Segment Boundaries Research Articles

Deep learning for the rapid automatic segmentation of forearm muscle boundaries from ultrasound datasets.

Ultrasound (US) is widely used in the clinical diagnosis and treatment of musculoskeletal diseases. However, the low efficiency and non-uniformity of artificial recognition hinder the application and popularization of US for this purpose. Herein, we developed an automatic muscle boundary segmentation tool for US image recognition and tested its accuracy and clinical applicability. Our dataset was constructed from a total of 465 US images of the flexor digitorum superficialis (FDS) from 19 participants (10 men and 9 women, age 27.4 ± 6.3years). We used the U-net model for US image segmentation. The U-net output often includes several disconnected regions. Anatomically, the target muscle usually only has one connected region. Based on this principle, we designed an algorithm written in C++ to eliminate redundantly connected regions of outputs. The muscle boundary images generated by the tool were compared with those obtained by professionals and junior physicians to analyze their accuracy and clinical applicability. The dataset was divided into five groups for experimentation, and the average Dice coefficient, recall, and accuracy, as well as the intersection over union (IoU) of the prediction set in each group were all about 90%. Furthermore, we propose a new standard to judge the segmentation results. Under this standard, 99% of the total 150 predicted images by U-net are excellent, which is very close to the segmentation result obtained by professional doctors. In this study, we developed an automatic muscle segmentation tool for US-guided muscle injections. The accuracy of the recognition of the muscle boundary was similar to that of manual labeling by a specialist sonographer, providing a reliable auxiliary tool for clinicians to shorten the US learning cycle, reduce the clinical workload, and improve injection safety.

Aggregate boundary recognition of asphalt mixture CT images based on convolutional neural networks

This study aims to propose an intelligent aggregate boundary segmentation algorithm based on convolutional neural networks (CNNs) and watershed algorithm for quickly recognising the boundary of aggregates on asphalt mixture CT images. CNN was concisely introduced. An aggregate boundary segmentation method for asphalt mixture CT images based on CNN and watershed algorithm was depicted in detail. The generalisation ability, that is, the effectiveness of image segmentation method by CNN and watershed algorithm was also evaluated. Results showed that the intelligent segmentation algorithm proposed by combining CNNs and watershed algorithm could effectively segment the aggregate boundaries on asphalt mixture CT images with different levels of boundary definition. The adhesion between aggregates on asphalt mixture CT images could be reduced using a custom multi-threshold segmentation (CMTS) method. The intelligent image segmentation algorithm had more accurate segmentation and more convenient operation than Canny and multi-threshold algorithms.

A design logic for sequential segmentation across organisms.

Multitudes of organisms display metameric compartmentalization of their body plan. Segmentation of these compartments happens sequentially in diverse phyla. In several sequentially segmenting species, periodically active molecular clocks and signaling gradients have been found. The clocks are proposed to control the timing of segmentation, while the gradients are proposed to instruct the positions of segment boundaries. However, the identity of the clock and gradient molecules differs across species. Furthermore, sequential segmentation of a basal chordate, Amphioxus, continues at late stages when the small tail bud cell population cannot establish long-range signaling gradients. Thus, it remains to be explained how a conserved morphological trait (i.e., sequential segmentation) is achieved by using different molecules or molecules with different spatial profiles. Here, we first focus on sequential segmentation of somites in vertebrate embryos and then draw parallels with other species. Thereafter, we propose a candidate design principle that has the potential to answer this puzzling question.

DRNet: Dual-stage refinement network with boundary inference for RGB-D semantic segmentation of indoor scenes

Semantic segmentation is a dense pixel prediction task, and its accuracy depends on the extraction of long-range contextual knowledge and refinement of segmentation boundaries. Most segmentation methods are based on feature extraction using a convolutional neural network, and layer-by-layer sampling and fusion are applied to solve inherent problems such as chaotic boundaries and scattered objects. Owing to the limited receptive field and loss of details during downsampling, the segmentation results may be unsatisfactory. To address existing shortcomings, we propose a dual-stage refinement network (DRNet) for semantic segmentation. In the first stage, we adopt an efficient spatiotemporal representation learning framework called UniFormer. We also use a novel boundary extractor and initial segmentation map generator to obtain rough segmentation results. In the second stage, we use the rough segmentation map and extracted boundary information in a graph reasoning module that restores the class boundary features while completing global modeling and local information inference. Benefiting from the acquisition of long-range dependencies between image pixels, contextual information promotes the distinction of pixel categories. In addition, edge information can increase the interclass distinguishability and refine the segmentation boundaries. Results from extensive experiments demonstrate that the proposed DRNet outperforms state-of-the-art semantic segmentation methods. The codes and results are available at: https://github.com/EnquanYang2022/DRNet.

Developing an experimental-computational workflow to study the biomechanics of the human conventional aqueous outflow pathway.

The aqueous humor actively interacts with the trabecular meshwork (TM), juxtacanalicular tissue (JCT), and Schlemm's canal (SC) through a dynamic fluid-structure interaction (FSI) coupling. Despite the fact that intraocular pressure (IOP) undergoes significant fluctuations, our understanding of the hyperviscoelastic biomechanical properties of the aqueous outflow tissues is limited. In this study, a quadrant of the anterior segment from a normal human donor eye was dynamically pressurized in the SC lumen, and imaged using a customized optical coherence tomography (OCT). The TM/JCT/SC complex finite element (FE) with embedded collagen fibrils was reconstructed based on the segmented boundary nodes in the OCT images. The hyperviscoelastic mechanical properties of the outflow tissues' extracellular matrix with embedded viscoelastic collagen fibrils were calculated using an inverse FE-optimization method. Thereafter, the 3D microstructural FE model of the TM, with adjacent JCT and SC inner wall, from the same donor eye was constructed using optical coherence microscopy and subjected to a flow load-boundary from the SC lumen. The resultant deformation/strain in the outflow tissues was calculated using the FSI method, and compared to the digital volume correlation (DVC) data. TM showed larger shear modulus (0.92MPa) compared to the JCT (0.47MPa) and SC inner wall (0.85MPa). Shear modulus (viscoelastic) was larger in the SC inner wall (97.65MPa) compared to the TM (84.38MPa) and JCT (56.30MPa). The conventional aqueous outflow pathway is subjected to a rate-dependent IOP load-boundary with large fluctuations. This necessitates addressing the biomechanics of the outflow tissues using hyperviscoelastic material-model. STATEMENT OF SIGNIFICANCE: While the human conventional aqueous outflow pathway is subjected to a large-deformation and time-dependent IOP load-boundary, we are not aware of any studies that have calculated the hyperviscoelastic mechanical properties of the outflow tissues with embedded viscoelastic collagen fibrils. A quadrant of the anterior segment of a normal humor donor eye was dynamically pressurized from the SC lumen with relatively large fluctuations. The TM/JCT/SC complex were OCT imaged and the mechanical properties of the tissues with embedded collagen fibrils were calculated using the inverse FE-optimization algorithm. The resultant displacement/strain in the FSI outflow model was validated versus the DVC data. The proposed experimental-computational workflow may significantly contribute to understanding of the effects of different drugs on the biomechanics of the conventional aqueous outflow pathway.

Solute drag-controlled grain growth in magnesium investigated by quasi in-situ orientation mapping and level-set simulations

Critical properties of metallic materials, such as the yield stress, corrosion resistance and ductility depend on the microstructure and its grain size and size distribution. Solute atoms that favorably segregate to grain boundaries produce a pinning atmosphere that exerts a drag pressure on the boundary motion, which strongly affects the grain growth behavior during annealing. In the current work, the characteristics of grain growth in an annealed Mg-1 wt.%Mn-1 wt.%Nd magnesium alloy were investigated by advanced experimental and modeling techniques. Systematic quasi in-situ orientation mappings with a scanning electron microscope were performed to track the evolution of local and global microstructural characteristics as a function of annealing time. Solute segregation at targeted grain boundaries was measured using three-dimensional atom probe tomography. Level-set computer simulations were carried with different setups of driving forces to explore their contribution to the microstructure development with and without solute drag. The results showed that the favorable growth advantage for some grains leading to a transient stage of abnormal grain growth is controlled by several drivers with varying importance at different stages of annealing. For longer annealing times, residual dislocation density gradients between large and smaller grains are no longer important, which leads to microstructure stability due to predominant solute drag. Local fluctuations in residual dislocation energy and solute concentration near grain boundaries cause different boundary segments to migrate at different rates, which affects the average growth rate of large grains and their evolved shape.

Abnormal data cleaning for wind turbines by image segmentation based on active shape model and class uncertainty

Wind power curve describes the relationship between wind speed and output power of wind turbine. In this paper, we propose a novel image-based abnormal data cleaning algorithm aiming to provide high quality data for wind power curve modeling and relevant applications. For image binarization, a multiscale space is built to determine the optimal scale at which the most stable as well as informative shape representations are extracted. The active shape model is utilized to model the power curve shape and generate a set of segmentation candidates for filtering abnormal data. An energy space is constructed based on class uncertainty map generated from dual feature images and eventually the optimal segmentation boundary is determined by an optimization process. Comparative experiments are conducted over the real-world datasets consisted of 37 wind turbines from two wind farms located in West and East China respectively. The accuracy reaches as high as 0.95, the precision rate P is 0.98, recall rate R is 0.99 and F1-score is 0.96, which demonstrates the superior performance of the proposed method. Besides, the reproducibility and sensitivity analyses are also carried out to assess the algorithm’s performance.

Towards Global Video Scene Segmentation with Context-Aware Transformer

Videos such as movies or TV episodes usually need to divide the long storyline into cohesive units, i.e., scenes, to facilitate the understanding of video semantics. The key challenge lies in finding the boundaries of scenes by comprehensively considering the complex temporal structure and semantic information. To this end, we introduce a novel Context-Aware Transformer (CAT) with a self-supervised learning framework to learn high-quality shot representations, for generating well-bounded scenes. More specifically, we design the CAT with local-global self-attentions, which can effectively consider both the long-term and short-term context to improve the shot encoding. For training the CAT, we adopt the self-supervised learning schema. Firstly, we leverage shot-to-scene level pretext tasks to facilitate the pre-training with pseudo boundary, which guides CAT to learn the discriminative shot representations that maximize intra-scene similarity and inter-scene discrimination in an unsupervised manner. Then, we transfer contextual representations for fine-tuning the CAT with supervised data, which encourages CAT to accurately detect the boundary for scene segmentation. As a result, CAT is able to learn the context-aware shot representations and provides global guidance for scene segmentation. Our empirical analyses show that CAT can achieve state-of-the-art performance when conducting the scene segmentation task on the MovieNet dataset, e.g., offering 2.15 improvements on AP.

Weakly-Supervised Semantic Segmentation for Histopathology Images Based on Dataset Synthesis and Feature Consistency Constraint

Tissue segmentation is a critical task in computational pathology due to its desirable ability to indicate the prognosis of cancer patients. Currently, numerous studies attempt to use image-level labels to achieve pixel-level segmentation to reduce the need for fine annotations. However, most of these methods are based on class activation map, which suffers from inaccurate segmentation boundaries. To address this problem, we propose a novel weakly-supervised tissue segmentation framework named PistoSeg, which is implemented under a fully-supervised manner by transferring tissue category labels to pixel-level masks. Firstly, a dataset synthesis method is proposed based on Mosaic transformation to generate synthesized images with pixel-level masks. Next, considering the difference between synthesized and real images, this paper devises an attention-based feature consistency, which directs the training process of a proposed pseudo-mask refining module. Finally, the refined pseudo-masks are used to train a precise segmentation model for testing. Experiments based on WSSS4LUAD and BCSS-WSSS validate that PistoSeg outperforms the state-of-the-art methods. The code is released at https://github.com/Vison307/PistoSeg.

Fast inference of genetic recombination rates in biobank scale data.

Although rates of recombination events across the genome (genetic maps) are fundamental to genetic research, the majority of current studies only use one standard map. There is evidence suggesting population differences in genetic maps, and thus estimating population-specific maps, are of interest. Although the recent availability of biobank-scale data offers such opportunities, current methods are not efficient at leveraging very large sample sizes. The most accurate methods are still linkage disequilibrium (LD)-based methods that are only tractable for a few hundred samples. In this work, we propose a fast and memory-efficient method for estimating genetic maps from population genotyping data. Our method, FastRecomb, leverages the efficient positional Burrows-Wheeler transform (PBWT) data structure for counting IBD segment boundaries as potential recombination events. We used PBWT blocks to avoid redundant counting of pairwise matches. Moreover, we used a panel-smoothing technique to reduce the noise from errors and recent mutations. Using simulation, we found that FastRecomb achieves state-of-the-art performance at 10-kb resolution, in terms of correlation coefficients between the estimated map and the ground truth. This is mainly because FastRecomb can effectively take advantage of large panels comprising more than hundreds of thousands of haplotypes. At the same time, other methods lack the efficiency to handle such data. We believe further refinement of FastRecomb would deliver more accurate genetic maps for the genetics community.

Instance segmentation method for weed detection using UAV imagery in soybean fields

Weed detection in crops is a new frontier of precision agriculture, which will enable the distinction between desirable and undesirable plants. Accurate and efficient weed detection and discrimination are fundamental for precision weed management. This study presents a novel approach combining the visible colour index with the instance segmentation method based on the encoder and decoder architecture. This approach tackles the challenge of accurately detecting and segmenting weeds amidst the dense distribution of weeds and soybean crops. The colour index was designed to highlight the contrast between plants and soil to mitigate the effects of illumination and background, while the integration of the ResNet101_v and DSASPP in the encoder-decoder architecture served to reinforce the extraction of abundant multi-scale semantic information to elevate the accuracy of weed patch boundary segmentation. Experimental results on real-world field imagery indicated that the proposed approach exhibited superior performance with an accuracy of 0.905 and an IoU score of 0.959 for weeds segmentation, an aAcc of 0.978, mIoU of 0.939 and a mAcc of 0.972 for overall performance. In comparison with Deeplabv3plus, Deeplabv3, FCN, U-net, FastFCN, Swin Transformer, and Vision Transformer, the proposed pipeline achieved comparable performance in terms of segmentation precision and execution time for differentiating between weeds and soybeans. This work represents a noteworthy contribution to the field of precision agriculture, offering a promising approach to the challenging task of weed detection and precision weed control.

О сегментации опухолей головного мозга по МРТ-изображениям с применением методов глубокого обучения

Segmentation of a brain tumor is one of the most difficult tasks in the analysis of medical images. The purpose of brain tumor segmentation is to create an accurate outline of brain tumor areas. Gliomas are the most common type of brain tumors. Diagnosis of patients with this disease is based on the analysis of the results of magnetic resonance imaging and segmentation of the tumor boundaries manually. However, due to the time-consuming nature of the manual segmentation process and errors, there is a need for a fast and reliable automatic segmentation algorithm. In recent years, deep learning methods have shown promising effectiveness in solving various computer vision problems, such as image classification, object detection and semantic segmentation. A number of methods based on deep learning have been applied to segmentation of brain tumors, and promising results have been achieved. The article proposes a hybrid method for solving the problem of segmentation of brain tumors based on its MRI images based on the U-Net architecture, the encoder of which uses a model of a deep convolutional neural network pre-trained on a set of ImageNet images. Among such models were used VGG16, VGG19, MobileNetV2, Inception, ResNet50, EfficientNetb7, InceptionResnetV2, DenseNet201, DenseNet121. Based on the hybrid method, the TL-U-Net model was implemented, and numerical experiments were carried out to train it with different encoder models for segmentation of brain tumors based on its MRI images. Computer experiments on a set of MRI images of the brain showed the effectiveness of the proposed approach, the best encoder model turned out to be the neural network Densenet121, which provided indicators of segmentation accuracy MeanIoU=90.34%, MeanDice=94.33%, accuracy=94.17%. The obtained estimates of segmentation accuracy are comparable or exceed similar estimates obtained by other researchers.

Three-dimensional propagation behavior of hydrogen-related intergranular cracks in high-strength martensitic steel

Using X-ray computed tomography and focused ion beam-scanning electron microscopy (FIB-SEM) serial sectioning, this study investigated the three-dimensional propagation behavior of hydrogen-related intergranular cracks in martensitic steel with a tensile strength of 1.2 GPa, focusing on segments of prior austenite grain boundaries (PAGB segments). X-ray computed tomography revealed that the crack morphology was more continuous in the hydrogen-charged specimens. Through FIB-SEM serial sectioning, we found that crack-tip blunting and ductile rupture of un-cracked ligaments were associated with specific PAGB segments in the uncharged specimen. In the case of hydrogen-related intergranular crack propagation, even very fine, low-angle PAGB segments (sub-micrometer size) can act as obstacles for crack propagation. Based on these results, we propose that the misorientation of each PAGB segment has a large influence on the local arrestability of intergranular crack propagation.

W-Net: A boundary-enhanced segmentation network for stroke lesions

Accurate lesion segmentation is a critical technology basis for the treatment and prognosis of stroke. Stroke lesion segmentation suffers from complex background and noise interferes due to the similar grayscale characteristics of tissue structures and imaging mode in medical image. To resolve the difficulty of fuzzy boundary for lesion segmentation, we propose a novel two-stage brain MRI lesion segmentation method, called W-Net. W-Net leverages the CNN and transformer-based method as the backbone network, and introduces boundary deformation module (BDM), boundary constraint module (BCM) to handle the tough condition of fuzzy boundary. The BDM utilizes circular convolution to correct the initial boundary and BCM exploits dilated convolution to dynamically constrain the object boundary. Moreover, we design a multi-task learning loss function to optimize the W-Net from both region and boundary perspectives. Extensive experiments are conducted using the two ischemic stroke lesion segmentation datasets, and the results are compared with other advanced models. The experimental results demonstrate that the proposed method is effective in preserving the edge information of stroke lesions, thus achieving competitive performance on the segmentation task.

Exploiting multi-granularity visual features for retinal layer segmentation in human eyes.

Accurate segmentation of retinal layer boundaries can facilitate the detection of patients with early ophthalmic disease. Typical segmentation algorithms operate at low resolutions without fully exploiting multi-granularity visual features. Moreover, several related studies do not release their datasets that are key for the research on deep learning-based solutions. We propose a novel end-to-end retinal layer segmentation network based on ConvNeXt, which can retain more feature map details by using a new depth-efficient attention module and multi-scale structures. In addition, we provide a semantic segmentation dataset containing 206 retinal images of healthy human eyes (named NR206 dataset), which is easy to use as it does not require any additional transcoding processing. We experimentally show that our segmentation approach outperforms state-of-the-art approaches on this new dataset, achieving, on average, a Dice score of 91.3% and mIoU of 84.4%. Moreover, our approach achieves state-of-the-art performance on a glaucoma dataset and a diabetic macular edema (DME) dataset, showing that our model is also suitable for other applications. We will make our source code and the NR206 dataset publicly available at (https://github.com/Medical-Image-Analysis/Retinal-layer-segmentation).

Dynamic BMP signaling mediates notochord segmentation in zebrafish

The vertebrate spine is a metameric structure composed of alternating vertebral bodies (centra) and intervertebral discs.1 Recent studies in zebrafish have shown that the epithelial sheath surrounding the notochord differentiates into alternating cartilage-like (col2a1/col9a2+) and mineralizing (entpd5a+) segments which serve as a blueprint for centra formation.2,3,4,5 This process also defines the trajectories of migrating sclerotomal cells that form the mature vertebral bodies.4 Previous work demonstrated that notochord segmentation is typically sequential and involves the segmented activation of Notch signaling.2 However, it is unclear how Notch is activated in an alternating and sequential fashion. Furthermore, the molecular components that define segment size, regulate segment growth, and produce sharp segment boundaries have not been identified. In this study, we uncover that a BMP signaling wave acts upstream of Notch during zebrafish notochord segmentation. Using genetically encoded reporters of BMP activity and signaling pathway components, we show that BMP signaling is dynamic as axial patterning progresses, leading to the sequential formation of mineralizing domains in the notochord sheath. Genetic manipulations reveal that type I BMP receptor activation is sufficient to ectopically trigger Notch signaling. Moreover, loss of Bmpr1ba and Bmpr1aa or Bmp3 function disrupts ordered segment formation and growth, which is recapitulated by notochord-specific overexpression of the BMP antagonist, Noggin3. Our data suggest that BMP signaling in the notochord sheath precedes Notch activation and instructs segment growth, facilitating proper spine morphogenesis.

Comparative Study of the Performance Improvement of an Axial Compressor with a Micro-Vortex Generator and a Suction Slot

Flow control methods have been gradually applied to improve the flow field characteristics of axial compressors. However, most of the current research is focused on the cascade, and few studies have taken a compressor as the research object. Therefore, a 3.5-stage transonic axial compressor is adopted to explore the characteristics of different flow control schemes. The effects of the micro-vortex generator (MVG), segmented boundary layer suction (BLS), and combined technology (COM) on the MVG and BLS are compared by the numerical simulation method. The research results are summarized as follows. The excessive accumulated low-energy fluid in the blade passage induces the occurrence of a corner stall in the last stage stator of an axial compressor. At this time, the compressor can still work, but the performance has accelerated deterioration. The flow characteristics are effectively improved when the MVG is introduced, the stall margin improvement ΔSM and the peak efficiency improvement Δηe of the last stage are 2.1% and 1.02%, respectively. Moreover, the BLS shows advantages in removing the three-dimensional reverse flow and decreasing the total pressure loss compared with the MVG, the ΔSM of the last stage is 2.69%, and the Δηe is 1.83%. When the combined technology is applied, it shows a significant advantage in delaying the occurrence of a corner stall, and the stall margin of the last stage is improved by 2.71%. Based on a quantitative analysis method using loss sources, the three flow control schemes show significant advantages in reducing the secondary flow loss sources and the wake loss sources. Above all, the BLS shows significant advantages in reducing the total pressure loss, and the COM shows an advantage in expanding the stable operating range. The research results will provide a reference for studies of flow control methods related to reducing flow losses and widening stability margins.

An Edge-Assisted Computing and Mask Attention Based Network for Lung Region Segmentation

Recent years have witnessed the success of encoder-decoder structure-based approaches in lung region segmentation of chest X-ray (CXR) images. However, accurate lung region segmentation is still challenging due to the following three issues: (1) inaccurate lung region segmentation boundaries, (2) existence of lesion-related artifacts (e.g., opacity and pneumonia), and (3) lack of the ability to utilize multiscale information. To address these issues, we propose an edge-assisted computing and mask attention based network (called EAM-Net), which consists of an encoder-decoder network, an edge-assisted computing module, and multiple mask attention modules. Based on the encoder-decoder structure, an edge-assisted computing module is first proposed, which integrates the feature maps of the shallow encoding layers for edge prediction, and uses the edge evidence map as a strong cue to guide the lung region segmentation, thereby refining the lung region segmentation boundaries. We further design a mask attention module after each decoding layer, which employs a mask attention operation to make the model focus on lung regions while suppressing the lesion-related artifacts. Besides, a multiscale aggregation loss is proposed to optimize EAM-Net. Extensive experiments on the JSRT, Shenzhen, and Montgomery datasets demonstrate that EAM-Net outperforms existing state-of-the-art lung region segmentation methods.

TPFR-Net: U-shaped model for lung nodule segmentation based on transformer pooling and dual-attention feature reorganization.

Accurate segmentation of lung nodules is the key to diagnosing the lesion type of lung nodule. The complex boundaries of lung nodules and the visual similarity to surrounding tissues make precise segmentation of lung nodules challenging. Traditional CNN based lung nodule segmentation models focus on extracting local features from neighboring pixels and ignore global contextual information, which is prone to incomplete segmentation of lung nodule boundaries. In the U-shaped encoder-decoder structure, variations of image resolution caused by up-sampling and down-sampling result in the loss of feature information, which reduces the reliability of output features. This paper proposes transformer pooling module and dual-attention feature reorganization module to effectively improve the above two defects. Transformer pooling module innovatively fuses the self-attention layer and pooling layer in the transformer, which compensates for the limitation of convolution operation, reduces the loss of feature information in the pooling process, and decreases the computational complexity of the Transformer significantly. Dual-attention feature reorganization module innovatively employs the dual-attention mechanism of channel and spatial to improve the sub-pixel convolution, minimizing the loss of feature information during up-sampling. In addition, two convolutional modules are proposed in this paper, which together with transformer pooling module form an encoder that can adequately extract local features and global dependencies. We use the fusion loss function and deep supervision strategy in the decoder to train the model. The proposed model has been extensively experimented and evaluated on the LIDC-IDRI dataset, the highest Dice Similarity Coefficient is 91.84 and the highest sensitivity is 92.66, indicating the model's comprehensive capability has surpassed state-of-the-art UTNet. The model proposed in this paper has superior segmentation performance for lung nodules and can provide a more in-depth assessment of lung nodules' shape, size, and other characteristics, which is of important clinical significance and application value to assist physicians in the early diagnosis of lung nodules.

Chronological-hybrid optimization enabled deep learning for boundary segmentation and osteoporosis classification using femur bone

ABSTRACT This paper devises a technique for diagnosing and classifying osteoporosis using the femur bone’s X-ray images. The devised approach uses the following phases: image acquisition, pre-processing, segmentation of femur boundary, measurement of femoral geometry, feature mining, and osteoporosis categorization. Here, femur boundary segmentation is done using the Deep Contour Aware Network (DCAN), which is trained by the Chronological Artificial Ecosystem Optimization (CAEO) algorithm. Then, the femoral geometry is measured and the most relevant features in the image are identified. Further, osteoporosis is classified using the Convolutional Neural Network (CNN) with Transfer Learning (TL), where the CNN is employed with the hyperparameters of the pre-trained model, LeNet. Here, the Chronological Artificial Ecosystem Tasmanian Devil Optimization (CAETDO) algorithm optimizes the classification process. The CAETDO-CNN-TL is examined based on metrics and have the values of 0.910, 0.927, 0.928, and 0.904 for precision, True Negative Rate (TNR), True Positive Rate (TPR), and accuracy, correspondingly.