Segmented Images Research Articles

Utilizing Fisheye Cameras and Mass Spectrometry Imaging to Obtain Probing Coordinates for Histological Samples

Detecting chemical compositions in histological samples using mass spectrometry often requires time-consuming user interaction with a probe to ensure it is positioned as closely as possible to a sampling region. Fisheye cameras can obtain images of samples with a wide field-of-view; however, barrel distortion makes it difficult to analyze them as a flat plane. This research project aimed to develop an application that unwarps live images of histological samples captured with a fisheye camera and calculates the exact image-to-real-world probing coordinates with minor user interaction. Python’s PyQt5 and OpenCV libraries were utilized to create interactive software that removes fisheye distortion, finds image-to-real-world coordinates, and segments images. Connected hardware consisted of a Prusa 3D printer, liquid micro junction-surface sampling probe, fisheye camera, and LED light strips. The printer provided a bed for capturing sample images and control of the camera and probe within 1/100 of a millimetre. The software was outlined to process at least 2 photos of checkerboards for the fisheye undistortion, capture a photo of a sample to be unwarped, and require the user to map 4 points from the image to probe coordinates. Using this approach, real-world coordinates could be found for any image location. The application successfully unwarps live photos and converts image-to-real-world coordinates with sub-millimeter accuracy. Users can take new calibration photos, control lighting, save photos, and check undistortion entirely within the application to ensure all parameters are satisfactory. Coordinates of significant regions are displayed or may be manually selected, making it easier to sample specific areas. The success of unwarping relies on the camera’s focus and height, lighting, and object positioning. Future development plans to improve robustness and coordinate accuracy, and reduce the amount of calibration needed.

Enhancing Surface Water Monitoring through Multi-Satellite Data-Fusion of Landsat-8/9, Sentinel-2, and Sentinel-1 SAR

Long revisit intervals and cloud susceptibility have restricted the applicability of earth observation satellites in surface water studies. Integrating multiple satellites offers potential for more frequent observations, yet combining different satellite sources, particularly optical and SAR satellites, presents complexities. This research explores the data-fusion potential and limitations of Landsat-8/9 Operational Land Imager (OLI), Sentinel-2 Multispectral Instrument (MSI), and Sentinel-1 Synthetic Aperture (SAR) satellites to enhance surface water monitoring. By focusing on segmented surface water images, we demonstrate that combining optical and SAR data is generally effective and straightforward using a simple statistical thresholding algorithm. Kappa coefficients(κ) ranging from 0.80 to 0.95 indicate very strong harmony for integration across reservoirs, lakes, and river environments. In vegetative environments, integration with S1SAR shows weak harmony, with κ values ranging from 0.27 to 0.45, indicating the need for further studies. Global revisit interval maps reveal significant improvement in median revisit intervals from 15.87 to 22.81 days using L8/9 alone, to 4.51 to 7.77 days after incorporating S2, and further to 3.48 to 4.62 days after adding S1SAR. Even during wet season months, multi-satellite fusion maintained the median revisit intervals to less than a week. Maximizing all available open-source earth observation satellites is integral for advancing studies requiring more frequent surface water observations, such as flood, inundation, and hydrological modeling.

Lung disease prediction based on CT images using REInf-net and world cup optimization based BI-LSTM classification

ABSTRACT A major global source of disability as well as mortality is respiratory illness. Though visual evaluation of computed tomography (CT) images and chest radiographs are a primary diagnostic for respiratory illnesses, it is limited in its ability to assess severity and predict patient outcomes due to low specificity and fundamental infectious organisms. In order to address these problems, world cup optimization-based Bi-LSTM classification and lung disease prediction on CT images using REINF-net were employed. To enhance the image quality, the gathered lung CT images are pre-processed using Lucy Richardson and CLAHE algorithms. For the purpose of lung infection segmentation, the pre-processed images are segmented using the REInf-net. The GLRLM method is used to extract features from the segmented images. In order to predict lung disease in CT images, the extracted features are trained using the Bi-LSTM based on world cup optimization. Accuracy, Precision, recall, Error and Specificity for the proposed model are 97.8%, 96.7%, 96.7%, 2.2% and 98.3%. These evaluated values are contrasted with the results of existing methods like WCO-BiLSTM, MLP, CNN and LSTM. Finally, the Lung disease prediction based on CT images using REINF-Net and world cup optimization based BI-LSTM classification performs better than the existing model.

Interrater Variability of ML-Based CT-FFR in Patients without Obstructive CAD before TAVR: Influence of Image Quality, Coronary Artery Calcifications, and Location of Measurement.

Objectives: CT-derived fractional flow reserve (CT-FFR) can improve the specificity of coronary CT-angiography (cCTA) for ruling out relevant coronary artery disease (CAD) prior to transcatheter aortic valve replacement (TAVR). However, little is known about the reproducibility of CT-FFR and the influence of diffuse coronary artery calcifications or segment location. The objective was to assess the reliability of machine-learning (ML)-based CT-FFR prior to TAVR in patients without obstructive CAD and to assess the influence of image quality, coronary artery calcium score (CAC), and the location of measurement within the coronary tree. Methods: Patients assessed for TAVR, without obstructive CAD on cCTA were evaluated with ML-based CT-FFR by two observers with differing experience. Differences in absolute values and categorization into hemodynamically relevant CAD (CT-FFR ≤ 0.80) were compared. Results in regard to CAD were also compared against invasive coronary angiography. The influence of segment location, image quality, and CAC was evaluated. Results: Of the screened patients, 109/388 patients did not have obstructive CAD on cCTA and were included. The median (interquartile range) difference of CT-FFR values was -0.005 (-0.09 to 0.04) (p = 0.47). Differences were smaller with high values. Recategorizations were more frequent in distal segments. Diagnostic accuracy of CT-FFR between both observers was comparable (proximal: Δ0.2%; distal: Δ0.5%) but was lower in distal segments (proximal: 98.9%/99.1%; distal: 81.1%/81.6%). Image quality and CAC had no clinically relevant influence on CT-FFR. Conclusions: ML-based CT-FFR evaluation of proximal segments was more reliable. Distal segments with CT-FFR values close to the given threshold were prone to recategorization, even if absolute differences between observers were minimal and independent of image quality or CAC.

NPSFF-Net: Enhanced Building Segmentation in Remote Sensing Images via Novel Pseudo-Siamese Feature Fusion

Building segmentation has extensive research value and application prospects in high-resolution remote sensing image (HRSI) processing. However, complex architectural contexts, varied building morphologies, and non-building occlusions make building segmentation challenging. Compared with traditional methods, deep learning-based methods present certain advantages in terms of accuracy and intelligence. At present, the most popular option is to first apply a single neural network to encode an HRSI, then perform a decoding process through up-sampling or using a transposed convolution operation, and then finally obtain the segmented building image with the help of a loss function. Although effective, this approach not only tends to lead to a loss of detail information, but also fails to fully utilize the contextual features. As an alternative, we propose a novel network called NPSFF-Net. First, using an improved pseudo-Siamese network composed of ResNet-34 and ResNet-50, two sets of deep semantic features of buildings are extracted with the support of transfer learning, and four encoded features at different scales are obtained after fusion. Then, information from the deepest encoded feature is enriched using a feature enhancement module, and the resolutions are recovered via the operations of skip connections and transposed convolutions. Finally, the discriminative features of buildings are obtained using the designed feature fusion algorithm, and the optimal segmentation model is obtained by fitting a cross-entropy loss function. Our method obtained intersection-over-union values of 89.45% for the Aerial Imagery Dataset, 71.88% for the Massachusetts Buildings Dataset, and 68.72% for the Satellite Dataset I.

MFLUnet: multi-scale fusion lightweight Unet for medical image segmentation.

Recently, the use of point-of-care medical devices has been increasing; however, many Unet and its latest variant networks have numerous parameters, high computational complexity, and slow inference speed, making them unsuitable for deployment on these point-of-care or mobile devices. In order to deploy in the real medical environment, we propose a multi-scale fusion lightweight network (MFLUnet), a CNN-based lightweight medical image segmentation model. For the information extraction ability and utilization efficiency of the network, we propose two modules, MSBDCB and EF module, which enable the model to effectively extract local features and global features and integrate multi-scale and multi-stage information while maintaining low computational complexity. The proposed network is validated on three challenging medical image segmentation tasks: skin lesion segmentation, cell segmentation, and ultrasound image segmentation. The experimental results show that our network has excellent performance without occupying almost any computing resources. Ablation experiments confirm the effectiveness of the proposed encoder-decoder and skip connection module. This study introduces a new method for medical image segmentation and promotes the application of medical image segmentation networks in real medical environments.

Deep learning-based virtual staining, segmentation, and classification in label-free photoacoustic histology of human specimens

In pathological diagnostics, histological images highlight the oncological features of excised specimens, but they require laborious and costly staining procedures. Despite recent innovations in label-free microscopy that simplify complex staining procedures, technical limitations and inadequate histological visualization are still problems in clinical settings. Here, we demonstrate an interconnected deep learning (DL)-based framework for performing automated virtual staining, segmentation, and classification in label-free photoacoustic histology (PAH) of human specimens. The framework comprises three components: (1) an explainable contrastive unpaired translation (E-CUT) method for virtual H&E (VHE) staining, (2) an U-net architecture for feature segmentation, and (3) a DL-based stepwise feature fusion method (StepFF) for classification. The framework demonstrates promising performance at each step of its application to human liver cancers. In virtual staining, the E-CUT preserves the morphological aspects of the cell nucleus and cytoplasm, making VHE images highly similar to real H&E ones. In segmentation, various features (e.g., the cell area, number of cells, and the distance between cell nuclei) have been successfully segmented in VHE images. Finally, by using deep feature vectors from PAH, VHE, and segmented images, StepFF has achieved a 98.00% classification accuracy, compared to the 94.80% accuracy of conventional PAH classification. In particular, StepFF’s classification reached a sensitivity of 100% based on the evaluation of three pathologists, demonstrating its applicability in real clinical settings. This series of DL methods for label-free PAH has great potential as a practical clinical strategy for digital pathology.

SCINet: A Segmentation and Classification Interaction CNN Method for Arteriosclerotic Retinopathy Grading.

As a common disease, cardiovascular and cerebrovascular diseases pose a great harm threat to human wellness. Even using advanced and comprehensive treatment methods, there is still a high mortality rate. Arteriosclerosis, as an important factor reflecting the severity of cardiovascular and cerebrovascular diseases, is imperative to detect the arteriosclerotic retinopathy. However, the detection of arteriosclerosis retinopathy requires expensive and time-consuming manual evaluation, while end-to-end deep learning detection methods also need interpretable design to high light task-related features. Considering the importance of automatic arteriosclerotic retinopathy grading, we propose a segmentation and classification interaction network (SCINet). We propose a segmentation and classification interaction architecture for grading arteriosclerotic retinopathy. After IterNet is used to segment retinal vessel from original fundus images, the backbone feature extractor roughly extracts features from the segmented and original fundus arteriosclerosis images and further enhances them through the vessel aware module. The last classifier module generates fundus arteriosclerosis grading results. Specifically, the vessel aware module is designed to highlight the important areal vessel features segmented from original images by attention mechanism, thereby achieving information interaction. The attention mechanism selectively learns the vessel features of segmentation region information under the proposed interactive architecture, which leads to reweighting the extracted features and enhances significant feature information. Extensive experiments have confirmed the effect of our model. SCINet has the best performance on the task of arteriosclerotic retinopathy grading. Additionally, the CNN method is scalable to similar tasks by incorporating segmented images as auxiliary information.

Deep Hybrid Model for Pest Detection: IoT‐UAV‐Based Smart Agriculture System

ABSTRACTModern technology is revolutionising traditional farming processes by introducing new and streamlined approaches. Despite these advancements, challenges such as disease identification, insect detection and weather forecasting persist. To address these issues, this work proposes a DHMPD‐based IoT‐UAV smart agriculture system focused on pest detection. The method involves several stages: data acquisition, preprocessing, data augmentation, segmentation, feature extraction and classification. During data acquisition, a ‘Pest data set’ is collected. Preprocessing includes Z‐score normalisation to produce better‐normalised images. Data augmentation involves rotating images to create different orientations. The segmentation stage uses an updated HDBSCAN process, which improves the distance calculation between pixels using hybridised Euclidean and Minkowski distances. Feature extraction retrieves various features from segmented images, including modified MBP features, colour‐based features and shape‐based features. After feature extraction, the classification phase is performed by a hybrid technique with DL approaches such as improved DBN and LSTM approaches. Finally, classification results are averaged to predict pest detection accurately. The approach's effectiveness is evaluated through various assessments, aiming to overcome current limitations and enhance smart agriculture systems. The proposed DHMPD method was compared with state‐of‐the‐art approaches and traditional classifiers, achieving a maximum accuracy of 0.936, outperforming conventional methods in accurately detecting pests. Hence, the proposed work holds immense promise to advance the capabilities of smart agriculture systems, offering practical solutions that can benefit farmers, agricultural researchers and industries involved in crop management and food production.

HexaLCSeg: A historical benchmark dataset from Hexagon satellite images for land cover segmentation [Software and Data Sets

HexaLCSeg: A historical benchmark dataset from Hexagon satellite images for land cover segmentation [Software and Data Sets

High‐resolution incoherent interference imaging without phase measurement

AbstractAiming for long‐distance, high‐resolution, passive imaging, we use fiber coupling to replace the spatial coupling of the segmented planar imaging detector for electro‐optical reconnaissance imaging system and propose a novel incoherent interference imaging method, which avoids the problem that the resolution of the imaging system is limited by the size of the silicon wafer. The method uses a phase retrieval algorithm and does not need to measure the phase in the system, effectively avoiding the intrinsic jitter problem of fiber optics. The feasibility of the method in achieving target image reconstruction was verified through simulation; furthermore, an indoor two‐dimensional discrete sampling imaging experiment for a simple four‐rod target was conducted, and an outdoor one‐dimensional feature identification experiment for a two‐rod target was also performed. The results showed that the resolution exceeds the diffraction limit and the reconstruction error of target size is less than 3.5%.

EcoFlow: Efficient Convolutional Dataflows on Low-Power Neural Network Accelerators

Dilated and transposed convolutions are widely used in modern convolutional neural networks (CNNs). These kernels are used extensively during CNN training and inference of applications such as image segmentation and high-resolution image generation. We find that commonly-used low-power CNN inference accelerators are <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">not</i> optimized for both these convolutional kernels. Dilated and transposed convolutions introduce significant zero padding when mapped to the underlying spatial architecture, significantly degrading performance and energy efficiency. Existing approaches that address this issue require significant design changes to the otherwise simple, efficient, and well-adopted architectures used to compute direct convolutions. To address this challenge, we propose EcoFlow, a new set of dataflows and mapping algorithms for dilated and transposed convolutions. These algorithms are tailored to execute efficiently on existing low-cost, small-scale spatial architectures and requires minimal changes to existing accelerators. At its core, EcoFlow eliminates zero padding through careful dataflow orchestration and data mapping tailored to the spatial architecture. We evaluate EcoFlow on CNN training workloads and Generative Adversarial Network (GAN) workloads. Experiments in our new cycle-accurate simulator show that, using a common CNN inference accelerator, EcoFlow 1) reduces end-to-end CNN training time between 7-85%, and 2) improves end-to-end GAN training performance between 29-42%, compared to state-of-the-art CNN dataflows.

Integrating Multi‐Scale Feature Boundary Module and Feature Fusion With CNN for Accurate Skin Cancer Segmentation and Classification

ABSTRACTThe skin, a crucial organ, plays a protective role in the human body, emphasizing the significance of early detection of skin diseases to prevent potential progression to skin cancer. The challenge lies in diagnosing these diseases at their early stages, where visual resemblance complicates differentiation, highlighting the need for an innovative automated method for precisely identifying skin lesions in biomedical images. This paper introduces a holistic methodology that combines DenseNet, multi‐scale feature boundary module (MFBM), and feature fusion and decoding engine (FFDE) to tackle challenges in existing deep‐learning image segmentation methods. Furthermore, a convolutional neural network model is designed for the classification of segmented images. The DenseNet encoder efficiently extracts features at four resolution levels, leveraging dense connectivity to capture intricate hierarchical features. The proposed MFBM plays a crucial role in extracting boundary information, employing parallel dilated convolutions with various dilation rates for effective multi‐scale information capture. To overcome potential disadvantages related to the conversion of features during segmentation, our approach ensures the preservation of context features. The proposed FFDE method adaptively fuses features from different levels, restoring skin lesion location information while preserving local details. The evaluation of the model is conducted on the HAM10000 dataset, which consists of 10 015 dermoscopy images, yielding promising results.

An improved semi-supervised segmentation of the retinal vasculature using curvelet-based contrast adjustment and generalized linear model

Diagnosis of most ophthalmic conditions, such as diabetic retinopathy, generally relies on an effective analysis of retinal blood vessels. Techniques that depend solely on the visual observation of clinicians can be tedious and prone to numerous errors. In this article, we propose a semi-supervised automated approach for segmenting blood vessels in retinal color images. Our method effectively combines some classical filters with a Generalized Linear Model (GLM). We first apply the Curvelet Transform along with the Contrast-Limited Histogram Adaptive Equalization (CLAHE) technique to significantly enhance the contrast of vessels in the retinal image during the preprocessing phase. We then use Gabor transform to extract features from the enhanced image. For retinal vasculature identification, we use a GLM learning model with a simple link identity function. Binarization is then performed using an automatic optimal threshold based on the maximum Youden index. A morphological cleaning operation is applied to remove isolated or unwanted segments from the final segmented image. The proposed model is evaluated using statistical parameters on images from three publicly available databases. We achieve average accuracies of 0.9593, 0.9553 and 0.9643, with Receiver Operating Characteristic (ROC) analysis yielding Area Under Curve (AUC) values of 0.9722, 0.9682 and 0.9767 for the CHASE_DB1, STARE and DRIVE databases, respectively. Compared to some of the best results from similar approaches published recently, our results exceed their performance on several datasets.

Automatic bony structure segmentation and curvature estimation on ultrasound cervical spine images - a feasibility study

Abstract The loss of cervical lordosis is a common degenerative disorder known to be associated with abnormal spinal alignment. In recent years, ultrasound (US) imaging has been widely applied in the assessment of spine deformity and has shown promising results. The objectives of this study are to automatically segment bony structures from the 3D US cervical spine image volume and to assess the cervical lordosis on the key sagittal frames. In this study, a portable ultrasound imaging system was applied to acquire cervical spine image volume. The nnU-Net was trained on to segment bony structures on the transverse images and validated by 5-fold-cross-validation. The volume data were reconstructed from the segmented image series. An energy function indicating intensity levels and integrity of bony structures was designed to extract the proxy key sagittal frames on both left and right sides for the cervical curve measurement. The mean absolute difference (MAD), standard deviation (SD) and correlation between the spine curvatures of the left and right sides were calculated for quantitative evaluation of the proposed method. The DSC value of the nnU-Net model in segmenting ROI was 0.973. For the measurement of 22 lamina curve angles, the MAD±SD and correlation between the left and right sides of the cervical spine were 3.591±3.432° and 0.926, respectively. The results indicate that our method has a high accuracy and reliability in the automatic segmentation of the cervical spine and shows the potential of diagnosing the loss of cervical lordosis using the 3D ultrasound imaging technique.

Multimodal Imaging of Posterior Corneal Opacities in Multicentric Osteolysis Nodulosis and Arthropathy (MONA).

Multicentric osteolysis nodulosis and arthropathy (MONA) syndrome is a rare autosomal recessive skeletal dysplasia. Caused by mutations in the matrix metalloproteinase 2 gene (MMP2) on chromosome 16q12, this syndrome has infrequently been associated with ophthalmic manifestations. Corneal opacities have been reported but not described or documented in detail. Complete ophthalmologic examination and multimodal anterior segment imaging were used to characterize the corneal findings in a patient with MONA syndrome. A 19-year-old with MONA syndrome was referred for an eye exam based upon MONA screening recommendations. Visually insignificant peripheral corneal opacities were noted. Anterior segment optical coherence tomography (AS-OCT) demonstrated posterior stromal and endothelial hyperreflectivity. Confocal microscopy demonstrated an acellular peripheral endothelium with a normal central endothelium. Corneal opacities can occur with MONA syndrome, which is caused by mutations in the MMP2 gene. In the patient presented here, the corneal opacities are peripheral, deep stromal, with sparing of the anterior stroma and epithelium.

Image segmentation of Komering script using bounding box

The development of deep learning technology is widely used for various purposes, including recognizing characters in a document. One of the scripts that can benefit from this deep learning technology is the Komering script, which is a local script in the South Sumatra region. However, there are challenges in reading documents written in this script, requiring a method to separate each character in a document. Therefore, there is a need for a technology that can automatically segment images of documents written in the Komering script. This research introduces an innovative technique for segmenting images of characters in documents that contain Komering script characters. The segmentation technique employs bounding box technology to separate each Komering script character, subsequently recognized by a pre-trained deep learning model. The bounding box approach imposes restrictions on the segmented object area. To recognize Komering characters, a deep learning model with a convolutional neural network (CNN) algorithm is employed.

Segmentation of image from a first-person-view unmanned aerial vehicle based on a simple ant algorithm

The object of this study is the process of image segmentation from the First Person View (FPV) of an unmanned aerial vehicle (UAV). The main hypothesis of the study assumes that the use of a simple ant algorithm could ensure the necessary quality of the segmented image. The segmentation method, unlike the known ones, takes into account the number of ants in the image, weight, initial amount and evaporation rate of the pheromone, the "greediness" of the algorithm and provides: – preliminary selection of individual channels of the Red-Green-Blue (RGB) color space; – preliminary placement of ants according to the uniform law; – determining the routes of ants; – taking into account the attractiveness of the route for each ant; – change (adjustment) in the concentration of ant pheromones; – calculation of the probability of movement (transition) of the ant on the movement route; – determination of the objective function at the j-th iteration and its minimization; – determining the coordinates of the route of movement (movement) of ants; – verification of the fulfillment of the stop condition; – determination of the best routes found by ants; – calculation of the brightness of the pixels of the segmented image in each channel of the RGB color space; – further combining the results of channel segmentation. An experimental study of image segmentation from UAV FPV based on a simple ant algorithm was conducted. The specified object of interest on the segmented image has a certain structure, unevenness of the contours, and can be further used for decoding, categorization, etc. Unlike the object of interest, the background ("garbage" objects) in the segmented image do not have a stable structure and can be further filtered out. It has been established that the segmented image by the known method based on the gradient module has a low contrast value, there are gaps in the segmented pixels of the object of interest. A segmented image using a method based on a simple ant algorithm is free from that drawback.

Evaluation of Anatomical Landmark Consistency in the Auditory Meatus Using Cone-Beam Computed Tomography (CBCT) Imaging

IntroductionThe present study evaluated the consistency and reliability of five anatomical landmarks in the external auditory meatus using CBCT (cone beam computed tomography) scans. The goal was to determine if these landmarks could reliably replace the commonly used point Porion for superimposing CBCT scans in clinical practice. Consistent anatomical features in the external auditory meatus are essential for accurate cranial assessments in dentistry and medicine. Despite their importance, these features have been underexplored, making this study a valuable contribution to improving clinical evaluation accuracy. Materials and methods22 CBCT scans were analyzed using ITK-Snap software. The shape of the External Auditory Meatus had been rendered by segmentation. These segmentations were used as a basis to locate the landmarks. The spread in location of the suggested landmarks concerning a reference coordinate system was treated as a measure for consistency. For that purpose, the absolute mean distance from all registered landmarks to the center of the location cloud was calculated for each dataset. ResultsThe smallest spread in locations was measured for the so-called Epitympanic Notch, with 4.3 mm on average (SD 1.7). However, recognizing this landmark in the segmented images appeared to be relatively difficult. The landmark with the second smallest spread in locations (4.4 mm (SD 2.2)) and an easier recognition in the segmented image was the most superior point of the external auditory meatus. The most inferior point appeared to be the least reliable of all five landmarks. ConclusionBased on the present study, a consistently reliable landmark in the External Auditory Meatus to replace point Porion could not be identified. The hypothesis that any landmark could suffice for superimposing CBCT scans was not confirmed.

Uni-to-Multi Modal Knowledge Distillation for Bidirectional LiDAR-Camera Semantic Segmentation.

Combining LiDAR points and images for robust semantic segmentation has shown great potential. However, the heterogeneity between the two modalities (e.g. the density, the field of view) poses challenges in establishing a bijective mapping between each point and pixel. This modality alignment problem introduces new challenges in network design and data processing for cross-modal methods. Specifically, 1) points that are projected outside the image planes; 2) the complexity of maintaining geometric consistency limits the deployment of many data augmentation techniques. To address these challenges, we propose a cross-modal knowledge imputation and transition approach. First, we introduce a bidirectional feature fusion strategy that imputes missing image features and performs cross-modal fusion simultaneously. This allows us to generate reliable predictions even when images are missing. Second, we propose a Uni-to-Multi modal Knowledge Distillation (U2MKD) framework, leveraging the transfer of informative features from a single-modality teacher to a cross-modality student. This overcomes the issues of augmentation misalignment and enables us to train the student effectively. Extensive experiments on the nuScenes, Waymo, and SemanticKITTI datasets demonstrate the effectiveness of our approach. Notably, our method achieves an 8.3 mIoU gain over the LiDAR-only baseline on the nuScenes validation set and achieves state-of-the-art performance on the three datasets.