Fully Convolutional Network Research Articles

Medical Image Segmentation

Medical image segmentation is a critical component in the development of computer-aided diagnosis and treatment planning systems. This paper provides a comprehensive survey of recent advances in segmentation techniques applied to various imaging modalities, including Magnetic Resonance Imaging (MRI). Traditional methods such as thresholding, region-growing, and active contours are reviewed alongside contemporary machine learning-based approaches, particularly deep learning models. The survey emphasizes the growing dominance of convolutional neural networks (CNNs) and their variants, including U-Net and Fully Convolutional Networks (FCNs), which have shown remarkable success in handling complex medical imaging challenges. Additionally, the paper discusses hybrid methods that combine classical techniques with artificial intelligence to improve accuracy and robustness in segmentation tasks. Key challenges such as class imbalance, boundary delineation, and computational efficiency are also highlighted. Future directions, including the integration of multi-modal data and advancements in self-supervised learning, are explored as potential solutions to overcome current limitations in medical image segmentation.

Research on target detection and recognition system of autonomous driving based on deep learning

Abstract. Autonomous driving technology relies heavily on advanced target detection and recognition systems to ensure safety and efficiency. This essay explores how deep learning has revolutionized these systems in four key areas: lane detection, detection of obscured vehicle parts, blind spot detection, and multi-target detection. Traditional methods often struggle under challenging conditions, but deep learning approaches, including Convolutional Neural Networks (CNNs) and Fully Convolutional Networks (FCNs), offer superior performance by analyzing complex features from raw data. Techniques like the Detection of Incomplete Vehicles using Deep Learning and Image Inpainting (DIDA) enhance safety by reconstructing obscured vehicle parts. In blind spot detection, models such as Sep-Res-SE blocks provide an effective, cost-efficient alternative to radar systems. The Adaptive Perceive SSD (AP-SSD) framework improves multi-target detection accuracy and real-time tracking by incorporating advanced feature extraction and temporal analysis. The essay concludes with future research directions aimed at refining real-time capabilities, expanding datasets, and exploring collaborative learning to further enhance autonomous driving technology.

VSegNet - A Variant SegNet for Improving Segmentation Accuracy in Medical Images with Class Imbalance and Limited Data

Deep learning methods for many medical image segmentation task encounter challenges like smaller datasets and class imbalance. This study proposes a variant SegNet (vSegNet) designed to deliver significantly accurate and reliable segmentation results on such datasets. The novelty lies in designing encoder and decoder blocks with an appropriate number of convolution layers and using the Dice score and Hausdorff distance (HD) as compound loss function in learning. This study used public datasets consisting of chest X-rays, axial CT slices, foot ulcer images, and subset of SPIDER dataset to benchmark segmentation task of the proposed neural network model with other popular networks like U-Net, SegNet, DeepLabv3+, VGG16, MobileNetV2, and fully convolutional network (FCN). For the segmentation of lungs in chest X-rays, vertebral body in CT, augmented data for the previous case, foot ulcer dataset, and segmentation of vertebrae, intervertebral disks and spinal canal in SPIDER dataset (MRI dataset) respectively, the proposed vSegNet performed with a Dice score of 0.96 ± 0.01, 0.90 ± 0.20, 0.95 ± 0.02, 0.86 ± 0.07, and 0.95 ± 0.01 and the HD of 14.33 ± 7.74, 8.45 ± 7.08, 7.99 ± 6.05, 29.32 ± 25.64, and 8.45 ± 2.81 with respect to the ground truth on the test dataset. These results highlight the effectiveness of the proposed model in delivering both higher segmentation accuracy and improved boundary delineation. The proposed network, vSegNet has demonstrated as an effective model for semantic segmentation on class-imbalanced smaller datasets, surpassing all other networks considered in this study in terms of mIoU, BF score, Dice score, HD, accuracy, precision, recall and F1 score on a variety of anatomical regions and medical imaging modalities.

Forecasting future earthquakes with deep neural networks: application to California

SUMMARY We use the spatial map of the logarithm of past estimated released earthquake energies as input of fully convolutional networks (FCN) to forecast future earthquakes. This model is applied to California and compared with an elaborated version of the epidemic type aftershock sequence (ETAS) model. Our long-term earthquake forecast simulations show that the FCN model is close to the ETAS model in forecasting earthquakes with $M \ge 3.0,\,\,4.0,\,\,{\rm{and\,\,}}5.0$ according to the Molchan diagram. Moreover, training and implementing the FCN model is 2000–4000 times faster than calibrating the ETAS model and generating its probabilistic forecasts. The FCN model is straightforward in terms of its neural network structure and feature engineering. It does not require extensive knowledge of statistical seismology or the analysis of earthquake catalogue completeness. Using the earthquake catalogue with $M \ge 0$ as FCN input can enhance the model's performance in some time–magnitude forecasting windows.

Lamb Wave TDTE Super-Resolution Imaging Assisted by Deep Learning

Abstract Ultrasonic Lamb waves undergo complex mode conversion and diffraction at non-penetrating defects, such as plate corrosion and cracks. Lamb wave imaging has a resolution limit due to the guided wave dispersion characteristics and Rayleigh criterion limitations. In this paper, a full convolutional network is designed to segment and reconstruct the received signals, enabling the automatic identification of target modalities. This approach eliminates clutter and mode conversion interference when calculating direct and accompanying acoustic fields in time-domain topological energy imaging. Subsequently, the measured accompanying acoustic field is reversed for adaptive focusing on defects and enhance the imaging quality. To circumvent the limitations of the Rayleigh criterion, the direct acoustic field and the accompanying acoustic field were fused to characterize the pixel distribution in the imaging region, achieving Lamb wave super-resolution imaging. Experimental results indicate that compared to the sign coherence factor-total focusing method (SCF-TFM), the proposed method achieves a 31.41% improvement in lateral resolution and a 29.53% increase in signal-to-noise ratio for single-blind-hole defects. In the case of multiple-blind-hole defects with spacings greater than the Rayleigh criterion resolution limit, it exhibits a 27.23% enhancement in signal-to-noise ratio. On the contrary, when the defect spacings are relatively smaller than the limit, this method has a higher resolution limit than SCF-TFM in super-resolution imaging.

Advances in Medical Image Segmentation: A Comprehensive Review of Traditional, Deep Learning and Hybrid Approaches.

Medical image segmentation plays a critical role in accurate diagnosis and treatment planning, enabling precise analysis across a wide range of clinical tasks. This review begins by offering a comprehensive overview of traditional segmentation techniques, including thresholding, edge-based methods, region-based approaches, clustering, and graph-based segmentation. While these methods are computationally efficient and interpretable, they often face significant challenges when applied to complex, noisy, or variable medical images. The central focus of this review is the transformative impact of deep learning on medical image segmentation. We delve into prominent deep learning architectures such as Convolutional Neural Networks (CNNs), Fully Convolutional Networks (FCNs), U-Net, Recurrent Neural Networks (RNNs), Adversarial Networks (GANs), and Autoencoders (AEs). Each architecture is analyzed in terms of its structural foundation and specific application to medical image segmentation, illustrating how these models have enhanced segmentation accuracy across various clinical contexts. Finally, the review examines the integration of deep learning with traditional segmentation methods, addressing the limitations of both approaches. These hybrid strategies offer improved segmentation performance, particularly in challenging scenarios involving weak edges, noise, or inconsistent intensities. By synthesizing recent advancements, this review provides a detailed resource for researchers and practitioners, offering valuable insights into the current landscape and future directions of medical image segmentation.

Sky-GVIO: Enhanced GNSS/INS/Vision Navigation with FCN-Based Sky Segmentation in Urban Canyon

Accurate, continuous, and reliable positioning is critical to achieving autonomous driving. However, in complex urban canyon environments, the vulnerability of stand-alone sensors and non-line-of-sight (NLOS) caused by high buildings, trees, and elevated structures seriously affect positioning results. To address these challenges, a sky-view image segmentation algorithm based on a fully convolutional network (FCN) is proposed for NLOS detection in global navigation satellite systems (GNSSs). Building upon this, a novel NLOS detection and mitigation algorithm (named S−NDM) uses a tightly coupled GNSS, inertial measurement units (IMUs), and a visual feature system called Sky−GVIO with the aim of achieving continuous and accurate positioning in urban canyon environments. Furthermore, the system combines single-point positioning (SPP) with real-time kinematic (RTK) methodologies to bolster its operational versatility and resilience. In urban canyon environments, the positioning performance of the S−NDM algorithm proposed in this paper is evaluated under different tightly coupled SPP−related and RTK−related models. The results exhibit that the Sky−GVIO system achieves meter-level accuracy under the SPP mode and sub-decimeter precision with RTK positioning, surpassing the performance of GNSS/INS/Vision frameworks devoid of S−NDM. Additionally, the sky-view image dataset, inclusive of training and evaluation subsets, has been made publicly accessible for scholarly exploration.

Deep Learning-Based Segmentation and Prediction of Liver Cancer Using CT Images

Background: Liver cancer presents a significant global health challenge, particularly in Asia, due to high incidence rates and the difficulty of early detection. Despite advances in medical imaging, the accurate segmentation and prediction of liver tumors remain critical for effective treatment planning. Deep-learning is a revolutionary tool in medical imaging, offering promising solutions for automated analysis of liver cancer using CT images. Methods: This study develops a framework based on deep-learning- for the segmentation, and prediction cancer of the liver from Computed Tomography images. The framework leverages advanced convolutional neural network (CNN) architectures, including U-Net and its variants, optimized with class balancing methods. The model is trained and authenticated using a high-quality annotated images, dataset of liver CT images. Key techniques such as transfer learning, data augmentation, and hyper parameter tuning are employed to enhance model performance. Evaluation metrics include accuracy, sensitivity, specificity, Dice-coefficient, and (IoU).Intersection over Union Results: The design of proposed program achieved significant improvements in segmentation, accuracy, with an average Dice, Similarity-Coefficient (DSC) of 0.96 for liver segmentation and 0.84 for the tumor segmentation. The results demonstrate the superior performance of the model's compared to the other traditional methods and other existing deep learning models. Cascaded fully convolutional networks (FCNs) combined with 3D Conditional Random Fields (CRFs) showed robust performance, achieving true value accuracy rates of approximately 99.55% for liver segmentation. Conclusion: This study presents a novel and effective deep learning framework for liver cancer segmentation and prediction, highlighting the clinical applicability and relevance of the developed models in real-world medical settings. The findings highlight the potential in deep-learning for enhancing liver cancer diagnosis, facilitating early detection, timely interventions, and improved patient outcomes. This research contributes in increasing the body of knowledge in medical image analysis, enlighten the way for future advancements in the field. Keywords: Computed Tomography, Deep Learning, Machine Learning, U-Net, 3D, Segmentation, Fully Convolutional Networks, Conditional Random Fields, Dice Similarity Coefficient, Intersection over Union, Conventional Neural Network

Partial Discharge Detection Technique Based on Full Convolutional Network with Channel-wise Convolution and Channel Mixing

Abstract The occurrence of partial discharge (PD) serves as an indication of power equipment failure, which, if not given due attention, may result in severe power accidents. Traditionally, the detection of PD involves applying various pre-processing techniques to the original signal and extracting specialized signal features that are subsequently classified using a classifier. The process is intricate and demands extensive expertise. One drawback of the conventional detection approach lies in its failure to consider the information conveyed between the phases of a three-phase electrical signal during the process of identifying PD in medium voltage three-phase cables. The present study proposes a novel method for detecting PDs, which is based on channel-wise convolution and channel mixing(C2CM) using full convolutional networks. Firstly, a set of pulses for each phase of a three-phase signal is extracted and combined to form a three-phase pulse set, serving as the input to the network. Subsequently, we introduce a FCN-C2CM network designed specifically for PD detection, capable of learning both the time-domain features of the input signal and the phase features of the three-phase signal, thereby enhancing recognition accuracy. We evaluated our approach using online testing tools on publicly available datasets, with experimental results demonstrating its superiority over existing.

Fully convolutional networks-based particle distribution analysis at multiphase interfaces

The intricate interplay between droplet dynamics and particle characteristics has led to the extensive use of particle-laden droplets in diverse applications, spanning industrial engineering and scientific research. Accurate particle distribution analysis at multiphase interfaces is crucial for understanding the complex behaviors of these soft matters, for example, liquid marbles (LMs). However, traditional image analysis methods for particle-laden interfaces often rely on manual processing, which is both time-consuming and susceptible to human error. In this study, we present a pixel-level image recognition technique based on fully convolutional networks (FCNs) to automatically analyze the interfacial particle distribution of LMs exhibiting clear core-shell structures. The FCNs-based method enables efficient and accurate segmentation of target particles on sessile LMs, utilizing experimental data obtained from an in-situ optical imaging device. This approach significantly expedites the analysis of particle distribution, including proportion, concentration, and cluster centers. We assess the performance of the proposed method using image data of monolayer LMs encapsulated by dispersed polystyrene microspheres, demonstrating its superiority over traditional techniques in terms of accuracy and generalization. This investigation aims to offer novel insights into determining pertinent particulate characteristics within similar solid-liquid hybrid microsystems.

Overcoming the Barrier of Incompleteness: A Hyperspectral Image Classification Full Model.

Deep learning-based methods have shown promising outcomes in many fields. However, the performance gain is always limited to a large extent in classifying hyperspectral image (HSI). We discover that the reason behind this phenomenon lies in the incomplete classification of HSI, i.e., existing works only focus on a certain stage that contributes to the classification, while ignoring other equally or even more significant phases. To address the above issue, we creatively put forward three elements needed for complete classification: the extensive exploration of available features, adequate reuse of representative features, and differential fusion of multidomain features. To the best of our knowledge, these three elements are being established for the first time, providing a fresh perspective on designing HSI-tailored models. On this basis, an HSI classification full model (HSIC-FM) is proposed to overcome the barrier of incompleteness. Specifically, a recurrent transformer corresponding to Element 1 is presented to comprehensively extract short-term details and long-term semantics for local-to-global geographical representation. Afterward, a feature reuse strategy matching Element 2 is designed to sufficiently recycle valuable information aimed at refined classification using few annotations. Eventually, a discriminant optimization is formulized in accordance with Element 3 to distinctly integrate multidomain features for the purpose of constraining the contribution of different domains. Numerous experiments on four datasets at small-, medium-, and large-scale demonstrate that the proposed method outperforms the state-of-the-art (SOTA) methods, such as convolutional neural network (CNN)-, fully convolutional network (FCN)-, recurrent neural network (RNN)-, graph convolutional network (GCN)-, and transformer-based models (e.g., accuracy improvement of more than 9% with only five training samples per class). The code will be available soon at https://github.com/jqyang22/ HSIC-FM.

MLFA-UNet: A multi-level feature assembly UNet for medical image segmentation

Medical image segmentation is crucial for accurate diagnosis and treatment in medical image analysis. Among the various methods employed, fully convolutional networks (FCNs) have emerged as a prominent approach for segmenting medical images. Notably, the U-Net architecture and its variants have gained widespread adoption in this domain. This paper introduces MLFA-UNet, an innovative architectural framework aimed at advancing medical image segmentation. MLFA-UNet adopts a U-shaped architecture and integrates two pivotal modules: multi-level feature assembly (MLFA) and multi-scale information attention (MSIA), complemented by a pixel-vanishing (PV) attention mechanism. These modules synergistically contribute to the segmentation process enhancement, fostering both robustness and segmentation precision. MLFA operates within both the network encoder and decoder, facilitating the extraction of local information crucial for accurately segmenting lesions. Furthermore, the bottleneck MSIA module serves to replace stacking modules, thereby expanding the receptive field and augmenting feature diversity, fortified by the PV attention mechanism. These integrated mechanisms work together to boost segmentation performance by effectively capturing both detailed local features and a broader range of contextual information, enhancing both accuracy and resilience in identifying lesions. To assess the versatility of the network, we conducted evaluations of MFLA-UNet across a range of medical image segmentation datasets, encompassing diverse imaging modalities such as wireless capsule endoscopy (WCE), colonoscopy, and dermoscopic images. Our results consistently demonstrate that MFLA-UNet outperforms state-of-the-art algorithms, achieving dice coefficients of 91.42%, 82.43%, 90.8%, and 88.68% for the MICCAI 2017 (Red Lesion), ISIC 2017, PH2, and CVC-ClinicalDB datasets, respectively.

Arc bubble edge detection method based on deep transfer learning in underwater wet welding.

The stability of arc bubble is a crucial indicator of underwater wet welding process. However, limited research exists on detecting arc bubble edges in such environments, and traditional algorithms often produce blurry and discontinuous results. To address these challenges, we propose a novel arc bubble edge detection method based on deep transfer learning for processing underwater wet welding images. The proposed method integrates two training stages: pre-training and fine-tuning. In the pre-training stage, a large source domain dataset is used to train VGG16 as a feature extractor. In the fine-tuning stage, we introduce the Attention-Scale-Semantics (ASS) model, which consists of a Convolutional Block Attention Module (CBAM), a Scale Fusion Module (SCM) and a Semantic Fusion Module (SEM). The ASS model is further trained on a small target domain dataset specific to underwater wet welding to fine-tune the model parameters. The CBAM can adaptively weight the feature maps, focusing on more crucial features to better capture edge information. The SCM training method maximizes feature utilization and simplifies training by combining multi-scale features. Additionally, the skip structure of SEM effectively mitigates semantic loss in the high-level network, enhancing the accuracy of edge detection. On the BSDS500 dataset and a self-constructed underwater wet welding dataset, the ASS model was evaluated against conventional edge detection models-Richer Convolutional Features (RCF), Fully Convolutional Network (FCN), and UNet-as well as state-of-the-art models LDC and TEED. In terms of Mean Absolute Error (MAE), accuracy, and other evaluation metrics, the ASS model consistently outperforms these models, demonstrating edge detection capabilities that are both effective and stable in detecting arc bubble edges in underwater wet welding images.

The Data Analysis of Enterprise Operational Risk Prediction Under Machine Learning

In the digital age, the financial sector faces increasingly severe risk management challenges. Traditional methods often rely on historical data and statistical models, which struggle to cope with the high volatility of the market. These methods exhibit poor adaptability in rapidly changing financial markets and often fail to meet demands in terms of accuracy and reliability. To address these issues, this study proposes a law risk prediction model based on deep learning—the WBIF model. This model integrates Bidirectional Long Short-Term Memory (BiLSTM) and Fully Convolutional Networks (FCN) and employs the Whale Optimization Algorithm (WOA) for parameter optimization. Experimental results show that compared to traditional models, the WBIF model reduces the Mean Absolute Error (MAE) by 51.73% on the UCI machine learning library dataset and improves accuracy by 12% on the Kaggle credit card fraud detection dataset.

Deep neural networks for automated damage classification in image-based visual data of reinforced concrete structures

Significant strides in deep learning for image recognition have expanded the potential of visual data in assessing damage to reinforced concrete (RC) structures. Our study proposes an automated technique, merging convolutional neural networks (CNNs) and fully convolutional networks (FCNs), to detect, classify, and segment building damage. These deep networks extract RC damage-related features from high-resolution smartphone images (3264 × 2448 pixels), categorized into two groups: damage (exposed reinforcement and spalled concrete) and undamaged area. With a labeled dataset of 2000 images, fine-tuning of network architecture and hyperparameters ensures effective training and testing. Remarkably, we achieve 98.75 % accuracy in damage classification and 95.98 % in segmentation, without overfitting. Both CNNs and FCNs play crucial roles in extracting features, showcasing the adaptability of deep learning. Our promising results validate the potential of these techniques for inspectors, providing an effective means to assess the severity of identified damage in image-based evaluations.

A Light-Weight Self-Supervised Infrared Image Perception Enhancement Method

Convolutional Neural Networks (CNNs) have achieved remarkable results in the field of infrared image enhancement. However, the research on the visual perception mechanism and the objective evaluation indicators for enhanced infrared images is still not in-depth enough. To make the subjective and objective evaluation more consistent, this paper uses a perceptual metric to evaluate the enhancement effect of infrared images. The perceptual metric mimics the early conversion process of the human visual system and uses the normalized Laplacian pyramid distance (NLPD) between the enhanced image and the original scene radiance to evaluate the image enhancement effect. Based on this, this paper designs an infrared image-enhancement algorithm that is more conducive to human visual perception. The algorithm uses a lightweight Fully Convolutional Network (FCN), with NLPD as the similarity measure, and trains the network in a self-supervised manner by minimizing the NLPD between the enhanced image and the original scene radiance to achieve infrared image enhancement. The experimental results show that the infrared image enhancement method in this paper outperforms existing methods in terms of visual perception quality, and due to the use of a lightweight network, it is also the fastest enhancement method currently.

A model-based deep learning framework for damage classification and detection in polycarbonate infused with AEROSIL under dynamic loading conditions

Composite 3D printing is a significant engineering application owing to its robustness, ability to achieve complex geometries, and ease of use. Polycarbonate, particularly when infused with AEROSIL, is an interesting thermoplastic and potential candidate for 3D printing with enhanced properties. The primary objective of this research is to develop a new model-based deep-learning framework to classify and detect damage in this material under dynamic loading conditions. To achieve this, a FASTCAM high-speed camera was placed in front of the SHPB test setup to capture dynamic damage. The test results were then used as label inputs for training the advanced deep learning algorithms, focusing on dense image recognition techniques for detailed damage analysis. The study involved a series of fully convolutional networks (FCNs), evaluating semantic segmentation with U-Net and instance segmentation with state-of-the-art frameworks such as YOLOv8 and Mask R–CNN. A comparative analysis revealed that deep learning models outperform traditional methods, providing efficient and accurate damage classification and detection. The U-Net model demonstrated the ability to recognize cubes and bars but was limited in detecting minor damage regardless of size. YOLO-V8, which specializes in case segmentation, achieved remarkable performance in detecting significant damage but struggled to accurately identify minor damage. By leveraging deep learning techniques, this study enables an efficient and accurate damage assessment, which is crucial for ensuring the reliability and safety of composite structures in various industries.

Advanced Global Prototypical Segmentation Framework for Few-Shot Hyperspectral Image Classification.

With the advancement of deep learning, related networks have shown strong performance for Hyperspectral Image (HSI) classification. However, these methods face two main challenges in HSI classification: (1) the inability to capture global information of HSI due to the restriction of patch input and (2) insufficient utilization of information from limited labeled samples. To overcome these challenges, we propose an Advanced Global Prototypical Segmentation (AGPS) framework. Within the AGPS framework, we design a patch-free feature extractor segmentation network (SegNet) based on a fully convolutional network (FCN), which processes the entire HSI to capture global information. To enrich the global information extracted by SegNet, we propose a Fusion of Lateral Connection (FLC) structure that fuses the low-level detailed features of the encoder output with the high-level features of the decoder output. Additionally, we propose an Atrous Spatial Pyramid Pooling-Position Attention (ASPP-PA) module to capture multi-scale spatial positional information. Finally, to explore more valuable information from limited labeled samples, we propose an advanced global prototypical representation learning strategy. Building upon the dual constraints of the global prototypical representation learning strategy, we introduce supervised contrastive learning (CL), which optimizes our network with three different constraints. The experimental results of three public datasets demonstrate that our method outperforms the existing state-of-the-art methods.

AN INTELLIGENT MODEL FOR BENIGN AND MALIGNANT PULMONARY NODULE ANALYSIS USING U-NET NETWORKS AND MULTILEVEL ATTENTION MECHANISMS

A key study area throughout the medical sector for image processing and analysis is medical image segmentation. The diagnosis and treatment strategies of doctors may have a solid foundation owing to accurate and effective medical image segmentation. Conventional approaches in this field rely on manual feature extraction, which makes segmentation complex, costs doctors’ time and energy, and involves a subjective evaluation that is readily susceptible to diagnostic errors. Researchers have applied convolutional neural network-based deep learning techniques to the segmentation of medical images as a result of their impressive advancements and successes in the field of computer vision. The research described here uses the U-Net network’s outstanding feature learning capabilities and end-to-end processing mode for lung CT image segmentation via fully convolutional network (FCN) research. However, focusing on valuable, crucial information aspects in the U-Net network is challenging. This study employed multilevel attention mechanisms on the basis of U-Net networks to enhance the model’s accuracy in lung CT image segmentation. These mechanisms were inspired by attention mechanisms. By improving the segmentation accuracy and optimizing the segmentation effect, the new model embeds a self-attention module in front of each upsampling layer in the U-Net model. This module provides more detailed information by stitching the self-attention module of the original image and then suppresses irrelevant and redundant information by using the effect of feature extraction of the upsampling layer. Several additional comparative experiments were conducted on the 2019nCoVR dataset. The outcomes demonstrate the efficacy of the optimized model described in this paper and its application results in improved segmentation effects in lung CT images. Additionally, the new model has distinct advantages over existing approaches that are typical of medical image segmentation, which represent its higher level of lung CT image segmentation.

A Multi-Stage Automatic Method Based on a Combination of Fully Convolutional Networks for Cardiac Segmentation in Short-Axis MRI

Magnetic resonance imaging (MRI) is a non-invasive technique used in cardiac diagnosis. Using it, specialists can measure the masses and volumes of the right ventricle (RV), left ventricular cavity (LVC), and myocardium (MYO). Segmenting these structures is an important step before this measurement. However, this process can be laborious and error-prone when done manually. This paper proposes a multi-stage method for cardiac segmentation in short-axis MRI based on fully convolutional networks (FCNs). This automatic method comprises three main stages: (1) the extraction of a region of interest (ROI); (2) MYO and LVC segmentation using a proposed FCN called EAIS-Net; and (3) the RV segmentation using another proposed FCN called IRAX-Net. The proposed method was tested with the ACDC and M&Ms datasets. The main evaluation metrics are end-diastolic (ED) and end-systolic (ES) Dice. For the ACDC dataset, the Dice results (ED and ES, respectively) are 0.960 and 0.904 for the LVC, 0.880 and 0.892 for the MYO, and 0.910 and 0.860 for the RV. For the M&Ms dataset, the ED and ES Dices are 0.861 and 0.805 for the LVC, 0.733 and 0.759 for the MYO, and 0.721 and 0.694 for the RV. These results confirm the feasibility of the proposed method.