Kernel Selection Research Articles

ASK-HAR: Attention-Based Multi-Core Selective Kernel Convolution Network for Human Activity Recognition

Human Activity Recognition (HAR) is an increasingly popular field of study aimed at automatically identifying and categorizing human movements and activities using various tracking devices and measurements, such as sensors and cameras. Smartphones have emerged as a prevalent sensor modality for HAR, offering abundant data on an individual’s movements through GPS, accelerometers, and gyroscopes. In conventional convolutional neural networks (CNNs), the artificial neurons within each feature layer typically possess identical receptive fields (RFs). This paper introduces a novel model, ASK-HAR (Attention-based Multi-Core Selective Kernel Convolution Network for HAR), which enhances HAR by performing kernel selection between multiple branches with different RFs using attention mechanisms. The selective kernel mechanism is leveraged to optimize HAR performance. Additionally, the CBAM attention module is employed for time series feature extraction and activity recognition within the overall framework. Extensive experiments conducted on benchmark HAR datasets, including UCI-HAR, USC-HAD, WISDM, PAMAP2, and DSADS, demonstrate that the ASK-HAR model consistently achieves high accuracy across all datasets.

Ultra-High-Resolution Photon-Counting Detector CTA of the Head and Neck: Image Quality Assessment and Vascular Kernel Optimization.

Background: Head and neck CTA requires fine detail evaluation, including characterization of potentially very small vessels and intrastent lumens. Blooming artifacts also hinder evaluation. Objectives: To evaluate image quality of ultra-high-resolution (UHR) photon-counting detector (PCD) CTA of the head and neck and to explore variation of such quality across reconstruction kernels. Methods: This prospective study included patients who underwent clinically indicated head and neck CTA from September 2023 to December 2023. Participants underwent PCD CTA in UHR mode. Reconstructions for each examination included a reference reconstruction (reflecting clinical protocols) using 0.8-mm slice thickness and Bv40 kernel, and six UHR reconstructions using 0.2-mm slice thickness and kernels of varying sharpness (Bv48-Bv80). Quantitative measures were recorded. Two radiologists independently evaluated qualitative measures using Likert scales (1=lowest quality; 5=highest quality). Results: The analysis included 103 participants (mean age, 61.3±13.0 years; 56 male, 48 female). Median vessel sharpness (in HU/mm) was 100.9 for reference reconstruction, and for UHR varied from 110.0 for Bv46 to 121.6 for Bv76 and 134.7 for Bv80. Median right internal carotid artery C2 luminal diameter was 3.8 mm for reference reconstruction, and for UHR increased from 4.1 mm for Bv48 to 4.9 mm for Bv80. For both readers, median overall image quality for reference reconstruction was 3, and for UHR was highest for Bv64 (5); calcified plaque blooming artifact for reference reconstruction was 1, and for UHR was highest for Bv72 (5) and Bv76 (5); stent blooming artifact for reference reconstruction was 1, and for UHR was highest for Bv76 (5) and Bv80 (5); soft-plaque delineation for reference reconstruction was 1, and for UHR was highest for Bv76 (5) or Bv80 (5); small-vessel visualization for reference reconstruction was 1, and for UHR was highest for Bv76 (5) or Bv80 (5). Conclusion: UHR-PCD CTA yielded reduced blooming artifact from calcified plaques or stents, and improved softplaque and small-vessel visualization. These advantages were more pronounced for strongest kernels, although subjective image quality was better for a weaker kernel. Clinical impact: The findings indicate benefits from use of UHR-PCD CTA for head and neck evaluation and may help guide such examinations' kernel selection.

Co-localization of quantitative trait loci for pod and kernel traits and development of molecular marker for kernel weight on chromosome Arahy05 in peanut (Arachis hypogaea L.)

Key messageStable QTL for pod and kernel traits were co-localized on chromosome Arahy05, and an INDEL marker at 106,411,957 on Arahy05 was developed and validated to be useful for marker-assisted selection of kernel weight.Pod and kernel traits, such as hundred pod weight (HPW), and hundred kernel weight (HKW), along with pod and kernel sizes, are pivotal determinants of yield in peanut breeding programs. This study sought to identify quantitative trait loci (QTL) that are associated with these pod and kernel traits in peanuts. To achieve this, a recombinant inbred line (RIL) population, was derived from a cross between Yuhua15, a cultivar known for its high yield, and a germplasm accession W1202. The investigation uncovered stable and major QTL that are significantly associated with both pod and kernel weight and were consistently co-localized on chromosomes Arahy05 and Arahy08. Furthermore, an INDEL marker was identified and characterized in the QTL interval on Arahy05. An extensive re-sequencing analysis comprising 395 germplasm accessions led to the discovery of two principal haplotypes within a 500-kb window flanking the aforementioned INDEL marker. The haplotypes exhibited a significant correlation with the HKW in our diverse panel of germplasm accessions. Notably, the 170 accessions harboring the haplotype associated with an increased HKW primarily represented botanical varieties, specifically Arachis hypogaea var. hypogaea and A. hypogaea var. hirsuta. On the other hand, the 137 accessions associated with the alternative haplotype, which corresponded to a reduced HKW, were predominately identified as belonging to botanical varieties within A. hypogaea subsp. fastigiata. The INDEL marker located on Arahy05, which demonstrates close linkage to the pod and kernel traits, would be an efficient approach for marker-assisted selection (MAS) of pod and kernel weight in breeding programs.

Optimization of Remote-Sensing Image-Segmentation Decoder Based on Multi-Dilation and Large-Kernel Convolution

Land-cover segmentation, a fundamental task within the domain of remote sensing, boasts a broad spectrum of application potential. We address the challenges in land-cover segmentation of remote-sensing imagery and complete the following work. Firstly, to tackle the issues of foreground–background imbalance and scale variation, a module based on multi-dilated rate convolution fusion was integrated into a decoder. This module extended the receptive field through multi-dilated convolution, enhancing the model’s capability to capture global features. Secondly, to address the diversity of scenes and background interference, a hybrid attention module based on large-kernel convolution was employed to improve the performance of the decoder. This module, based on a combination of spatial and channel attention mechanisms, enhanced the extraction of contextual information through large-kernel convolution. A convolution kernel selection mechanism was also introduced to dynamically select the convolution kernel of the appropriate receptive field, suppress irrelevant background information, and improve segmentation accuracy. Ablation studies on the Vaihingen and Potsdam datasets demonstrate that our decoder significantly outperforms the baseline in terms of mean intersection over union and mean F1 score, achieving an increase of up to 1.73% and 1.17%, respectively, compared with the baseline. In quantitative comparisons, the accuracy of our improved decoder also surpasses other algorithms in the majority of categories. The results of this paper indicate that our improved decoder achieves significant performance improvement compared with the old decoder in remote-sensing image-segmentation tasks, which verifies its application potential in the field of land-cover segmentation.

ICNoduleNet: Enhancing Pulmonary Nodule Detection Performance on Sharp Kernel CT Imaging.

Thoracic computed tomography (CT) currently plays the primary role in pulmonary nodule detection, where the reconstruction kernel significantly impacts performance in computer-aided pulmonary nodule detectors. The issue of kernel selection affecting performance has been overlooked in pulmonary nodule detection. This paper first introduces a novel pulmonary nodule detection dataset named Reconstruction Kernel Imaging for Pulmonary Nodule Detection (RKPN) for quantifying algorithm differences between the two imaging types. The dataset contains pairs of images taken from the same patient on the same date, featuring both smooth (B31f) and sharp kernel (B60f) reconstructions. All other imaging parameters and pulmonary nodule labels remain entirely consistent across these pairs. Extensive quantification reveals mainstream detectors perform better on smooth kernel imaging than on sharp kernel imaging. To address suboptimal detection on the sharp kernel imaging, we further propose an image conversion-based pulmonary nodule detector called ICNoduleNet. A lightweight 3D slice-channel converter (LSCC) module is introduced to convert sharp kernel images into smooth kernel images, which can sufficiently learn inter-slice and inter-channel feature information while avoiding introducing excessive parameters. We conduct thorough experiments that validate the effectiveness of ICNoduleNet, it takes sharp kernel images as input and can achieve comparable or even superior detection performance to the baseline that uses the smooth kernel images. The evaluation shows promising results and proves the effectiveness of ICNoduleNet.

Writer Independent Offline Signature Verification System using Global and Local Geometric Features

The objective of this paper is to propose an offline signature verification system (OSVS) designed for writer-independent applications. The system shall differentiate between original and forged signatures. The system encompasses key stages such as pre-processing, feature extraction, and model training and testing, employing the Support Vector Machine algorithm for classification. One challenge in creating a good OSVS is to find proper signature features to be used in the training/classification phases. In this research, global and local features are utilized. This includes signature area, mean, standard deviation, perimeter, number of connected components, number of vertical and horizontal edges, number of end points, number of branch points, and number of lines. The contributions of this paper are on several aspects of the offline signature verification process. Investigation in this study include data pre-processing techniques (normalization vs. standardization), kernel selection (Poly vs. RBF), dataset distribution for training and testing (80%-20% vs. 5-fold), and variations of the C and gamma parameters (C=1, 10, 100 and gamma=1, 10, 100). Improving the recognition rate involves removing of features with little or bad effect on the recognition rate. An algorithm for model-agnostic feature importance is executed, revealing that the most crucial features in the classification process are mean, standard deviation, perimeter, number of connected components, and number of end points. Signatures are classified as either original or forgery, and the model's performance is assessed on the CEDAR dataset. Experimental results shows a 95.65% accuracy of the proposed system when utilizing standardization with the RBF kernel.

Fast Transform Kernel Selection Based on Frequency Matching and Probability Model for AV1

Fast Transform Kernel Selection Based on Frequency Matching and Probability Model for AV1

Multi-kernel subspace stable clustering with exact rank constraints

The identification of cancer subtypes plays a critical role in the early diagnosis of cancer and the delivery of appropriate treatment. Clustering based on comprehensive multi-omic data often yields superior results than clustering using a single data type since each omic type may contain complementary information. However, existing clustering methods cannot fully explore the internal structural information within multi-omic data and can lead to unstable clustering results. This paper proposes a novel multi-omic clustering framework, Multi-Kernel Subspace Stable Clustering with Exact Rank Constraints (MKSSC-ERC), which can effectively explore the correlation and complementary information between different omics types. Specifically, we design a kernel selection criterion that considers both accuracy and diversity and each omic data gets a base kernel. Then, we extract a consistent affinity matrix for these base kernels using subspace segmentation with exact rank constraints, which can refine the extracted shared structures and make them robust to sample noises. In addition, we devise a strategy for selecting the optimal initial cluster centers, which significantly enhances the stability of clustering results. Simulation studies on benchmark multi-omic datasets illustrate substantial gains over existing state-of-the-art methods in survival analysis. MKSSC-ERC is available at https://github.com/xuzihan66/MKSSC-ERC.

Join multiple Riemannian manifold representation and multi‐kernel non‐redundancy for image clustering

AbstractImage clustering has received significant attention due to the growing importance of image recognition. Researchers have explored Riemannian manifold clustering, which is capable of capturing the non‐linear shapes found in real‐world datasets. However, the complexity of image data poses substantial challenges for modelling and feature extraction. Traditional methods such as covariance matrices and linear subspace have shown promise in image modelling, and they are still in their early stages and suffer from certain limitations. However, these include the uncertainty of representing data using only one Riemannian manifold, limited feature extraction capacity of single kernel functions, and resulting incomplete data representation and redundancy. To overcome these limitations, the authors propose a novel approach called join multiple Riemannian manifold representation and multi‐kernel non‐redundancy for image clustering (MRMNR‐MKC). It combines covariance matrices with linear subspace to represent data and applies multiple kernel functions to map the non‐linear structural data into a reproducing kernel Hilbert space, enabling linear model analysis for image clustering. Additionally, the authors use matrix‐induced regularisation to improve the clustering kernel selection process by reducing redundancy and assigning lower weights to identical kernels. Finally, the authors also conducted numerous experiments to evaluate the performance of our approach, confirming its superiority to state‐of‐the‐art methods on three benchmark datasets.

M-SKSNet: Multi-Scale Spatial Kernel Selection for Image Segmentation of Damaged Road Markings

It is a challenging task to accurately segment damaged road markings from images, mainly due to their fragmented, dense, small-scale, and blurry nature. This study proposes a multi-scale spatial kernel selection net named M-SKSNet, a novel model that integrates a transformer and a multi-dilated large kernel convolutional neural network (MLKC) block to address these issues. Through integrating multiple scales of information, the model can extract high-quality and semantically rich features while generating damage-specific representations. This is achieved by leveraging both the local and global contexts, as well as self-attention mechanisms. The performance of M-SKSNet is evaluated both quantitatively and qualitatively, and the results show that M-SKSNet achieved the highest improvement in F1 by 3.77% and in IOU by 4.6%, when compared to existing models. Additionally, the effectiveness of M-SKSNet in accurately extracting damaged road markings from images in various complex scenarios (including city roads and highways) is demonstrated. Furthermore, M-SKSNet is found to outperform existing alternatives in terms of both robustness and accuracy.

Photon-Counting Detector CT Angiography Versus Digital Subtraction Angiography in Patients with Peripheral Arterial Disease

Rationale and ObjectivesThis study aims to compare the diagnostic confidence of photon-counting detector CT angiography (PCD-CTA) depending on the used vascular reformatting kernels with digital subtraction angiography (DSA) as diagnostic reference standard in peripheral arterial occlusive disease (PAOD). Material and MethodsIn 39 patients, 45 lower extremity PCD-CTA with subsequent DSA were analyzed. Advanced PAOD (Fontaine stage 4) was ascertained in 77.8% of patients. CTA post-processing comprised three vascular kernels (Bv36/48/56). Objective image quality assessment included vessel attenuation, image noise, contrast-to-noise (CNR) and signal-to-noise ratios (SNR). Subjective evaluation of calcium blooming, vessel sharpness, luminal attenuation and image noise was performed by three radiologists. Diagnostic performance and concordance to DSA were assessed. ResultsThe luminal attenuation remained kernel-independent constant. With sharper kernels, image noise increased substantially, while SNR and CNR decreased. Subjective reduction of calcium blooming and increased vessel sharpness were noted for the sharp Bv56 kernel. While sensitivity in stenosis quantification was comparable between kernels (81.6% vs. 81.5% vs. 81.0%, p=0.797), specificity increased slightly higher sharpness (71.1% vs. 76.9% vs. 79.6%, p=0.067). Diagnostic concordance of stenosis ratings compared to DSA increased likewise (Bv36 vs. Bv56, p=0.002). Severe crural vessel calcifications had no influence on sensitivity, regardless of kernel selection. Contrarily, specificity was substantially worse in severely calcified tibial vessels but could be improved by using the sharp Bv56 kernel (Bv36 vs. Bv56 p=0.024). Diagnostic confidence was highest for Bv56. ConclusionIn lower leg PCD-CTA, sharp convolution kernels increase diagnostic confidence compared to DSA by improved vessel delineation and reduced calcium blooming with acceptable image noise.

Hybrid Parameter Search and Dynamic Model Selection for Mixed-Variable Bayesian Optimization

This article presents a new type of hybrid model for Bayesian optimization (BO) adept at managing mixed variables, encompassing both quantitative (continuous and integer) and qualitative (categorical) types. Our proposed new hybrid models (named hybridM) merge the Monte Carlo Tree Search structure (MCTS) for categorical variables with Gaussian Processes (GP) for continuous ones. hybridM leverages the upper confidence bound tree search (UCTS) for MCTS strategy, showcasing the tree architecture’s integration into Bayesian optimization. Our innovations, including dynamic online kernel selection in the surrogate modeling phase and a unique UCTS search strategy, position our hybrid models as an advancement in mixed-variable surrogate models. Numerical experiments underscore the superiority of hybrid models, highlighting their potential in Bayesian optimization. Supplementary materials for this article are available online.

Deformable Dynamic Sampling and Dynamic Predictable Mask Mining for Image Inpainting.

Existing image inpainting methods often produce artifacts that are caused by using vanilla convolution layers as building blocks that treat all image regions equally and generate holes at random locations with equal probability. This design does not differentiate the missing regions and valid regions in inference and does not consider the predictability of missing regions in training. To address these issues, we propose a deformable dynamic sampling (DDS) mechanism which is built on deformable convolutions (DCs), and a constraint is proposed to avoid the deformably sampled elements falling into the corrupted regions. Furthermore, to select both valid sample locations and suitable kernels dynamically, we equip DCs with content-aware dynamic kernel selection (DKS). In addition, to further encourage the DDS mechanism to find meaningful sampling locations, we propose to train the inpainting model with mined predictable regions as holes. During training, we jointly train a mask generator with the inpainting network to generate hole masks dynamically for each training sample. Thus, the mask generator can find large yet predictable missing regions as a better alternative to random masks. Extensive experiments demonstrate the advantages of our method over state-of-the-art methods qualitatively and quantitatively.

Improvement of Kernel SVM to Enhance Accuracy in Chronic Kidney Disease

Chronic Kidney Disease (CKD) is a highly serious health issue, affecting millions of people worldwide. Early diagnosis and accurate prediction of chronic kidney disease are key factors in successful treatment. One of the approaches used for diagnosing this disease is through machine learning algorithms, specifically the Support Vector Machine (SVM) method. By collecting CKD data that includes various clinical parameters, initial kernel selection as well as various kernels are tested. However, the accuracy of the SVM method can be further improved for better diagnosis. The objective of this research is to enhance accuracy, optimize parameters, and improve the SVM kernel by incorporating the Particle Swarm Optimization (PSO) algorithm. The results of this study indicate that the use of PSO method to improve SVM kernels can significantly enhance accuracy in CKD diagnosis compared to conventional SVM approaches, potentially aiding medical practitioners in early disease diagnosis and better CKD management, which in turn can improve patient prognosis and quality of life

A new adaptive multi-kernel relevance vector regression for structural reliability analysis

Surrogate models have been widely used in structural reliability analysis to improve the computational efficiency and the accuracy of failure probability. Recently, several multi-kernel relevance vector regression (MKRVR) models have been studied to evaluate the failure probability. However, existing multiple kernel functions for relevance vector regression models are fixed choices, which increases the number of calls to the limit state function (LSF) and leads to inaccurate results. To address the problem, this paper presents a new adaptive MKRVR model combined with Monte Carlo simulation (MCS). Firstly, a stepwise kernel selection strategy is developed to adaptively select better-performing kernel functions and eliminate redundant kernel functions for constructing the MKRVR model. Secondly, a new active learning function is proposed by considering the probability of mis-prediction and spatial locations of the existing sampling point to identify the new training sample points. Thirdly, a hybrid efficient stopping criterion is adopted to terminate the learning process automatically. Three benchmark examples and one practical engineering example are introduced to demonstrate the effectiveness of the proposed method. Results show that the proposed method can provide accurate failure probability by less number of calls to the LSF than existing fixed kernel-based methods.

An improved YOLOv5s model for assessing apple graspability in automated harvesting scene.

With continuously increasing labor costs, an urgent need for automated apple- Qpicking equipment has emerged in the agricultural sector. Prior to apple harvesting, it is imperative that the equipment not only accurately locates the apples, but also discerns the graspability of the fruit. While numerous studies on apple detection have been conducted, the challenges related to determining apple graspability remain unresolved. This study introduces a method for detecting multi-occluded apples based on an enhanced YOLOv5s model, with the aim of identifying the type of apple occlusion in complex orchard environments and determining apple graspability. Using bootstrap your own atent(BYOL) and knowledge transfer(KT) strategies, we effectively enhance the classification accuracy for multi-occluded apples while reducing data production costs. A selective kernel (SK) module is also incorporated, enabling the network model to more precisely identify various apple occlusion types. To evaluate the performance of our network model, we define three key metrics: APGA, APTUGA, and APUGA, representing the average detection accuracy for graspable, temporarily ungraspable, and ungraspable apples, respectively. Experimental results indicate that the improved YOLOv5s model performs exceptionally well, achieving detection accuracies of 94.78%, 93.86%, and 94.98% for APGA, APTUGA, and APUGA, respectively. Compared to current lightweight network models such as YOLOX-s and YOLOv7s, our proposed method demonstrates significant advantages across multiple evaluation metrics. In future research, we intend to integrate fruit posture and occlusion detection to f]urther enhance the visual perception capabilities of apple-picking equipment.

On the Use of Kullback–Leibler Divergence for Kernel Selection and Interpretation in Variational Autoencoders for Feature Creation

This study presents a novel approach for kernel selection based on Kullback–Leibler divergence in variational autoencoders using features generated by the convolutional encoder. The proposed methodology focuses on identifying the most relevant subset of latent variables to reduce the model’s parameters. Each latent variable is sampled from the distribution associated with a single kernel of the last encoder’s convolutional layer, resulting in an individual distribution for each kernel. Relevant features are selected from the sampled latent variables to perform kernel selection, which filters out uninformative features and, consequently, unnecessary kernels. Both the proposed filter method and the sequential feature selection (standard wrapper method) were examined for feature selection. Particularly, the filter method evaluates the Kullback–Leibler divergence between all kernels’ distributions and hypothesizes that similar kernels can be discarded as they do not convey relevant information. This hypothesis was confirmed through the experiments performed on four standard datasets, where it was observed that the number of kernels can be reduced without meaningfully affecting the performance. This analysis was based on the accuracy of the model when the selected kernels fed a probabilistic classifier and the feature-based similarity index to appraise the quality of the reconstructed images when the variational autoencoder only uses the selected kernels. Therefore, the proposed methodology guides the reduction of the number of parameters of the model, making it suitable for developing applications for resource-constrained devices.

Influence of tree-based multi-layer node information on scoring accuracy and speed of speaker verification

In the speaker verification task based on Gaussian Mixture Model-Universal Background Model (GMM-UBM), by constructing the UBM as a tree structure, the kernel Gaussians suitable for different speakers can be quickly selected, which speeds up the modeling of speaker acoustic space by GMM. The tree-based kernel selection algorithm (TBKS) introduces a beam-width, which increases the candidate range of kernels and improves the kernel selection accuracy. In this paper, we improve the TBKS algorithm by introducing a recall rate to adjust the number of nodes recalled in each layer of the tree structure. This adjustment refines the quantity and resolution of Gaussian distributions in various subspaces within the acoustic space, compensating for the loss caused by discarding some significant Gaussians erroneously. Speaker verification experiments are carried out based on the Aishell2 dataset. The results reveal that the modified TBKS algorithm reduces EER by 7.5% relatively and increses computational reduction factor to 42.93, enhancing both recognition accuracy and speed. In addition, the test speech is spliced into different lengths and common environmental noise is added to verify the universality of the improved algorithm.

Confectionery Groundnut Genotypes (Arachis hypogaea L.): A Study on Genetic Variability Parameters with Special Reference to Confectionery Traits Improvement

Diverse groundnut parents were used to generate 36 advanced breeding lines to develop into a confectionery genotype for yield, quality, and main biotic stress. Groundnut use is increasing in both food and confectionary due to the steady change in groundnut consumption patterns. Advanced breeding lines evaluated in randomized complete block design including parents and checks during kharif and summer, 2021 respectively. High genetic variability (%) and heritability (%) observed for hundred seed weight (21.08 & 86.34), pod yield per plant (25.37 & 87.66), pod yield (31.63 & 83.71), late leaf spot (32.71 & 79.16), rust disease (39.86 & 87.06), oil content (2.55 & 79.60), Blanchability (18.76 & 97.00), oleic acid (8.52 & 72.90), linoleic acid (16.32 & 73.40) and O/L ratio (24.64 & 73.60) respectively. The correlation between hundred seed weight, sound mature kernels, shelling percentage and pod yield per plant with kernel yield showed positive association. Rust and LLS disease revealed a high positive connection indicating that an increase in late leaf spot disease incidence enhances the incidence of rust disease and vice versa in groundnut. Substantial negative association was found between two crucial quality parameters such as oil content and protein content, as these two characteristics determine the quality of groundnut for confectionary purposes. Path analysis indicates kernel yield had the greatest direct positive effect on pod yield, followed by seed weight, oil content and sound mature kernel and direct selection for these characteristics would be effective. The path analysis of quality attributes revealed that at the genotypic level, oil content, protein content, and O/L ratio have the most direct effect on hundred seed weight.

MAS-Net:Multi-modal Assistant Segmentation Network For Lumbar Intervertebral Disc.

Objective.Despite advancements in medical imaging technology, the diagnosis and positioning of lumbar disc diseases still heavily rely on the expertise and experience of medical professionals. This process is often time-consuming, labor-intensive, and susceptible to subjective factors. Achieving automatic positioning and segmentation of lumbar intervertebral disc (LID) is the first and critical step in intelligent diagnosis of lumbar disc diseases. However, due to the complexity of the vertebral body and the ambiguity of the soft tissue boundaries of the LID, accurate and intelligent segmentation of LIDs remains challenging. The study aims to accurately and intelligently segment and locate LIDs by fully utilizing multi-modal lumbar magnetic resonance Images (MRIs).Approach.A novel multi-modal assistant segmentation network (MAS-Net) is proposed in this paper. The architecture consists of four key components: the multi-branch fusion encoder (MBFE), the cross-modality correlation evaluation (CMCE), the channel fusion transformer (CFT), and the selective Kernel (SK) based decoder. The MBFE module captures and integrates various modal features, while the CMCE module facilitates the fusion process between the MBFE and decoder. The CFT module selectively guides the flow of information between the MBFE and decoder and effectively utilizes skip connections from multiple layers. The SK module computes the significance of each channel using global pooling operations and applies weights to the input feature maps to improve the models recognition of important features.Main results.The proposed MAS-Net achieved a dice coefficient of 93.08% on IVD3Seg and 93.22% on DualModalDisc dataset, outperforming the current state-of-the-art network, accurately segmenting the LIDs, and generating a 3D model that can precisely display the LIDs.Significance.MAS-Net automates the diagnostics process and addresses challenges faced by doctors. Simplifying and enhancing the clarity of visual representation, multi-modal MRI allows for better information complementation and LIDs segmentation. By successfully integrating data from various modalities, the accuracy of LID segmentation is improved.