos NTB é de 88% na amostra II.

DETECTION OF VISIBLE BOUNDARIES FROM UAV IMAGES USING U-NET

Abstract. Peoples land rights are secure if they are registered in a formal cadastral system. More than 70% of global land rights are not registered in any formal cadastral system. The contemporary efforts are on accelerating the cadastral mapping process as a basis of defining land rights boundaries. Proposed surveying techniques are indirect ones – delineation of visible parcel boundaries from remote sensing imagery. This research aims at automizing the procedure of visible boundary delineation from Unmanned Aerial Vehicle (UAV) imagery through deep learning. U-Net architecture was selected to train the model and predict visible boundaries. The model was trained on an available edge detection dataset, which was the closest to our domain problem. The model was tested on a tiled UAV images. The U-Net architecture was implemented in Keras and written in Python, running on top of the TensorFlow library. The training was done through Google Colaboratory. The evaluation metrics of the trained model indicated 0.95 overall accuracy. The average percentage of correctly detected visible boundaries was almost 80% for the tiled UAV images. This percentage is very satisfying since the model was trained on everyday imagery which is very different from UAV ones. The automatic boundary detection by using U-Net is applicable mostly for rural areas where the visibility of the boundaries is continuous. In cases where the boundaries are not visible, manual delineations are still required.

A deep learning based medical image segmentation technique in Internet-of-Medical-Things domain

Medical Image Segmentation is the process of automatic or semi-automatic detection of boundaries within a 2D or 3D image in Internet-of-Medical-Things (IoMT) domain. The main difficulty of medical image segmentation is the high variability in medical images. For example, CT images contain a large amount of noise, and complex boundaries. In this paper, we propose an adaptive fully dense(AFD) neural network for CT image segmentation. By adding the horizontal connections in UNet structure, it can extract various features from all layers adaptively. And it use ensemble training for the output to extract more edge information in the multiple rounds training. We have validated our method on two data sets, a natural scene image data set and a liver cancer CT image data set. The experimental results demonstrate that it performs better than state-of-the-art segmentation methods. And our method yields superior segmentation results for CT images with complex boundaries.

Fast automatic detection of geological boundaries from multivariate log data using recurrence

Manual interpretation of data collected from drill holes for mineral or oil and gas exploration is time-consuming and subjective. Identification of geological boundaries and distinctive rock physical property domains is the first step of interpretation. We introduce a multivariate technique, that can identify geological boundaries from petrophysical or geochemical data. The method is based on time-series techniques that have been adapted to be applicable for detecting transitions in geological spatial data. This method allows for the use of multiple variables in detecting different lithological layers. Additionally, it reconstructs the phase space of a single drill-hole or well to be applicable for further investigations across other holes or wells. The computationally cheap method shows efficiency and accuracy in detecting boundaries between lithological layers, which we demonstrate using examples from mineral exploration boreholes and an offshore gas exploration well.

Gabor phase response based scheme for accurate pectoral muscle boundary detection

Pectoral muscle segmentation is a very important step in any Computer Aided Detection (CAD) system, for automatic detection of different possible signs of malignancy in the mammographic image. Presence of high-density glandular tissues overlapping pectoral muscle boundary, presence of pectoral muscle of small size and low density, presence of pectoralis minor etc., poses a great challenge for researchers working on the problem of automatic pectoral muscle boundary detection. Gabor filter has been used by many researchers to detect textural edges. The ability that it can be tuned along a particular direction, makes it a very important image processing tool. In the proposed research work a Gabor filter tuned in the direction of pectoral muscle boundary is applied on the region of interest (ROI) containing pectoral muscle. Finally, the magnitude and phase response of Gabor filtered ROI image along with an edge-connect and region merge algorithm is used to detect pectoral muscle boundary. The proposed method is validated using 200 mammograms from the miniMIAS database and 200 mammograms from INbreast database. With an accuracy of 97.50% for miniMIAS database and 94.50% for INbreast database, the proposed method outperforms many state-of-the-art methods which show its suitability for CAD Systems.

Automatic Detection of Geoelectric Boundaries According to Lateral Logging Sounding Data by Applying a Deep Convolutional Neural Network

Abstract —Lateral logging sounding (LLS) is currently the only widely used Russian method of resistivity measurements, sensitive to vertical electrical resistivity in vertical wells. However, interpreting data measured by this method in thin-layered sections is difficult and requires the utilization of resource-intensive numerical simulation algorithms. Today, the development of computational methods and an increase in computer performance allow us to invert LLS data in the class of two-dimensional axisymmetric models. However, in virtue of the large number of difficulties associated with the nonlocal responses of the probes and their asymmetry, this process requires the active participation of a log analyst. One of the first issues is the creation of an initial approximation of the geoelectric model. It consists in splitting the target interval into layers within which the properties of the medium can be considered constant in the vertical direction, since LLS signals have a very complex shape in the intervals of alternation of beds with different resistivities. We propose applying a fully connected convolutional artificial neural network to automatically create sectional layering suitable for constructing the initial approximation of the geoelectric model for two-dimensional LLS data inversion, including vertical resistivity estimation. The neural network was trained and tested on the synthetic and field data measured in West Siberia. Based on the results of the testing, we established the workability of the proposed approach.

Boundary segmentation based on modified random walks for vascular Doppler optical coherence tomography images

Vascular Doppler optical coherence tomography (DOCT) images with weak boundaries are usually difficult for most algorithms to segment. We propose a modified random walk (MRW) algorithm with a novel regularization for the segmentation of DOCT vessel images. Based on MRW, we perform automatic boundary detection of the vascular wall from intensity images and boundary extraction of the blood flowing region from Doppler phase images. Dice, sensitivity, and specificity coefficients were adopted to verify the segmentation performance. The experimental study on DOCT images of the mouse femoral artery showed the effectiveness of our proposed method, yielding three-dimensional visualization and quantitative evaluation of the vessel.