Parkinson’s disease (PD) is a prevalent neurodegenerative disorder globally. The eye’s retina is an extension of the brain, and clinical evidence has suggested the great potential of retinal pathology as surrogate biomarkers for early PD diagnosis. In particular, recent studies have shown that texture features extracted from retinal layers based on optical coherence tomography (OCT) images are strongly associated with PD-related retinal pathology. Additionally, frequency domain learning techniques can improve the representational capabilities of deep neural networks (DNNs) by decomposing frequency components that involve rich texture features, which remain underexplored for automated early PD diagnosis in OCT. To bridge this gap, we propose an Adaptive Wavelet Filter (AWF) that serves as the Practical Texture Amplifier, which fully leverages the merits of texture features from the retinal pathology view. Specifically, AWF first enhances feature map diversities and refines feature representations via channel mixer, then emphasizes informative texture feature representations with the well-designed adaptive wavelet filtering token mixer with the aid of frequency domain learning. By embedding AWFs into the DNN stem, AWFNet is constructed for automated early PD screening from OCT images. Additionally, we introduce a novel Balanced Confidence (BC) loss to boost early PD screening performance and trustworthiness of AWFNet, by mining the potential of sample-wise predicted probabilities across all classes and class frequency prior. The extensive experiments manifest the superiority of AWFNet with BC over state-of-the-art methods in terms of early PD screening performance and trustworthiness.

Photovoltaic (PV) panel segmentation from high-resolution remote sensing imagery is a critical task in advancing low-carbon energy monitoring, yet it faces challenges from complex backgrounds, inter-class confusion, and boundary dispersion. Existing methods leveraging convolutional neural networks and transformers have improved segmentation performance but struggle with misidentification of PV-like background objects and imprecise boundary delineation under varying environmental conditions. To address these issues, we propose an enhanced semantic segmentation model integrating U-Net with a transformer architecture, incorporating an adaptive hierarchical feature fusion module (AHFM) and an inter-class feature disambiguation module (IFDM). The AHFM combines multi-scale local and global features through attention mechanisms to mitigate boundary dispersion, and the IFDM employs channel and spatial attention with depth-separable convolutions to reduce misidentification caused by inter-class confusion. Extensive experiments on a self-built PV dataset confirm the model's effectiveness. In comparative experiments, the Intersection over Union (IoU) of this model reaches 89.73%, outperforming the second-best model by 2.49% while reducing computational complexity by 18.6%. Visualization results demonstrate that, compared with the masks generated by the comparison models, this model effectively suppresses the false feature activation of similar objects and achieves pixel-level adherence to target boundaries. Ablation experiments indicate that with an increase of only 6.88M parameters, the synergy of the dual modules improves the IoU of the baseline model by 3.59%. Furthermore, the model demonstrates good robustness, with an IoU standard deviation of only +/- 2.72% across 10 independent runs, and it maintains stable, high-precision convergence under substantial variations in core hyperparameters.

The rapid growth of renewable energy underscores the importance of accurately extracting photovoltaic (PV) panels from remote sensing images. However, existing methods often suffer from incomplete segmentation in distributed PV regions and blurred boundary representations. To address these issues, this paper proposes a Multi-branch Wavelet Network (MWNet). MWNet employs a multi-branch decoder with adaptive weighting to enhance multi-scale feature fusion. It further integrates a wavelet upsampling module to preserve high-frequency details, improving boundary accuracy and structural completeness. Experiments on mainstream PV datasets demonstrate that MWNet achieves superior segmentation accuracy and robustness, particularly in detecting distributed PV panels and delineating fine-grained edges.

Efficient convolutional neural network (CNN) architecture design has attracted growing research interests. However, they typically apply single receptive field (RF), small asymmetric RFs, or pyramid RFs to learn different feature representations, still encountering two significant challenges in medical image classification tasks: i) They have limitations in capturing diverse lesion characteristics efficiently, e.g., tiny, coordination, small and salient, which have unique roles on the classification results, especially imbalanced medical image classification. ii) The predictions generated by those CNNs are often unfair/biased, bringing a high risk when employing them to real-world medical diagnosis conditions. To tackle these issues, we develop a new concept, Expert-Like Reparameterization of Heterogeneous Pyramid Receptive Fields (ERoHPRF), to simultaneously boost medical image classification performance and fairness. This concept aims to mimic the multi-expert consultation mode by applying the well-designed heterogeneous pyramid RF bag to capture lesion characteristics with varying significances effectively via convolution operations with multiple heterogeneous kernel sizes. Additionally, ERoHPRF introduces an expertlike structural reparameterization technique to merge its parameters with the two-stage strategy, ensuring competitive computation cost and inference speed through comparisons to a single RF. To manifest the effectiveness and generalization ability of ERoHPRF, we incorporate it into mainstream efficient CNN architectures. The extensive experiments show that our proposed ERoHPRF maintains a better trade-off than state-of-the-art methods in terms of medical image classification, fairness, and computation overhead. The code of this paper is available at https://github.com/XiaoLing12138/Expert-Like-Reparameterization-of-Heterogeneous-Pyramid-Receptive-Fields.

To solve the problems of difficulty in multi-source data fusion and insufficient mining of timing dependencies in high-strength steel laser welding quality prediction, this study proposes a high-strength steel laser welding quality prediction method that jointly improves the back propagation neural network and K-means clustering. This study conducts pattern analysis through K-means clustering to divide historical parameters into representative process categories, providing a basis for parameter optimization. It uses empirical mode decomposition to adaptively decompose and denoise monitoring signals during the welding process, and extract multi-scale time-frequency features. This research uses long short-term memory to capture the timing dependencies in monitoring signals. Finally, it integrates static and dynamic features through a back-propagation neural network with improved structure (introducing ReLU and Sigmoid activation functions) to achieve high-precision prediction of molten pool geometric dimensions (bead width, penetration depth, and reinforcement height). Experimental results showed that this method performed well in multiple dimensions: In terms of convergence, the loss value stabilized at 0.010 after 500 rounds of training. In terms of prediction accuracy, the average absolute error under 500 groups of samples was 0.03.

Accurate depth estimation is crucial for 3D reconstruction and precise navigation in ophthalmic fundus surgery. However, acquiring annotated data remains challenging due to the impracticality of depth sensors under surgical microscopes.To overcome this limitation, we introduce RetinalDepth-64K, a novel synthetic dataset comprising 64,000 stereo image pairs across 1,280 diverse scenes, developed through a Real2Sim2Real pipeline that transforms real-world fundus surgery videos into synthetic data and facilitates model deployment in real scenarios. We analyzed key characteristics such as intricate retinal textures from real-world videos to guide the Real-to-Sim phase, enabling realistic data synthesis.To improving dataset fidelity for depth estimation, we created 3D eye models using Blender with ultra-wide-field retinal textures, glass-modeled aqueous humor, and dynamic instrument trajectories, enhanced by post-processing to ensure photorealism.The dataset provides RGB images, depth maps, normal maps, and instrument segmentation masks from binocular view, supporting the training of monocular, binocular, and video-based depth estimation models to enhance robustness. In the Sim-to-Real phase, quantitative and qualitative experiments show that finetuning foundation models with RetinalDepth-64K produces accurate depth predictions for synthetic data. Comparative analysis on results of zeroshot and finetuned models further validates robust generalization to real fundus surgery scenes, offering significant potential to enhance surgical precision and support the training of novice surgeons through reliable depth cues.As the first dataset of its kind for retinal surgery, RetinalDepth-64K offers a vital resource for advancing 3D reconstruction and surgical navigation in ophthalmology.

Precise diabetic retinopathy (DR) grading is essential for developing personalized and effective treatment plans. Although deep neural networks (DNNs) have achieved promising DR grading results, constructing a precise and trustworthy DR grading model remains challenging due to limited high-quality medical image data, high computational costs, and imbalanced data distributions. To tackle these challenges, we explore the transferability of feature representations from pre-trained vision foundation models (VFMs) to fundus images through adapter learning, aiming to build an efficient imbalanced DR grading model. Unlike classical full-tuning, which fine-tunes all pre-trained parameters of VFMs, adapter learning achieves competitive performance by adding negligible finetuned parameter number. Motivated by the above analysis, we develop a Token Pyramid Pooling-Driven Style Adapter Learning (TPDSAL) to better capture task-specific feature representations from VFMs, which fully exploits pathological distribution prior of DR and the inherent fundus imaging characteristics. Besides, we propose a novel dual-view balanced loss (DVB) to improve imbalanced DR grading performance and trustworthiness, which explores the potential of training class frequencies in sample-wise predicted logit space and sample-wise loss value space simultaneously. Extensive experiments on four public fundus image datasets manifest the superiority of our TPDSAL with DVB over competitive transfer tuning and loss methods in terms of imbalanced grading performance and trustworthiness. Further analysis suggests that clinical prior knowledge utilization is beneficial for adapter learning in capturing task-specific feature representations from VFMs.

Surgical scene reconstruction is a critical prerequisite for intraoperative navigation in robotic surgery. Although deep neural networks (DNNs) have advanced surgical scene reconstruction, their performance degrades severely when confronted with texture-sparse biological tissues and dynamic illumination changes. To this end, we develop an end-to-end SurgFM-SLAM framework for surgical scene reconstruction by exploring geometric representation priors of pretrained 3D foundation models (FMs) and SLAM. To be specific, we first design a frame-similarity sampling strategy to keep surgical scene consistency among sampled frames, and then develop a surgical foundation model (SurgFM) by employing a low-rank adaptation (LoRA) method to mine 3D strong structural geometry representations from 3D FMs with the aid of parameter-efficient finetuning techniques, aiming to effectively capture dynamic illumination conditions and informative textures from surgical environments. Finally, we embed SurgFM into the SLAM backend to construct SurgFM-SLAM to perform robust surgical scene reconstruction in an end-to-end manner, including tracking, mapping, and relocalization. Extensive experiments on the SimCol dataset demonstrate that SurgFM-SLAM achieves competitive performance across depth estimation, camera pose estimation, and 3D reconstruction through comparisons to state-of-the-art methods. Additionally, zero-shot generalization test on the C3VD and SCARED datasets manifests the generalization of SurgFM-SLAM. The project of this paper is available at: https://gumlau.github.io/SurgFM-SLAM/ .

Hyperspectral image classification (HSIC) plays a key role in remote sensing, but the interpretation of scene features with limited samples remains challenging. In this letter, we propose a multiscale spectral morphological fusion network (MSMFNet) for unsupervised domain adaptation (UDA) in HSIC. This method consists of several key steps. First, the network uses principal component analysis (PCA) to extract spectral features, which are then combined with extended morphological profiles (EMPs) to enhance spatial structure representation. Then, a multiscale heterogeneous feature aggregation (MHFA) module is introduced to capture heterogeneous features across different scales and directions. Next, the multiscale global attention (MGA) module generates multilevel responses, exploring correlations between local and global information, thereby improving the representation of fine-grained features and boosting cross-domain feature fusion and classification performance. Finally, contrastive learning (CL) is applied to efficiently extract domain-invariant features. The experimental results demonstrate that MSMFNet achieves superior performance in cross-domain adaptation and fine-grained feature discrimination, achieving accuracies of 77.48% on the Houston dataset and 93.66% on the Pavia dataset, with Kappa coefficients that exceed the state of the art by 2.46 and 2.48, respectively.

Intracranial aneurysm (IA) lesion segmentation is significant for its treatment and prognosis. Although exiting deep network-based instance methods have good IA lesion segmentation results based on digital subtraction angiography (DSA) images, they still face great challenges with instance confidence bias and imprecise boundary segmentation, which may negatively affect IA diagnosis. To tackle these problems, this paper proposes a novel graph confidence intercalibration network (GCINet) to automatically segment IA lesions from DSA images. To be specific, we design a graph confidence intercalibration (GCI) module to mitigate instance confidence bias by dynamically adjusting their confidence distributions. At the same time, we propose an edge space perception (ESP) module to correct ambiguous segmentation boundaries. Extensive experiments on a clinical IA-DSA and a publicly available LiTS dataset demonstrate that our GCINet outperforms state-of-the-art methods. Additionally, visual analysis and ablation studies are provided to verify the effectiveness of each module in GCINet.

Deformable retinal image registration is crucial in clinical diagnosis and longitudinal studies of retinal diseases. Most existing deep deformable retinal image registration methods focus on fully convolutional network (FCN) architecture design, which fails to model long-range dependencies among pixels – a significant factor in deformable retinal image registration. Transformers based on the self-attention mechanism, can capture global context dependencies, complementing local convolution. However, multi-scale spatial feature fusion and pixel-wise position selection are also crucial for the deformable retinal image registration, are often ignored by both FCNs and transformers. To fully leverage the merits of FCNs, multi-scale spatial attention and transformers, we propose a hierarchical hybrid architecture, reparameterized multi-scale transformer (RMFormer), for deformable retinal image registration. In RMFormer, we specifically develop a reparameterized multi-scale spatial attention to adaptively fuse multi-scale spatial features, with the assistance of the reparameterizing technique, thereby highlighting informative pixel-wise positions in a lightweight manner. The experimental results on two publicly available datasets demonstrate the superiority of our RMFormer over state-of-the-art methods and show that it is data-efficient in a limited medical image regime. Additionally, we are the first to provide a visualization analysis to explain how our proposed method affects the deformable retinal image registration process. The source code of our work is available at https://github.com/Tloops/RMFormer .

As an affordable and convenient eye scan, fundus photography holds the potential for preventing vision impairment, especially in resource-limited regions. However, fundus image degradation is common under intricate imaging environments, impacting following diagnosis and treatment. Consequently, image quality assessment (IQA) and enhancement (IQE) are essential for ensuring the clinical value and reliability of fundus images. While existing reviews offer some overview of this field, a comprehensive analysis of the interplay between IQA and IQE, along with their clinical deployment challenges, is lacking. This paper addresses this gap by providing a thorough review of fundus IQA and IQE algorithms, research advancements, and practical applications. We outline the fundamentals of the fundus photography imaging system and the associated interferences, and then systematically summarize the paradigms in fundus IQA and IQE. Furthermore, we discuss the practical challenges and solutions in deploying IQA and IQE, as well as offer insights into potential future research directions.

Retinal image registration is vital for diagnostic therapeutic applications within the field of ophthalmology. Existing public datasets, focusing on adult retinal pathologies with high-quality images, have limited number of image pairs and neglect clinical challenges. To address this gap, we introduce COph100, a novel and challenging dataset known as the Comprehensive Ophthalmology Retinal Image Registration dataset for infants with a wide range of image quality issues constituting the public "RIDIRP" database. COph100 consists of 100 eyes, each with 2 to 9 examination sessions, amounting to a total of 491 image pairs carefully selected from the publicly available dataset. We manually labeled the corresponding ground truth image points and provided automatic vessel segmentation masks for each image. We have assessed COph100 in terms of image quality and registration outcomes using state-of-the-art algorithms. This resource enables a robust comparison of retinal registration methodologies and aids in the analysis of disease progression in infants, thereby deepening our understanding of pediatric ophthalmic conditions.

Cataract is the leading ocular disease of blindness and visual impairment globally. Deep neural networks (DNNs) have achieved promising cataracts recognition performance based on anterior segment optical coherence tomography (AS-OCT) images; however, they have poor explanations, limiting their clinical applications. In contrast, visual features extracted from original AS-OCT images and their transform forms (e.g., AS-OCT-based histograms) have good explanations but have not been fully exploited. Motivated by these observations, an explainable machine learning framework to recognize cataracts severity levels automatically using AS-OCT images was proposed, consisting of three stages: visual feature extraction, feature importance explanation and selection, and recognition. First, the intensity histogram and intensity-based statistical methods are applied to extract visual features from original AS-OCT images and AS-OCT-based histograms. Subsequently, the SHapley Additive exPlanations and Pearson correlation coefficient methods are applied to analyze the feature importance and select significant visual features. Finally, an ensemble multi-class ridge regression method is applied to recognize the cataracts severity levels based on the selected visual features. Experiments on a clinical AS-OCT-NC dataset demonstrate that the proposed framework not only achieves competitive performance through comparisons with DNNs, but also has a good explanation ability, meeting the requirements of clinical diagnostic practice.

The morphologies of vessel-like structures, such as blood vessels and nerve fibres, play significant roles in disease diagnosis, e.g., Parkinson’s disease. Although deep network-based refinement segmentation and topology-preserving segmentation methods recently have achieved promising results in segmenting vessel-like structures, they still face two challenges: (1) existing methods often have limitations in rehabilitating subsection ruptures in segmented vessel-like structures; (2) they are typically overconfident in predicted segmentation results. To tackle these two challenges, this paper attempts to leverage the potential of spatial interconnection relationships among subsection ruptures from the structure rehabilitation perspective. Based on this perspective, we propose a novel Vessel-like Structure Rehabilitation Network (VSR-Net) to both rehabilitate subsection ruptures and improve the model calibration based on coarse vessel-like structure segmentation results. VSR-Net first constructs subsection rupture clusters via a Curvilinear Clustering Module (CCM). Then, the well-designed Curvilinear Merging Module (CMM) is applied to rehabilitate the subsection ruptures to obtain the refined vessel-like structures. Extensive experiments on six 2D/3D medical image datasets show that VSR-Net significantly outperforms state-of-the-art (SOTA) refinement segmentation methods with lower calibration errors. Additionally, we provide quantitative analysis to explain the morphological difference between the VSR-Net’s rehabilitation results and ground truth (GT), which are smaller compared to those between SOTA methods and GT, demonstrating that our method more effectively rehabilitates vessel-like structures.

Precise segmentation and uncertainty estimation are crucial for error identification and correction in medical diagnostic assistance. Existing methods mainly rely on pixel-wise uncertainty estimations. They (1) neglect the global context, leading to erroneous uncertainty indications, and (2) bring attention interference, resulting in the waste of extensive details and potential understanding confusion. In this paper, we propose a novel structural uncertainty estimation method, based on Convolutional Neural Networks (CNN) and Active Shape Models (ASM), named SU-ASM, which incorporates global shape information for providing precise segmentation and uncertainty estimation. The SU-ASM consists of three components. Firstly, multi-task generation provides multiple outcomes to assist ASM initialization and shape optimization via a multi-task learning module. Secondly, information fusion involves the creation of a Combined Boundary Probability (CBP) and along with a rapid shape initialization algorithm, Key Landmark Template Matching (KLTM), to enhance boundary reliability and select appropriate shape templates. Finally, shape model fitting where multiple shape templates are matched to the CBP while maintaining their intrinsic shape characteristics. Fitted shapes generate segmentation results and structural uncertainty estimations. The SU-ASM has been validated on cardiac ultrasound dataset, ciliary muscle dataset of the anterior eye segment, and the chest X-ray dataset. It outperforms state-of-the-art methods in terms of segmentation and uncertainty estimation.

Salient and small lesions (e.g., microaneurysms on fundus) both play significant roles in real-world disease diagnosis under medical image examinations. Although deep neural networks (DNNs) have achieved promising medical image classification performance, they often have limitations in capturing both salient and small lesion information, restricting performance improvement in imbalanced medical image classification. Recently, with the advent of DNN-based style transfer in medical image generation, the roles of clinical styles have attracted great interest, as they are crucial indicators of lesions. Motivated by this observation, we propose a novel Adaptive Dual-Axis Style-based Recalibration (ADSR) module, leveraging the potential of clinical styles to guide DNNs in effectively learning salient and small lesion information from a dual-axis perspective. ADSR first emphasizes salient lesion information via global style-based adaptation, then captures small lesion information with pixel-wise style-based fusion. We construct an ADSR-Net for imbalanced medical image classification by stacking multiple ADSR modules. Additionally, DNNs typically adopt cross-entropy loss for parameter optimization, which ignores the impacts of class-wise predicted probability distributions. To address this, we introduce a new Class-wise Statistics Loss (CWS) combined with CE to further boost imbalanced medical image classification results. Extensive experiments on five imbalanced medical image datasets demonstrate not only the superiority of ADSR-Net and CWS over state-of-the-art (SOTA) methods but also their improved confidence calibration results. For example, ADSR-Net with the proposed loss significantly outperforms CABNet50 by 21.39% and 27.82% in F1 and B-ACC while reducing 3.31% and 4.57% in ECE and BS on ISIC2018.

ABSTRACT Given that the retina shares embryonic origin with the central nervous system, past evidence has attempted to prove parallel pathology of neurovegetative diseases in the retina. Retinal imaging techniques provide in vivo structural and functional data with advantages of high resolution and low cost in a noninvasive way. In recent literature, the retina presents significant alterations related to Parkinson's disease (PD) and other neurodegenerative diseases through multimodal retinal images due to its neural accessibility. However, current findings remain fragmented and inconsistent causing the existing reviews limited in a certain modality or incomplete without mention of interdisciplinary integration. The purpose of this paper is to systematically review and synthesize the application of retinal imaging techniques in PD and other neurodegenerative diseases. We begin with the connections between the retina and the brain and the main retinal imaging modalities. We then summarize the retinal changes in patients with PD and other neurodegenerative diseases during the past decade. Additionally, we discuss the application of artificial intelligence in PD prediction and retinal probe. By integrating histopathological insights with advanced imaging analytics, we highlight retinal changes as biomarkers for neurodegeneration, which accelerate their clinical translation for early diagnosis and monitoring of PD in the future. and other neurodegenerative diseases.

Precise multi-class retinal disease recognition faces challenges from inter/intra-class variations and imbalanced distributions. While Convolution Neural Network (CNNs) effectively capture salient lesions, they struggle with subtle lesions and exhibit bias toward frequent diseases. We propose a Retinal Lesion Fusion Network (RLF-Net) with two novel modules: a Retinal Lesion Feature Fusion (RLFF) module combining a SAlient Lesion Enhancement (SALE) block, SUbtle Lesion Enhancement (SULE) block, and Fast Fourier Transform Fusion (FFTF) block to adaptively integrate multi-scale lesion features and a Retinal Screening of Diseases (RSD) module mitigating class imbalance by equally weighting disease-specific feature differences. Additionally, we design a hybrid loss merging supervised contrastive learning and cross-entropy to enhance discriminative power. Evaluations on a clinical Fundus Fluorescein Angiography (FFA) dataset and two public fundus benchmarks demonstrate RLF-Net's superiority over state-of-the-art methods. Our approach advances multi-class retinal diagnosis by addressing critical limitations in feature representation and class imbalance, particularly improving recognition of subtle lesions and rare diseases through synergistic feature fusion and balanced optimization strategies.

Pathology context and expert experience play significant roles in clinical ocular disease diagnosis. Although deep neural networks (DNNs) have good ocular disease recognition results, they often ignore exploring the clinical pathology context and expert experience priors to improve ocular disease recognition performance and decision-making interpretability. To this end, we first develop a novel Pathology Recalibration Module (PRM) to leverage the potential of pathology context prior via the combination of the well-designed pixel-wise context compression operator and pathology distribution concentration operator; then this paper applies a novel expert prior Guidance Adapter (EPGA) to further highlight significant pixel-wise representation regions by fully mining the expert experience prior. By incorporating PRM and EPGA into the modern DNN, the PCRNet is constructed for automated ocular disease recognition. Additionally, we introduce an Integrated Loss (IL) to boost the ocular disease recognition performance of PCRNet by considering the effects of sample-wise loss distributions and training label frequencies. The extensive experiments on three ocular disease datasets demonstrate the superiority of PCRNet with IL over state-of-the-art attention-based networks and advanced loss methods. Further visualization analysis explains the inherent behavior of PRM and EPGA that affects the decision-making process of DNNs.