Residual Block Research Articles

Residual trio feature network for efficient super-resolution

Deep learning-based approaches have demonstrated impressive performance in single-image super-resolution (SISR). Efficient super-resolution compromises the reconstructed image’s quality to have fewer parameters and Flops. Ensured efficiency in image reconstruction and improved reconstruction quality of the model are significant challenges. This paper proposes a trio branch module (TBM) based on structural reparameterization. TBM achieves equivalence transformation through structural reparameterization operations, which use a complex network structure in the training phase and convert it to a more lightweight structure in the inference, achieving efficient inference while maintaining accuracy. Based on the TBM, we further design a lightweight version of the enhanced spatial attention mini (ESA-mini) and the residual trio feature block (RTFB). Moreover, the multiple RTFBs are combined to construct the residual trio network (RTFN). Finally, we introduce a localized contrast loss for better applicability to the super-resolution task, which enhances the reconstruction quality of the super-resolution model. Experiments show that the RTFN framework proposed in this paper outperforms other state-of-the-art efficient super-resolution methods in terms of inference speed and reconstruction quality.

Automatic assessment of nailfold capillaroscopy software: a pilot study

IntroductionCapillaroscopy is a simple method of nailfold capillary imaging, used to diagnose diseases from the systemic sclerosis spectrum. However, the assessment of the capillary image is time-consuming and subjective. This makes it difficult to use for a detailed comparison of studies assessed by various physicians. This pilot study aimed to validate software used for automatic capil­lary counting and image classification as normal or pathological.Material and methodsThe study was based on the assessment of 200 capillaroscopic images obtained from patients suffering from systemic sclerosis or scleroderma spectrum diseases and healthy people. Dinolite MEDL4N Pro was used to perform capillaroscopy. Each image was analysed manually and described using working software. The neural network was trained using the fast.ai library (based on PyTorch). The ResNet-34 deep residual neural network was chosen; 10-fold cross-validation with the validation and test set was performed, using the Darknet-YoloV3 state of the art neural network in a GPU-optimized (P5000 GPU) environment. For the calculation of 1 mm capillaries, an additional detection mechanism was designed.ResultsThe results obtained under neural network training were compared to the results obtained in manual analysis. The sensitivity of the automatic tool relative to manual assessment in classification of correct vs. pathological images was 89.0%, specificity 89.4% for the training group, in validation 89.0% and 86.9% respectively. For the average number of capillaries in 1 mm the precision of real images detected within the region of interest was 96.48%.ConclusionsThe pilot software for fully automatic capillaroscopic image assessment can be a useful tool for the rapid classification of a normal and altered capillaroscopy pattern. In addition, it allows one to quickly calculate the number of capillaries. In the future, the tool will be developed and will make it possible to obtain full imaging characteristics independent of the experience of the examiner.

Deep learning based apparent diffusion coefficient map generation from multi‐parametric MR images for patients with diffuse gliomas

AbstractPurposeApparent diffusion coefficient (ADC) maps derived from diffusion weighted magnetic resonance imaging (DWI MRI) provides functional measurements about the water molecules in tissues. However, DWI is time consuming and very susceptible to image artifacts, leading to inaccurate ADC measurements. This study aims to develop a deep learning framework to synthesize ADC maps from multi‐parametric MR images.MethodsWe proposed the multiparametric residual vision transformer model (MPR‐ViT) that leverages the long‐range context of vision transformer (ViT) layers along with the precision of convolutional operators. Residual blocks throughout the network significantly increasing the representational power of the model. The MPR‐ViT model was applied to T1w and T2‐fluid attenuated inversion recovery images of 501 glioma cases from a publicly available dataset including preprocessed ADC maps. Selected patients were divided into training (N = 400), validation (N = 50), and test (N = 51) sets, respectively. Using the preprocessed ADC maps as ground truth, model performance was evaluated and compared against the Vision Convolutional Transformer (VCT) and residual vision transformer (ResViT) models with the peak signal‐to‐noise ratio (PSNR), structural similarity index measure (SSIM), and mean squared error (MSE).ResultsThe results are as follows using T1w + T2‐FLAIR MRI as inputs: MPR‐ViT—PSNR: 31.0 ± 2.1, MSE: 0.009 ± 0.0005, SSIM: 0.950 ± 0.015. In addition, ablation studies showed the relative impact on performance of each input sequence. Both qualitative and quantitative results indicate that the proposed MR‐ViT model performs favorably against the ground truth data.ConclusionWe show that high‐quality ADC maps can be synthesized from structural MRI using a MPR‐ViT model. Our predicted images show better conformality to the ground truth volume than ResViT and VCT predictions. These high‐quality synthetic ADC maps would be particularly useful for disease diagnosis and intervention, especially when ADC maps have artifacts or are unavailable.

U-net with ResNet-34 backbone for dual-polarized C-band baltic sea-ice SAR segmentation

Abstract In this study, the U-net with ResNet-34, i.e. a residual neural network with 34 layers, backbone semantic segmentation network is applied to C-band sea-ice SAR imagery over the Baltic Sea. Sentinel-1 Extra Wide Swath mode HH/HV-polarized SAR data acquired during the winter season 2018–2019, and corresponding segments derived from the daily Baltic Sea ice charts were used for training the segmentation algorithm. C-band SAR image mosaics of the winter season 2020–2021 were then used to evaluate the segmentation. The major objective was to study the suitability of semantic segmentation of SAR imagery for automated SAR segmentation and also to find a potential replacement for the outdated iterated conditional modes (ICM) algorithm currently in operational use. The results compared to the daily Baltic Sea ice charts and the operational ICM segmentation and visual interpretation were encouraging from the operational point of view. Open water areas were located very well and the oversegmentation produced by ICM was significantly reduced. The correspondence between the ice chart polygons and the segmentation results was also reasonably good. Based on the results, the studied method is a potential candidate to replace the operational ICM SAR segmentation used in the Copernicus Marine Service automated sea-ice products at Finnish Meteorological Institute.

CrackNet: A Hybrid Model for Crack Segmentation with Dynamic Loss Function

Cracks are a common form of damage in infrastructure, posing significant risks to both personal safety and property. Along with the development of deep learning, visual-based crack automatic detection has been widely studied. However, this task is still challenging due to complex crack topology, noisy backgrounds, unbalanced categories, etc. To address these challenges, this research proposes a novel hybrid network, named CrackNet, which leverages the strengths of both CNN and transformer. On the encoder side, CNNs are employed to extract multi-level local features, while transformers are used to model global dependencies. Additionally, a strip pooling module is introduced to suppress irrelevant regions and enhance the network’s ability to segment narrow and elongated cracks. On the decoder side, an attention-based skip connection strategy and a mixed up-sampling module are implemented to restore detailed information. Furthermore, a joint learning loss combining Dice and cross-entropy with dynamic weighting is proposed to mitigate the effects of severe class imbalance. CrackNet is trained and evaluated on three public crack datasets, and experimental results show that the proposed model outperforms several well-known deep neural networks, with a particularly noticeable improvement in recall rate.

MFHARFNet: multi-branch feature hybrid and adaptive receptive field network for image segmentation

Abstract Accurate segmentation of medical images is crucial for disease diagnosis and understanding disease changes. Deep learning methods, utilizing encoder-decoder structures, have demonstrated cutting-edge performance in various medical image segmentation tasks. However, the pooling operation in the encoding stage results in feature loss, which makes the network lack the ability to fuse multi-scale information at different levels, hinders its effective perception of multi-scale information, and leads to poor segmentation performance. Drawing inspiration from the U-shaped network, this study introduces a multi-branch feature hybrid attention and adaptive receptive field network (MFHARFNet) for medical image segmentation. Building upon the encoder-decoder framework, we initially devise a multi-branch feature hybrid attention module (MFHAM) to seamlessly integrate feature maps of varying scales, capturing both fine-grained features and coarse-grained semantics across the entire scale. Furthermore, we redesign the skip connection to amalgamate feature information from different branches in the encoder stage and efficiently transmit it to the decoder, providing the decoder with global context feature maps at different levels. Finally, the adaptive receptive field (ARF) module is introduced in the decoder feature reconstruction stage to adapt and focus on related fields, ensuring the model’s adaptation to different segmentation target features, and achieving different weights for the output of different convolution kernels to improve segmentation performance. We comprehensively evaluate our method on medical image segmentation tasks, by using four public datasets across CT and MRI. Remarkably, MFHARFNet method consistently outperforms other state-of-the-art methods, exceeding UNet by 2.1%, 0.9%, 6.6% and 1.0% on Dice on ATLAS, LiTs, BraTs2019 and Spine and intervertebral disc datasets, respectively. In addition, MFHARFNet minimizes network parameters and computational complexity as much as possible. The source codes are in https://github.com/OneHundred99/MFHARFNet.

REDalign: accurate RNA structural alignment using residual encoder-decoder network

BackgroundRNA secondary structural alignment serves as a foundational procedure in identifying conserved structural motifs among RNA sequences, crucially advancing our understanding of novel RNAs via comparative genomic analysis. While various computational strategies for RNA structural alignment exist, they often come with high computational complexity. Specifically, when addressing a set of RNAs with unknown structures, the task of simultaneously predicting their consensus secondary structure and determining the optimal sequence alignment requires an overwhelming computational effort of O(L6)\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$O(L^6)$$\\end{document} for each RNA pair. Such an extremely high computational complexity makes these methods impractical for large-scale analysis despite their accurate alignment capabilities.ResultsIn this paper, we introduce REDalign, an innovative approach based on deep learning for RNA secondary structural alignment. By utilizing a residual encoder-decoder network, REDalign can efficiently capture consensus structures and optimize structural alignments. In this learning model, the encoder network leverages a hierarchical pyramid to assimilate high-level structural features. Concurrently, the decoder network, enhanced with residual skip connections, integrates multi-level encoded features to learn detailed feature hierarchies with fewer parameter sets. REDalign significantly reduces computational complexity compared to Sankoff-style algorithms and effectively handles non-nested structures, including pseudoknots, which are challenging for traditional alignment methods. Extensive evaluations demonstrate that REDalign provides superior accuracy and substantial computational efficiency.ConclusionREDalign presents a significant advancement in RNA secondary structural alignment, balancing high alignment accuracy with lower computational demands. Its ability to handle complex RNA structures, including pseudoknots, makes it an effective tool for large-scale RNA analysis, with potential implications for accelerating discoveries in RNA research and comparative genomics.

Multivariate variational mode decomposition and 1D residual neural network for subtle feature recognition of rolling bearings

Multivariate variational mode decomposition and 1D residual neural network for subtle feature recognition of rolling bearings

Coupling design features of material surface treatment for ceramic products based on ResNet

Abstract Ceramic products is one of the important carriers of various civilizations, reflecting the lifestyle, aesthetic concepts, and technological level of society at that time. In order to study the surface treatment design features of ceramic craft products, this article analyzed the ceramic features through computer vision technology and used residual neural networks to detect the surface treatment features of ceramic craft products. The extracted texture features were classified to study and analyze the coupling features of different glazes, colors, and shapes on the formation of different textures. This study used ResNeXt50-SSD, which combined ResNeXt50 and SSD (Single Shot MultiBox Detector) algorithms, to compare feature detection with LeNet-5, VGG-16, and MobileNetV2 network models. From the experimental findings, it can be concluded that ResNeXt50-SSD was the most effective for feature recognition of ceramic craft products, with precision, recall, and mAP of 94.3, 92.1, and 89.5%, respectively. Therefore, the combination of ResNeXt50 and SSD algorithms is an effective method for detecting surface treatment features of ceramic craft products.

Documenting customary land boundaries using unmanned aerial vehicle imagery and artificial intelligence

AbstractThe use of computer vision and deep learning in boundary documentation for land registration stems from the ongoing demand for appropriate mapping approaches of unregistered land rights to eradicate the global challenge of tenure insecurity. Previous research has yielded promising results towards automated extraction of photo‐visible cadastral boundaries from high‐resolution imagery. Nonetheless, the extraction of invisible cadastral boundaries is still a challenge. This study investigates the place of sensor/s on‐board unmanned aerial vehicles and deep learning algorithms in detecting cadastral boundaries. It develops a participatory boundary marking procedure using low‐cost markers to bring monument to previously invisible and ill‐defined cadastral boundaries. After that, the researchers trained and tested the accuracy of a convolutional neural network, namely single shot multi‐box detector (SSD) based on Residual Neural Network (ResNet) and Visual Geometry Group (VGG) backbone networks to automatically detect cadastral boundary markers from unmanned aerial vehicle imagery. SSD based on ResNet34 performed best with 0.88 precision, 0.92 recall and 0.91 F measure or (F1) score. VGG19‐based SSD yielded a precision of 0.47, recall of 0.53 and F1 score of 0.50. The horizontal accuracy of the cadastral map generated varied from 0.089 to 0.496 m per parcel, with a standard deviation of 0.120 m. Results show that this approach is practical for cadastral mapping in rural areas.

Channel reconstruction through improvised deep learning architecture for high-speed networks

Efficient acquisition of channel state information (CSI) is quite complicated process but immensely essential to exploit probable benefits of massive multiple input multiple output (MIMO) systems. Therefore, a deep learningbased model is utilized to estimate channel feedback in a massive MIMO system. The proposed improvised deep learning-based channel estimation (IDLCE) model enhances channel reconstruction efficiency by using multiple convolutional layers and residual blocks. The proposed IDLCE model utilizes encoder network to compress CSI matrices where decoder network is used to downlink reconstruct CSI matrices. Here, an additional quantization block is incorporated to improve feedback reconstruction accuracy by reducing channel errors. A COST 2,100 model is adopted to analyse performance efficiency for both indoor and outdoor scenarios. Further, deep learning-based model is used to train thousands of parameter and correlation coefficients much faster and to minimize computational complexity. The proposed IDLCE model evaluate performance in terms of normalized mean square error (NMSE), correlation efficiency and reconstruction accuracy and compared against varied state-of-art-channel estimation techniques. Excellent performance results are obtained with large improvement in channel reconstruction accuracy

MSSTGNN: Multi-Scaled Spatio-Temporal Graph Neural Networks for short- and long-term traffic prediction

Accurate traffic prediction plays a crucial role in ensuring traffic safety and minimizing property damage. The utilization of STGNNs in traffic prediction has gained significant attention from researchers aiming to capture the intricate time-varying relationships within traffic data. While existing STGNNs commonly rely on Euclidean distance to assess the similarity between nodes, which may fall short in reflecting POI or regional functions. The traffic network exhibits static from a macro perspective, whereas undergoes dynamic changes in the micro perspective. Previous researchers incorporating self-attention to capture time-varying features for constructing dynamic graphs have faced challenges in overlooking the connections between nodes due to the Softmax polarization effect, which tends to amplify extreme value differences, and fails to accurately represent the true relationships between nodes. To solve this problem, we introduce the Multi-Scaled Spatio-Temporal Graph Neural Networks (MSSTGNN), which aims to comprehensively capture characteristics within traffic from multiscale viewpoints to construct multi-perspective graphs. We employ a trainable matrix to enhance the predefined adjacency matrices and to construct an optimal dynamic graph based on both trend and period. Additionally, a graph aggregate technique is proposed to effectively merge trend and periodic dynamic graphs. TCN is developed to model nonstationary traffic data, and we leverage the skip and residual connections to increase the model depth. A two-stage learning approach and a novel MSELoss function is designed to enhance the model's performance. The experimental results demonstrate that the MSSTGNN model outperforms the existing methods, achieving state-of-the-art performances across multiple real-world datasets.

Combining residual convolutional LSTM with attention mechanisms for spatiotemporal forest cover prediction

Understanding spatiotemporal variations in forest cover is crucial for effective forest resource management. However, existing models often lack accuracy in simultaneously capturing temporal continuity and spatial correlation. To address this challenge, we developed ResConvLSTM-Att, a novel hybrid model integrating residual neural networks, Convolutional Long Short-Term Memory (ConvLSTM) networks, and attention mechanisms. We evaluated ResConvLSTM-Att against four deep learning models: LSTM, combined convolutional neural network and LSTM (CNN-LSTM), ConvLSTM, and ResConvLSTM for spatiotemporal prediction of forest cover in Tasmania, Australia. ResConvLSTM-Att achieved outstanding prediction performance, with an average root mean square error (RMSE) of 6.9% coverage and an impressive average coefficient of determination of 0.965. Compared with LSTM, CNN-LSTM, ConvLSTM, and ResConvLSTM, ResConvLSTM-Att achieved RMSE reductions of 31.2%, 43.0%, 10.1%, and 6.5%, respectively. Additionally, we quantified the impacts of explanatory variables on forest cover dynamics. Our work demonstrated the effectiveness of ResConvLSTM-Att in spatiotemporal data modelling and prediction.

MDDU-Net: A multi-scale dense connectivity hybrid dilated convolutional U-Net for segmentation in prostate ultrasound images

Ultrasound is one of the most commonly used imaging tools in prostate biopsy and brachytherapy. However, due to issues such as blurred image boundaries, similar intensity distributions, and noise, accurate segmentation of the prostate gland has been challenged. We propose a multi-scale dense connectivity hybrid dilated convolutional U-Net (MDDU-Net) network model segmentation of 2D prostate ultrasound images to improve the segmentation accuracy. Our proposed hybrid dilated convolution (HDC) module consists of standard and dilated convolution, which can expand the receptive field of the network, focus on global information, capture more features, and improve the learning ability of the model. Introducing dense connectivity (Dense) maximizes the flow of information between layers, retains more details, and helps mitigate gradient vanishing. The output features of each convolution layer in the bottleneck and decoder are fused using a multi-scale fusion block (MSFB) to obtain a feature map with richer semantic and detailed information. A dense skip connection provides more raw information to the decoder. To verify the superiority of our proposed method, we performed a 5-fold cross-validation, quantitatively analyzed it using five commonly used medical image evaluation metrics, and compared it with several state-of-the-art segmentation methods. The results show that our method performs better in prostate ultrasound image segmentation and can achieve more accurate prostate segmentation.

Simultaneous single image super‐resolution and blind Gaussian denoising via slim ghost full‐frequency residual blocks

AbstractGiven that super‐resolution (SR) aims to recover lost information, and low‐resolution (LR) images in real‐world conditions might be corrupted with multiple degradations, considering basic bicubic down‐sampling as the sole degradation significantly limits the performance of most existing SR models. This paper presents a model for simultaneous super‐resolution and blind additive white Gaussian noise (AWGN) denoising with two components (netdeg and netSR) that is based on a generative adversarial network (GAN) to achieve detailed results. netdeg, featuring residual and innovative cost‐effective ghost residual blocks with a frequency separation module for obtaining long‐range information, blindly restores a clean version of the LR image. netSR leverages slim ghost full‐frequency residual blocks to process low‐frequency (LF) and high‐frequency (HF) information via static large convolutions and pixel‐wise highlighted input‐adaptive dynamic convolutions, respectively. To address the susceptibility of dynamic layers to noise and preserve feature diversity while reducing model’s costs, static and dynamic layer features are combined and highlighted. Diverse and non‐redundant features are then processed using ghost‐style blocks. The proposed model achieves comparable SR results in bicubic down‐sampling scenarios, outperform existing SR methods in the complex task of concurrent SR and AWGN denoising, and demonstrate robustness in handling images corrupted with varying levels of AWGN.

A Strip Steel Surface Defect Salient Object Detection Based on Channel, Spatial and Self-Attention Mechanisms

Strip steel is extensively utilized in industries such as automotive manufacturing and aerospace due to its superior machinability, economic benefits, and adaptability. However, defects on the surface of steel strips, such as inclusions, patches, and scratches, significantly affect the performance and service life of the product. Therefore, the salient object detection of surface defects on strip steel is crucial to ensure the quality of the final product. Many factors, such as the low contrast of surface defects on strip steel, the diversity of defect types, complex texture structures, and irregular defect distribution, hinder existing detection technologies from accurately identifying and segmenting defect areas against complex backgrounds. To address the above problems, we propose a novel detector called S3D-SOD for the salient object detection of strip steel surface defects. For the encoding stage, a residual self-attention block is proposed to explore semantic information cues of high-level features to locate and guide low-level feature information. In addition, we apply a general residual channel and spatial attention to low-level features, enabling the model to adaptively focus on the key channels and spatial areas of feature maps with high resolutions, thereby enhancing the encoder features and accelerating the convergence of the model. For the decoding stage, a simple residual decoder block with an upsampling operation is proposed to realize the integration and interaction of feature information between different layers. Here, the simple residual decoder block is used for feature integration due to the following observation: backbone networks like ResNet and the Swin Transformer, after being pretrained on the large dataset ImageNet and then fine-tuned on a smaller dataset for strip steel surface defects, are capable of extracting feature maps that contain both general image features and the specific characteristics required for the salient object detection of strip steel surface defects. The experimental results on the SD-saliency-900 dataset show that S3D-SOD is better than advanced methods, and it has strong generalization ability and robustness.

HepNet: Deep Neural Network for Classification of Early-Stage Hepatic Steatosis Using Microwave Signals.

Hepatic steatosis, a key factor in chronic liver diseases, is difficult to diagnose early. This study introduces a classifier for hepatic steatosis using microwave technology, validated through clinical trials. Our method uses microwave signals and deep learning to improve detection to reliable results. It includes a pipeline with simulation data, a new deep-learning model called HepNet, and transfer learning. The simulation data, created with 3D electromagnetic tools, is used for training and evaluating the model. HepNet uses skip connections in convolutional layers and two fully connected layers for better feature extraction and generalization. Calibration and uncertainty assessments ensure the model's robustness. Our simulation achieved an F1-score of 0.91 and a confidence level of 0.97 for classifications with entropy ≤0.1, outperforming traditional models like LeNet (0.81) and ResNet (0.87). We also use transfer learning to adapt HepNet to clinical data with limited patient samples. Using 1H-MRS as the standard for two microwave liver scanners, HepNet achieved high F1-scores of 0.95 and 0.88 for 94 and 158 patient samples, respectively, showing its clinical potential.

Effective encoder-decoder network for pupil light reflex segmentation in facial photographs of ptosis patients

Accurate segmentation of pupil light reflexes is essential for the reliable assessment of ptosis severity, a condition characterized by the drooping of the upper eyelid. This study introduces a novel encoder-decoder network specialized in reflex segmentation by focusing on addressing issues related to very small regions of interest from an architectural perspective. Specifically, the proposed network is designed to exploit low-level features effectively by integrating a multi-level skip connection and a 1 × 1 convolution-enhanced initial encoding stage. Assessed using a photograph image dataset from Chung-Ang University Hospital, which includes 87 healthy subjects, 64 with ptosis, and 257 with Graves’ orbitopathy (collected between January 2010 and February 2023), the proposed network outperforms five conventional encoder-decoders. Over 30 trials, the proposed network achieved a mean Dice coefficient of 0.767 and an Intersection over Union of 0.653, indicating a statistically significant improvement in the segmentation of reflex. Our findings show that an elaborate design based on the lowest-level skip connection and 1 × 1 convolution at initial stage enhances the segmentation of pupil light reflexes. The source code of the proposed network is available at https://github.com/tkdgur658/ReflexNet.

A Multi-Scale Feature Fusion Deep Learning Network for the Extraction of Cropland Based on Landsat Data

In the face of global population growth and climate change, the protection and rational utilization of cropland are crucial for food security and ecological balance. However, the complex topography and unique ecological environment of the Qinghai-Tibet Plateau results in a lack of high-precision cropland monitoring data. Therefore, this paper constructs a high-quality cropland dataset for the YarlungZangbo-Lhasa-Nyangqv River region of the Qinghai-Tibet Plateau and proposes an MSC-ResUNet model for cropland extraction based on Landsat data. The dataset is annotated at the pixel level, comprising 61 Landsat 8 images in 2023. The MSC-ResUNet model innovatively combines multiscale features through residual connections and multiscale skip connections, effectively capturing features ranging from low-level spatial details to high-level semantic information and further enhances performance by incorporating depthwise separable convolutions as part of the feature fusion process. Experimental results indicate that MSC-ResUNet achieves superior accuracy compared to other models, with F1 scores of 0.826 and 0.856, and MCC values of 0.816 and 0.847, in regional robustness and temporal transferability tests, respectively. Performance analysis across different months and band combinations demonstrates that the model maintains high recognition accuracy during both growing and non-growing seasons, despite the study area’s complex landforms and diverse crops.

Predicting an opaque bubble layer during small-incision lenticule extraction surgery based on deep learning

AimThis study aimed to predict the formation of OBL during femtosecond laser SMILE surgery by employing deep learning technology.MethodsThis was a cross-sectional, retrospective study conducted at a university hospital. Surgical videos were randomly divided into a training (3,271 patches, 73.64%), validation (704 patches, 15.85%), and internal verification set (467 patches, 10.51%). An artificial intelligence (AI) model was developed using a SENet-based residual regression deep neural network. Model performance was assessed using the mean absolute error (EMA), Pearson’s correlation coefficient (r), and determination coefficient (R2).ResultsFour distinct types of deep neural network models were established. The modified deep residual neural network prediction model with channel attention built on the PyTorch framework demonstrated the best predictive performance. The predicted OBL area values correlated well with the Photoshop-based measurements (EMA = 0.253, r = 0.831, R2 = 0.676). The ResNet (EMA = 0.259, r = 0.798, R2 = 0.631) and Vgg19 models (EMA = 0.31, r = 0.758, R2 = 0.559) both displayed satisfactory predictive performance, while the U-net model (EMA = 0.605, r = 0.331, R2 = 0.171) performed poorest.ConclusionWe used a panoramic corneal image obtained before the SMILE laser scan to create a unique deep residual neural network prediction model to predict OBL formation during SMILE surgery. This model demonstrated exceptional predictive power, suggesting its clinical applicability across a broad field.