Adaptive weighted multi-teacher distillation for efficient medical imaging segmentation with limited data

PurposeConvolution operator-based neural networks have shown great success in medical image segmentation over the past decade. The U-shaped network with a codec structure is one of the most widely used models. Transformer, a technology used in natural language processing, can capture long-distance dependencies and has been applied in Vision Transformer to achieve state-of-the-art performance on image classification tasks. Recently, researchers have extended transformer to medical image segmentation tasks, resulting in good models.MethodsThis review comprises publications selected through a Web of Science search. We focused on papers published since 2018 that applied the transformer architecture to medical image segmentation. We conducted a systematic analysis of these studies and summarized the results.ResultsTo better comprehend the benefits of convolutional neural networks and transformers, the construction of the codec and transformer modules is first explained. Second, the medical image segmentation model based on transformer is summarized. The typically used assessment markers for medical image segmentation tasks are then listed. Finally, a large number of medical segmentation datasets are described.ConclusionEven if there is a pure transformer model without any convolution operator, the sample size of medical picture segmentation still restricts the growth of the transformer, even though it can be relieved by a pretraining model. More often than not, researchers are still designing models using transformer and convolution operators.

DSST: A dual student model guided student–teacher framework for semi-supervised medical image segmentation

Cold SegDiffusion: A novel diffusion model for medical image segmentation

Deep learning models, especially U-Net and its derivate models, have been widely used in medical image segmentation. These approaches have achieved promising results in many medical image segmentation tasks with a limited number of training samples. We aim on enhancing medical image segmentation by using spatial continuity information in a proposed Multi-Encoder Parse-Decoder Network (MEPDNet) based on the fact that most of the medical images are sampled continuously. Sequential images are input into parameter shared encoders for getting feature maps, which are then fused by a fusion block. A V$\Lambda$-block is structured to parse the fused feature map to extract the hidden continuity information. The reconstructed feature map is fed into a decoder for generating segmentation masks. Experiments on three datasets show MEPDNet outperforms other state-of-the-art segmentation models while using the least parameters.

In recent years, generative adversarial networks (GANs) have gained tremendous popularity for various imaging related tasks such as artificial image generation to support AI training. GANs are especially useful for medical imaging-related tasks where training datasets are usually limited in size and heavily imbalanced against the diseased class. We present a systematic review, following the PRISMA guidelines, of recent GAN architectures used for medical image analysis to help the readers in making an informed decision before employing GANs in developing medical image classification and segmentation models. We have extracted 54 papers that highlight the capabilities and application of GANs in medical imaging from January 2015 to August 2020 and inclusion criteria for meta-analysis. Our results show four main architectures of GAN that are used for segmentation or classification in medical imaging. We provide a comprehensive overview of recent trends in the application of GANs in clinical diagnosis through medical image segmentation and classification and ultimately share experiences for task-based GAN implementations.

Medical image segmentation tasks hitherto have achieved excellent progresses with large-scale datasets, which empowers us to train potent deep convolutional neural networks (DCNNs). However, labeling such large-scale datasets is laborious and error-prone, which leads the noisy (or incorrect) labels to be an ubiquitous problem in the real-world scenarios. In addition, data collected from different sites usually exhibit significant data distribution shift (or domain shift). As a result, noisy label and domain shift become two common problems in medical imaging application scenarios, especially in medical image segmentation, which degrade the performance of deep learning models significantly. In this paper, we identify a novel problem hidden in medical image segmentation, which is unsupervised domain adaptation on noisy labeled data, and propose a novel algorithm named "Self-Cleansing Unsupervised Domain Adaptation" (S-CDUA) to address such issue. S-CUDA sets up a realistic scenario to solve the above problems simultaneously where training data (i.e., source domain) not only shows domain shift w.r.t. unsupervised test data (i.e., target domain) but also contains noisy labels. The key idea of S-CUDA is to learn noise-excluding and domain invariant knowledge from noisy supervised data, which will be applied on the highly corrupted data for label cleansing and further data-recycling, as well as on the test data with domain shift for supervised propagation. To this end, we propose a novel framework leveraging noisy-label learning and domain adaptation techniques to cleanse the noisy labels and learn from trustable clean samples, thus enabling robust adaptation and prediction on the target domain. Specifically, we train two peer adversarial networks to identify high-confidence clean data and exchange them in companions to eliminate the error accumulation problem and narrow the domain gap simultaneously. In the meantime, the high-confidence noisy data are detected and cleansed in order to reuse the contaminated training data. Therefore, our proposed method can not only cleanse the noisy labels in the training set but also take full advantage of the existing noisy data to update the parameters of the network. For evaluation, we conduct experiments on two popular datasets (REFUGE and Drishti-GS) for optic disc (OD) and optic cup (OC) segmentation, and on another public multi-vendor dataset for spinal cord gray matter (SCGM) segmentation. Experimental results show that our proposed method can cleanse noisy labels efficiently and obtain a model with better generalization performance at the same time, which outperforms previous state-of-the-art methods by large margin. Our code can be found at https://github.com/zzdxjtu/S-cuda.

DEAttack: A differential evolution based attack method for the robustness evaluation of medical image segmentation

Federated learning is a distributed learning framework designed to facilitate collaborative training of deep learning models at the network edge, ensuring the privacy and security of users while effectively responding to the challenges posed by data silos within the healthcare domain. However, deploying intricate medical segmentation network models in realistic settings often leads to substantial communication overhead for federated learning clients, thereby severely undermining the operational efficiency of federated systems. In light of these issues, we propose an innovative federated learning approach known as FedSKD, which is intended to alleviate the significant communication overhead incurred during medical image segmentation tasks in federated learning. This method strikes a delicate balance between communication costs and model training accuracy. A key feature of the FedSKD framework is incorporating a self-knowledge distillation mechanism, which encompasses both prediction map distillation loss and feature map distillation loss. This integration plays a key role in enhancing model accuracy without the need for training an additional resource-intensive mentor model, thereby optimizing computational efficiency. Additionally, we introduce parameter difference dynamic compression technology to compress communications, resulting in a reduction of communication costs by more than 30% within the federated system. Furthermore, extensive experiments utilizing FedSKD were conducted on widely recognized public medical segmentation datasets such as KiTS19 and COVID-19. The experimental findings unequivocally demonstrate that our FedSKD not only significantly diminishes the expenses associated with federated communication but also enhances overall model accuracy.

Medical image segmentation is a critical task in computer-assisted diagnosis and therapy planning. Accurate and efficient fully automated segmentation methods are essential for accelerating decision-making processes. Compared with natural image segmentation methods, current deep learning–based segmentation models for medical imaging face challenges related to data scarcity, inconsistent annotations, and high computational cost. Consequently, this thesis aims to develop efficiency and generalised medical image segmentation models, which are fundamental to the clinical adoption and industrial deployment of this technology in healthcare settings. This thesis systematically addresses these challenges through a series of progressively advanced deep learning models. The foundational work, CTranS, forms a baseline medical image segmentation model that combines a convolutional neural network (CNN) based encoder with a Transformer-based decoder. This initial contribution leverages the strengths of CNNs for local feature extraction and Transformers for global contextual understanding, achieving state-of-the-art (SOTA) performance. Subsequently, to enhance the generalisability, MO-CTranS is introduced, which is capable of learning from a mixture of multiple partially labelled datasets. MO-CTranS employs a CNN encoder and Transformer decoder with task-specific tokens to handle label inconsistencies and class imbalances, tackling the challenges of training a single unified model from fragmented and heterogeneously annotated medical data. This innovation moves beyond single-task, fully-annotated model training towards a more scalable and clinically realistic approach. Despite the superior performance of CTranS and MO-CTranS, the computational efficiency of the Transformer blocks hinders their clinical application. Thus, CRFTrans is proposed to enhance the model efficiency. As a novel recursive Vision Transformer layer designed to replace the commonly used cascaded Transformer layers, this approach represents a fundamental shift from simply stacking layers to a recursive refinement process, demonstrating that comparable performance can be achieved with significantly reduced computational complexity, thus facilitating deployment in resource-constrained clinical environments. However, these methods still depend on extensive manual annotation, which remains challenging and time-consuming in clinical practice. Thus, to further enhance both efficiency and generality, a novel semi-supervised medical image segmentation framework, referred to as SSL-MedSAM2, has been developed. SSL-MedSAM2 integrates a training-free few-shot segmentation branch for pseudo labelling and a fully supervised learning branch for label refinement. This design allows the model to leverage a small fraction of labelled images together with a large unlabelled set, dramatically decreasing the annotation effort. Together, these contributions result in deep learning models that are significantly more data-efficient and broadly applicable than conventional approaches. The overall impact of this work is to facilitate the reliable and scalable deployment of segmentation models in real-world clinical settings by bridging the gap among accuracy, efficiency, and generalisability.

AMAKD: Adversarial multi-modal attention-based knowledge distillation for robust behaviour anomaly detection in real-world environments

MSKD: Structured knowledge distillation for efficient medical image segmentation

The encoder-decoder-based deep convolutional neural networks (CNNs) have made great improvements in medical image segmentation tasks. However, due to the inherent locality of convolution, CNNs generally are demonstrated to have limitations in obtaining features across layers and long-range features from the medical image. In this study, we develop a local-long range hybrid features network (LLRHNet), which inherits the merits of the iterative aggregation mechanism and the transformer technology, as a medical image segmentation model. LLRHNet adopts encoder-decoder architecture as the backbone which iteratively aggregates the projection and up-sampling to fuse local low-high resolution features across isolated layers. The transformer adopts the multi-head self-attention mechanism to extract long-range features from the tokenized image patches and fuses these features with the local-range features extracted by down-sampling operation in the backbone network. These hybrid features are used to assist the cascaded up-sampling operations to local the position of the target tissues. LLRHNet is evaluated on two multiple lesions medical image data sets, including a public liver-related segmentation data set (3DIRCADb) and an in-house stroke and white matter hyperintensity (SWMH) segmentation data set. Experimental results denote that LLRHNet achieves state-of-the-art performance on both data sets.

Low-level texture feature/knowledge is also of vital importance for characterizing the local structural pattern and global statistical properties, such as boundary, smoothness, regularity, and color contrast, which may not be well addressed by high-level deep features. In this paper, we aim to re-emphasize the low-level texture information in deep networks for semantic segmentation and related knowledge distillation tasks. To this end, we take full advantage of both structural and statistical texture knowledge and propose a novel Structural and Statistical Texture Knowledge Distillation (SSTKD) framework for semantic segmentation. Specifically, Contourlet Decomposition Module (CDM) is introduced to decompose the low-level features with iterative Laplacian pyramid and directional filter bank to mine the structural texture knowledge, and Texture Intensity Equalization Module (TIEM) is designed to extract and enhance the statistical texture knowledge with the corresponding Quantization Congruence Loss (QDL). Moreover, we propose the Co-occurrence TIEM (C-TIEM) and generic segmentation frameworks, namely STLNet++ and U-SSNet, to enable existing segmentation networks to harvest the structural and statistical texture information more effectively. Extensive experimental results on three segmentation tasks demonstrate the effectiveness of the proposed methods and their state-of-the-art performance on seven popular benchmark datasets, respectively.

Numerous transformer-based medical image segmentation methods have been proposed and achieved good segmentation results. However, it is still a challenge to train and deploy transformer networks to mobile medical devices due to a large number of model parameters. To resolve the training and model parameter problems, in this paper, we propose a Transformer-based network for Medical Image Segmentation using Knowledge Distillation named MISTKD. The MISTKD consists of a teacher network and a student network. It achieves comparable performance to state-of-the-art transformer works using fewer parameters by employing the teacher network to train the student network. The training can be implemented by extracting the sequence in the teacher and student encoder networks during the training procedure. The losses between sequences are further calculated, thus the student network can learn from the teacher network. The experimental results on Synapse show that the proposed work achieves competitive performance using only one-eighth parameters.

RSKD: Enhanced medical image segmentation via multi-layer, rank-sensitive knowledge distillation in Vision Transformer models