Recurrent Contrastive Learning progressively expands the support region of tail classes by recurrently reusing historical feature states across training phases and devise a Temporal Memory Queue (TMQ) to preserve corpus-level features across training phases and provide diversified global references for contrastive learning.
Abstract
Medical image classification often suffers from class imbalance due to the inherent disparities in disease incidence. Existing approaches, such as class resampling and loss reweighting, mainly improve learning within the observed feature distribution, but do not explicitly enlarge the latent support region of tail classes. As a result, tail-class representations remain overly compact and are easily encroached upon by head classes, leading to biased decision boundaries. In this work, we propose Recurrent Contrastive Learning (RCL) for imbalanced medical image classification. RCL progressively expands the support region of tail classes by recurrently reusing historical feature states across training phases. Specifically, we adopt DINOv3 with LoRA adapters as the backbone to provide robust feature embeddings. We then devise a Temporal Memory Queue (TMQ) to preserve corpus-level features across training phases and provide diversified global references for contrastive learning. Based on TMQ, we construct Temporal Anchors (TARs) to form an anchor field around tail classes. This field enlarges the support region of tail classes, suppresses head-class encroachment, and improves inter-class separation. Extensive experiments on three imbalanced medical datasets demonstrate that RCL achieves consistent improvements over strong baselines. The code is available at https://github.com/dndins/RCL.
Experiments show that IRRL achieves balanced classification performance, with favorable F1-score and Matthews Correlation Coefficient results that reflect improved minority-class recognition quality, and robustness and consistency of the proposed representation learning strategy.
Medical image datasets often suffer from limited training samples and severe class imbalance. This scarcity hinders deep learning models from learning representative features and leads to degraded performance on minority classes. In this work, we adapt the Unconditional Training at Lower Resolutions (UTLO) framework to long-tailed brain MRI data and conduct systematic evaluations on a brain tumor dataset. UTLO enables knowledge sharing between frequent and rare classes by training the generator unconditionally at lower resolutions to capture class-agnostic structures, while employing conditional generation at higher resolutions to refine class-specific details. Meanwhile, the discriminator is optimized with a multi-objective loss across low- and high-resolution stages to improve robustness and alleviate mode collapse under limited data. We evaluate the generated images using Frechet Inception Distance (FID), Kernel Inception Distance (KID), Inception Score (IS), and their few-shot variants, i.e., FID-FS (FID on few-shot/rare classes) and KID-FS (KID on few-shot/rare classes). Qualitative visualization is also provided to assess perceptual realism and diversity. Experimental results demonstrate that UTLO achieves stable generation performance across categories and yields noticeable improvements on rare classes.
Deep learning has shown strong potential in medical image analysis, but most existing methods rely on large-scale annotations and a closed-world assumption that rarely holds in clinical practice. Although Generalized Category Discovery (GCD) has advanced rapidly on natural images, it remains underexplored in medical imaging. To address this issue, we propose MedXplore, a unified framework for reliable and unbiased medical GCD, optimizing from both perceptual and decision levels. Specifically, at the perceptual level, taking a frequency domain perspective, Frequency-SNR Adaptive Attention and Consistency (FAAC) performs learnable full-spectrum filtering and global-local energy contrast activation to not only highlight local abnormal signals relative to the global context, but also provide reliable semantic anchors for patch consistency learning. At the decision level, Adaptive Cosine-Angular Margin (ACAM) adjusts angular margins using semantic difficulty and feature confidence to balance intra-class compactness and inter-class separability. Together, the two modules improve lesion-sensitive representation learning and mitigate old-class bias. Experiments on multiple benchmarks show an average \textbf{8.5\%} gain in \textit{All} accuracy over the strongest competing methods. On Kvasir, MedXplore reduces false-old errors from 14.50\% to 0.80\%, demonstrating strong robustness under severe old-new ambiguity.
The increasing availability of electronic clinical records has enabled new opportunities for predictive modeling in healthcare. However, clinical data are characterized by heterogeneous patient information, limited sample availability, and highly imbalanced outcome distributions, which may hinder the ability of predictive models to capture complex patterns and generalize across underrepresented groups. This work investigates deep learning-based representation learning strategies for predicting the final post-stroke functional state derived from the modified Rankin Scale (mRS) using clinical tabular data. The proposed methodology learns informative patient representations while addressing class imbalance without altering the original data distribution, avoiding limitations of conventional resampling strategies. Specifically, the proposed framework combines a TabTransformer encoder with Supervised Contrastive Learning and Focal Loss to jointly optimize discriminative representation learning and imbalance-aware classification. By obtaining more separable latent representations and reducing the bias toward majority-class predictions, the approach aims to improve the identification of patients with unfavorable functional outcomes. Experimental results show that the proposed model achieves the highest accuracy (0.92) and macro-averaged F1-score (0.80), with balanced minority-class precision (0.67) and recall (0.64), unlike the other models, which showed a trade-off between these metrics. These results demonstrate improved minority-class identification while maintaining overall predictive performance in clinical outcome prediction.
María N. Moreno-García, Araceli Rodríguez Vico, V. F. López Batista et al.· Information· 0 citations
Despite the growing number of public datasets, annotated medical images remain scarce. Supervised learning methods achieve strong performance on many benchmarks, however require large amounts of labeled data, which are costly and time-consuming to obtain in the medical domain. To address this limitation, contrastive self-supervised learning (SSL) has emerged as a promising alternative for learning useful representations from unlabeled data. In this work, we investigate two SSL frameworks, SimSiam and SimCLR, for retinal disease classification from fundus images. We focus on understanding how augmentation strategies and training parameters influence representation learning under resource-constrained settings. Given limited data and computational capacity, we explore the feasibility of training SSL models with small batch sizes incorporated with retinal-specific augmentation techniques. Through a series of experiments, we assess the quality of learned representations via linear evaluation and fine-tuning across downstream tasks, including multi-disease classification and diabetic retinopathy grading. Our results show that tailoring augmentation strategies to the characteristics of retinal images plays a critical role in improving performance. Even under constrained settings, lightweight SSL frameworks can learn transferable representations that reduce dependence on large annotated datasets and achieve competitive results.
Medical image analysis has undergone transformative progress with the application of deep learning models. However, existing architectures often struggle to effectively balance local feature extraction with global contextual understanding, which is crucial for complex diagnostic tasks such as Retinopathy of Prematurity (ROP) detection. In this study, we present a pretrained lightweight Conformer model tailored for medical image classification. The model integrates convolutional layers for capturing fine-grained spatial features with transformer blocks that capture long-range dependencies, creating a unified architecture capable of robust representation learning. We evaluate the model across multiple benchmark medical imaging datasets, including ROP, BloodMNIST, RetinalMNIST and other MedMNIST benchmark datasets. With 93.61% accuracy on the ROP dataset and 99.12% accuracy on BloodMNIST, experimental results show competitive classification performance while lowering model complexity to 12.4 million parameters and 3.2 GFLOPs. Experimental results demonstrate that the comparative studies versus CNN-based and transformer-based architectures, such as ResNet50, Swin-Tiny, ConvNeXt-Tiny, Vision Transformer, and MedViT. The findings show that in clinical settings with limited resources, the suggested lightweight Conformer offers a practical and computationally efficient alternative for medical image interpretation. Furthermore, the lightweight design ensures computational efficiency, making it suitable for deployment in resource-constrained healthcare environments. These findings validate the lightweight Conformer model’s potential for scalable, accurate, and real-time medical image classification.
Sreelekshmi Vijayasree, Adithya K. Krishna, Akarsh S. Nair et al.· Journal of Imaging· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
