The precise spatial and quantitative delineation of indistinct-boundary medical objects is paramount for the accuracy of diagnostic protocols, efficacy of surgical interventions, and reliability of postoperative assessments. Despite their significance, the effective segmentation and instantaneous three-dimensional reconstruction are significantly impeded by the paucity of representative samples in available datasets and noise artifacts. To surmount these challenges, we introduced Stochastic Defect Injection (SDi) to augment the representational diversity of challenging indistinct-boundary objects within training corpora. Consequently, we propose the Dual-Encoder Fourier Group Harmonics Network (DEFN) to tailor noise filtration, amplify detailed feature recognition, and bolster representation across diverse medical imaging scenarios. By incorporating Dynamic Weight Composing (DWC) loss dynamically adjusts model's focus based on training progression, DEFN achieves SOTA performance on the OIMHS public dataset, showcasing effectiveness in indistinct boundary contexts. Source code for DEFN is available at: https://github.com/IMOP-lab/DEFN-pytorch.
Referring Expression Segmentation (RES) is an emerging task in computer vision, which segments the target instances in images based on text descriptions. However, its development is plagued by the expensive segmentation labels. To address this issue, we propose a new learning task for RES called Omni-supervised Referring Expression Segmentation (Omni-RES), which aims to make full use of unlabeled, fully labeled and weakly labeled data, e.g., referring points or grounding boxes, for efficient RES training. To accomplish this task, we also propose a novel yet strong baseline method for Omni-RES based on the recently popular teacher-student learning, where the weak labels are not directly transformed into supervision signals but used as a yardstick to select and refine high-quality pseudo-masks for teacher-student learning. To validate the proposed Omni-RES method, we apply it to a set of state-of-the-art RES models and conduct extensive experiments on a bunch of RES datasets. The experimental results yield the obvious merits of Omni-RES than the fully-supervised and semi-supervised training schemes. For instance, with only 10% fully labeled data, Omni-RES can help the base model achieve 100% fully supervised performance, and it also outperform the semi-supervised alternative by a large margin, e.g., +14.93% on RefCOCO and +14.95% on RefCOCO+, respectively. More importantly, Omni-RES also enable the use of large-scale vision-langauges like Visual Genome to facilitate low-cost RES training, and achieve new SOTA performance of RES, e.g., 80.66 on RefCOCO.
Brain hemorrhage is potentially a fatal condition that results from internal bleeding in the human brain. In this study, Computed Tomography (CT) scan images have been used for segmentation tasks to pinpoint the area of hemorrhage. Unique data augmentation techniques using non-linear transformations like, Twirl and Spherical have been used along with traditional data augmentation techniques to increase variation in the dataset. The hemorrhagic portion of the brain in images that are easy to distinguish have been annotated to perform the segmentation task. The segmentation task was applied using U-Net and U-Net++ architecture. U-Net architecture has shown 84.33% Intersection over Union (IoU) and 91.34% dice coefficient score whereas U-Net++ has achieved 17.06% IoU and 28.27% dice coefficient score after applying some non-linear transformations on the dataset.
In high-quality radiotherapy delivery, precise segmentation of targets and healthy structures is essential. This study proposes Radiomics features as a superior measure for assessing the segmentation ability of physicians and auto-segmentation tools, in comparison to the widely used Dice Similarity Coefficient (DSC). The research involves selecting reproducible radiomics features for evaluating segmentation accuracy by analyzing radiomics data from 2 CT scans of 10 lung tumors, available in the RIDER Data Library. Radiomics features were extracted using PyRadiomics, with selection based on the Concordance Correlation Coefficient (CCC). Subsequently, CT images from 10 patients, each segmented by different physicians or auto-segmentation tools, were used to assess segmentation performance. The study reveals 206 radiomics features with a CCC greater than 0.93 between the two CT images, indicating robust reproducibility. Among these features, seven exhibit low Intraclass Correlation Coefficients (ICC), signifying increased sensitivity to segmentation differences. Notably, ICCs of original shape features, including sphericity, elongation, and flatness, ranged from 0.1177 to 0.995. In contrast, all DSC values exceeded 0.778. This research demonstrates that radiomics features, particularly those related to shape and energy, can capture subtle variations in tumor segmentation characteristics, unlike DSC. As a result, Radiomics features with ICC prove superior for evaluating a physician's tumor segmentation ability and the performance of auto-segmentation tools. The findings suggest that these new metrics can be employed to assess novel auto-segmentation methods and enhance the training of individuals in medical segmentation, thus contributing to improved radiotherapy practices.
One of the common and promising deep learning approaches used for medical image segmentation is transformers, as they can capture long-range dependencies among the pixels by utilizing self-attention. Despite being successful in medical image segmentation, transformers face limitations in capturing local contexts of pixels in multimodal dimensions. We propose a Medical Image Segmentation Transformer (MIST) incorporating a novel Convolutional Attention Mixing (CAM) decoder to address this issue. MIST has two parts: a pre-trained multi-axis vision transformer (MaxViT) is used as an encoder, and the encoded feature representation is passed through the CAM decoder for segmenting the images. In the CAM decoder, an attention-mixer combining multi-head self-attention, spatial attention, and squeeze and excitation attention modules is introduced to capture long-range dependencies in all spatial dimensions. Moreover, to enhance spatial information gain, deep and shallow convolutions are used for feature extraction and receptive field expansion, respectively. The integration of low-level and high-level features from different network stages is enabled by skip connections, allowing MIST to suppress unnecessary information. The experiments show that our MIST transformer with CAM decoder outperforms the state-of-the-art models specifically designed for medical image segmentation on the ACDC and Synapse datasets. Our results also demonstrate that adding the CAM decoder with a hierarchical transformer improves segmentation performance significantly. Our model with data and code is publicly available on GitHub.
This paper proposes a thermal-infrared (TIR) remote target detection system for maritime rescue using deep learning and data augmentation. We established a self-collected TIR dataset consisting of multiple scenes imitating human rescue situations using a TIR camera (FLIR). Additionally, to address dataset scarcity and improve model robustness, a synthetic dataset from a 3D game (ARMA3) to augment the data is further collected. However, a significant domain gap exists between synthetic TIR and real TIR images. Hence, a proper domain adaptation algorithm is essential to overcome the gap. Therefore, we suggest a domain adaptation algorithm in a target-background separated manner from 3D game-to-real, based on a generative model, to address this issue. Furthermore, a segmentation network with fixed-weight kernels at the head is proposed to improve the signal-to-noise ratio (SNR) and provide weak attention, as remote TIR targets inherently suffer from unclear boundaries. Experiment results reveal that the network trained on augmented data consisting of translated synthetic and real TIR data outperforms that trained on only real TIR data by a large margin. Furthermore, the proposed segmentation model surpasses the performance of state-of-the-art segmentation methods.
Skin lesion segmentation plays a crucial role in the computer-aided diagnosis of melanoma. Deep Learning models have shown promise in accurately segmenting skin lesions, but their widespread adoption in real-life clinical settings is hindered by their inherent black-box nature. In domains as critical as healthcare, interpretability is not merely a feature but a fundamental requirement for model adoption. This paper proposes IARS SegNet an advanced segmentation framework built upon the SegNet baseline model. Our approach incorporates three critical components: Skip connections, residual convolutions, and a global attention mechanism onto the baseline Segnet architecture. These elements play a pivotal role in accentuating the significance of clinically relevant regions, particularly the contours of skin lesions. The inclusion of skip connections enhances the model's capacity to learn intricate contour details, while the use of residual convolutions allows for the construction of a deeper model while preserving essential image features. The global attention mechanism further contributes by extracting refined feature maps from each convolutional and deconvolutional block, thereby elevating the model's interpretability. This enhancement highlights critical regions, fosters better understanding, and leads to more accurate skin lesion segmentation for melanoma diagnosis.
In this report, we present our approach to the EPIC-KITCHENS VISOR Hand Object Segmentation Challenge, which focuses on the estimation of the relation between the hands and the objects given a single frame as input. The EPIC-KITCHENS VISOR dataset provides pixel-wise annotations and serves as a benchmark for hand and active object segmentation in egocentric video. Our approach combines the baseline method, i.e., Point-based Rendering (PointRend) and the Segment Anything Model (SAM), aiming to enhance the accuracy of hand and object segmentation outcomes, while also minimizing instances of missed detection. We leverage accurate hand segmentation maps obtained from the baseline method to extract more precise hand and in-contact object segments. We utilize the class-agnostic segmentation provided by SAM and apply specific hand-crafted constraints to enhance the results. In cases where the baseline model misses the detection of hands or objects, we re-train an object detector on the training set to enhance the detection accuracy. The detected hand and in-contact object bounding boxes are then used as prompts to extract their respective segments from the output of SAM. By effectively combining the strengths of existing methods and applying our refinements, our submission achieved the 1st place in terms of evaluation criteria in the VISOR HOS Challenge.
Ubiquitous sensors today emit high frequency streams of numerical measurements that reflect properties of human, animal, industrial, commercial, and natural processes. Shifts in such processes, e.g. caused by external events or internal state changes, manifest as changes in the recorded signals. The task of streaming time series segmentation (STSS) is to partition the stream into consecutive variable-sized segments that correspond to states of the observed processes or entities. The partition operation itself must in performance be able to cope with the input frequency of the signals. We introduce ClaSS, a novel, efficient, and highly accurate algorithm for STSS. ClaSS assesses the homogeneity of potential partitions using self-supervised time series classification and applies statistical tests to detect significant change points (CPs). In our experimental evaluation using two large benchmarks and six real-world data archives, we found ClaSS to be significantly more precise than eight state-of-the-art competitors. Its space and time complexity is independent of segment sizes and linear only in the sliding window size. We also provide ClaSS as a window operator with an average throughput of 1k data points per second for the Apache Flink streaming engine.
Muscle volume is a useful quantitative biomarker in sports, but also for the follow-up of degenerative musculo-skelletal diseases. In addition to volume, other shape biomarkers can be extracted by segmenting the muscles of interest from medical images. Manual segmentation is still today the gold standard for such measurements despite being very time-consuming. We propose a method for automatic segmentation of 18 muscles of the lower limb on 3D Magnetic Resonance Images to assist such morphometric analysis. By their nature, the tissue of different muscles is undistinguishable when observed in MR Images. Thus, muscle segmentation algorithms cannot rely on appearance but only on contour cues. However, such contours are hard to detect and their thickness varies across subjects. To cope with the above challenges, we propose a segmentation approach based on a hybrid architecture, combining convolutional and visual transformer blocks. We investigate for the first time the behaviour of such hybrid architectures in the context of muscle segmentation for shape analysis. Considering the consistent anatomical muscle configuration, we rely on transformer blocks to capture the longrange relations between the muscles. To further exploit the anatomical priors, a second contribution of this work consists in adding a regularisation loss based on an adjacency matrix of plausible muscle neighbourhoods estimated from the training data. Our experimental results on a unique database of elite athletes show it is possible to train complex hybrid models from a relatively small database of large volumes, while the anatomical prior regularisation favours better predictions.
To address prevalent issues in medical imaging, such as data acquisition challenges and label availability, transfer learning from natural to medical image domains serves as a viable strategy to produce reliable segmentation results. However, several existing barriers between domains need to be broken down, including addressing contrast discrepancies, managing anatomical variability, and adapting 2D pretrained models for 3D segmentation tasks. In this paper, we propose ProMISe,a prompt-driven 3D medical image segmentation model using only a single point prompt to leverage knowledge from a pretrained 2D image foundation model. In particular, we use the pretrained vision transformer from the Segment Anything Model (SAM) and integrate lightweight adapters to extract depth-related (3D) spatial context without updating the pretrained weights. For robust results, a hybrid network with complementary encoders is designed, and a boundary-aware loss is proposed to achieve precise boundaries. We evaluate our model on two public datasets for colon and pancreas tumor segmentations, respectively. Compared to the state-of-the-art segmentation methods with and without prompt engineering, our proposed method achieves superior performance. The code is publicly available at https://github.com/MedICL-VU/ProMISe.
Extracting long tracks and lineages from videomicroscopy requires an extremely low error rate, which is challenging on complex datasets of dense or deforming cells. Leveraging temporal context is key to overcoming this challenge. We propose DistNet2D, a new deep neural network (DNN) architecture for 2D cell segmentation and tracking that leverages both mid- and long-term temporal information. DistNet2D considers seven frames at the input and uses a post-processing procedure that exploits information from the entire video to correct segmentation errors. DistNet2D outperforms two recent methods on two experimental datasets, one containing densely packed bacterial cells and the other containing eukaryotic cells. It is integrated into an ImageJ-based graphical user interface for 2D data visualization, curation, and training. Finally, we demonstrate the performance of DistNet2D on correlating the size and shape of cells with their transport properties over large statistics, for both bacterial and eukaryotic cells.
Unsupervised monocular depth estimation techniques have demonstrated encouraging results but typically assume that the scene is static. These techniques suffer when trained on dynamical scenes, where apparent object motion can equally be explained by hypothesizing the object's independent motion, or by altering its depth. This ambiguity causes depth estimators to predict erroneous depth for moving objects. To resolve this issue, we introduce Dynamo-Depth, an unifying approach that disambiguates dynamical motion by jointly learning monocular depth, 3D independent flow field, and motion segmentation from unlabeled monocular videos. Specifically, we offer our key insight that a good initial estimation of motion segmentation is sufficient for jointly learning depth and independent motion despite the fundamental underlying ambiguity. Our proposed method achieves state-of-the-art performance on monocular depth estimation on Waymo Open and nuScenes Dataset with significant improvement in the depth of moving objects. Code and additional results are available at https://dynamo-depth.github.io.
In this paper, we propose a new multi-modal task, termed audio-visual instance segmentation (AVIS), which aims to simultaneously identify, segment and track individual sounding object instances in audible videos. To facilitate this research, we introduce a high-quality benchmark named AVISeg, containing over 90K instance masks from 26 semantic categories in 926 long videos. Additionally, we propose a strong baseline model for this task. Our model first localizes sound source within each frame, and condenses object-specific contexts into concise tokens. Then it builds long-range audio-visual dependencies between these tokens using window-based attention, and tracks sounding objects among the entire video sequences. Extensive experiments reveal that our method performs best on AVISeg, surpassing the existing methods from related tasks. We further conduct the evaluation on several multi-modal large models. Unfortunately, they exhibits subpar performance on instance-level sound source localization and temporal perception. We expect that AVIS will inspire the community towards a more comprehensive multi-modal understanding. Dataset and code is available at https://github.com/ruohaoguo/avis.
Recent works have shown that the computational efficiency of 3D medical image (e.g. CT and MRI) segmentation can be impressively improved by dynamic inference based on slice-wise complexity. As a pioneering work, a dynamic architecture network for medical volumetric segmentation (i.e. Med-DANet) has achieved a favorable accuracy and efficiency trade-off by dynamically selecting a suitable 2D candidate model from the pre-defined model bank for different slices. However, the issues of incomplete data analysis, high training costs, and the two-stage pipeline in Med-DANet require further improvement. To this end, this paper further explores a unified formulation of the dynamic inference framework from the perspective of both the data itself and the model structure. For each slice of the input volume, our proposed method dynamically selects an important foreground region for segmentation based on the policy generated by our Decision Network and Crop Position Network. Besides, we propose to insert a stage-wise quantization selector to the employed segmentation model (e.g. U-Net) for dynamic architecture adapting. Extensive experiments on BraTS 2019 and 2020 show that our method achieves comparable or better performance than previous state-of-the-art methods with much less model complexity. Compared with previous methods Med-DANet and TransBTS with dynamic and static architecture respectively, our framework improves the model efficiency by up to nearly 4.1 and 17.3 times with comparable segmentation results on BraTS 2019.
While traditional methods relies on depth sensors, the current trend leans towards utilizing cost-effective RGB images, despite their absence of depth cues. This paper introduces an interesting approach to detect grasping pose from a single RGB image. To this end, we propose a modular learning network augmented with grasp detection and semantic segmentation, tailored for robots equipped with parallel-plate grippers. Our network not only identifies graspable objects but also fuses prior grasp analyses with semantic segmentation, thereby boosting grasp detection precision. Significantly, our design exhibits resilience, adeptly handling blurred and noisy visuals. Key contributions encompass a trainable network for grasp detection from RGB images, a modular design facilitating feasible grasp implementation, and an architecture robust against common image distortions. We demonstrate the feasibility and accuracy of our proposed approach through practical experiments and evaluations.
This paper studies the problem of weakly open-vocabulary semantic segmentation (WOVSS), which learns to segment objects of arbitrary classes using mere image-text pairs. Existing works turn to enhance the vanilla vision transformer by introducing explicit grouping recognition, i.e., employing several group tokens/centroids to cluster the image tokens and perform the group-text alignment. Nevertheless, these methods suffer from a granularity inconsistency regarding the usage of group tokens, which are aligned in the all-to-one v.s. one-to-one manners during the training and inference phases, respectively. We argue that this discrepancy arises from the lack of elaborate supervision for each group token. To bridge this granularity gap, this paper explores explicit supervision for the group tokens from the prototypical knowledge. To this end, this paper proposes the non-learnable prototypical regularization (NPR) where non-learnable prototypes are estimated from source features to serve as supervision and enable contrastive matching of the group tokens. This regularization encourages the group tokens to segment objects with less redundancy and capture more comprehensive semantic regions, leading to increased compactness and richness. Based on NPR, we propose the prototypical guidance segmentation network (PGSeg) that incorporates multi-modal regularization by leveraging prototypical sources from both images and texts at different levels, progressively enhancing the segmentation capability with diverse prototypical patterns. Experimental results show that our proposed method achieves state-of-the-art performance on several benchmark datasets. The source code is available at https://github.com/Ferenas/PGSeg.
Recent advances in Vision Transformers (ViT) and Stable Diffusion (SD) models with their ability to capture rich semantic features of the image have been used for image correspondence tasks on natural images. In this paper, we examine the ability of a variety of pre-trained ViT (DINO, DINOv2, SAM, CLIP) and SD models, trained exclusively on natural images, for solving the correspondence problems on medical images. While many works have made a case for in-domain training, we show that the models trained on natural images can offer good performance on medical images across different modalities (CT,MR,Ultrasound) sourced from various manufacturers, over multiple anatomical regions (brain, thorax, abdomen, extremities), and on wide variety of tasks. Further, we leverage the correspondence with respect to a template image to prompt a Segment Anything (SAM) model to arrive at single shot segmentation, achieving dice range of 62%-90% across tasks, using just one image as reference. We also show that our single-shot method outperforms the recently proposed few-shot segmentation method - UniverSeg (Dice range 47%-80%) on most of the semantic segmentation tasks(six out of seven) across medical imaging modalities.
In this paper, we study a challenging task of zero-shot referring image segmentation. This task aims to identify the instance mask that is most related to a referring expression without training on pixel-level annotations. Previous research takes advantage of pre-trained cross-modal models, e.g., CLIP, to align instance-level masks with referring expressions. %Yet, CLIP only considers image-text pair level alignment, which neglects fine-grained image region and complex sentence matching. Yet, CLIP only considers the global-level alignment of image-text pairs, neglecting fine-grained matching between the referring sentence and local image regions. To address this challenge, we introduce a Text Augmented Spatial-aware (TAS) zero-shot referring image segmentation framework that is training-free and robust to various visual encoders. TAS incorporates a mask proposal network for instance-level mask extraction, a text-augmented visual-text matching score for mining the image-text correlation, and a spatial rectifier for mask post-processing. Notably, the text-augmented visual-text matching score leverages a $P$ score and an $N$-score in addition to the typical visual-text matching score. The $P$-score is utilized to close the visual-text domain gap through a surrogate captioning model, where the score is computed between the surrogate model-generated texts and the referring expression. The $N$-score considers the fine-grained alignment of region-text pairs via negative phrase mining, encouraging the masked image to be repelled from the mined distracting phrases. Extensive experiments are conducted on various datasets, including RefCOCO, RefCOCO+, and RefCOCOg. The proposed method clearly outperforms state-of-the-art zero-shot referring image segmentation methods.
Coronary angiography analysis is a common clinical task performed by cardiologists to diagnose coronary artery disease (CAD) through an assessment of atherosclerotic plaque's accumulation. This study introduces an end-to-end machine learning solution developed as part of our solution for the MICCAI 2023 Automatic Region-based Coronary Artery Disease diagnostics using x-ray angiography imagEs (ARCADE) challenge, which aims to benchmark solutions for multivessel coronary artery segmentation and potential stenotic lesion localisation from X-ray coronary angiograms. We adopted a robust baseline model training strategy to progressively improve performance, comprising five successive stages of binary class pretraining, multivessel segmentation, fine-tuning using class frequency weighted dataloaders, fine-tuning using F1-based curriculum learning strategy (F1-CLS), and finally multi-target angiogram view classifier-based collective adaptation. Unlike many other medical imaging procedures, this task exhibits a notable degree of interobserver variability. %, making it particularly amenable to automated analysis. Our ensemble model combines the outputs from six baseline models using the weighted ensembling approach, which our analysis shows is found to double the predictive accuracy of the proposed solution. The final prediction was further refined, targeting the correction of misclassified blobs. Our solution achieved a mean F1 score of $37.69\%$ for coronary artery segmentation, and $39.41\%$ for stenosis localisation, positioning our team in the 5th position on both leaderboards. This work demonstrates the potential of automated tools to aid CAD diagnosis, guide interventions, and improve the accuracy of stent injections in clinical settings.