28 August 2023 in IEEE Transactions on Image Processing
Breast tumor segmentation of ultrasound images provides valuable information of tumors for early detection and diagnosis. Accurate segmentation is challenging due to low image contrast between areas of interest; speckle noises, and large inter-subject variations in tumor shape and size. This paper proposes a novel Multi-scale Dynamic Fusion Network (MDF-Net) for breast ultrasound tumor segmentation. It employs a two-stage end-to-end architecture with a trunk sub-network for multiscale feature selection and a structurally optimized refinement sub-network for mitigating impairments such as noise and inter-subject variation via better feature exploration and fusion. The trunk network is extended from UNet++ with a simplified skip pathway structure to connect the features between adjacent scales. Moreover, deep supervision at all scales, instead of at the finest scale in UNet++, is proposed to extract more discriminative features and mitigate errors from speckle noise via a hybrid loss function. Unlike previous works, the first stage is linked to a loss function of the second stage so that both the preliminary segmentations and refinement subnetworks can be refined together at training. The refinement sub-network utilizes a structurally optimized MDF mechanism to integrate preliminary segmentation information (capturing general tumor shape and size) at coarse scales and explores inter-subject variation information at finer scales. Experimental results from two public datasets show that the proposed method achieves better Dice and other scores over state-of-the-art methods. Qualitative analysis also indicates that our proposed network is more robust to tumor size/shapes, speckle noise and heavy posterior shadows along tumor boundaries. An optional post-processing step is also proposed to facilitate users in mitigating segmentation artifacts. The efficiency of the proposed network is also illustrated on the “Electron Microscopy neural structures segmentation dataset”. It outperforms a state-of-the-art algorithm based on UNet-2022 with simpler settings. This indicates the advantages of our MDF-Nets in other challenging image segmentation tasks with small to medium data sizes.
27 April 2024 in Biomedical Signal Processing and Control
26 July 2021 in IEEE Access
Person re-identification problems usually suffer from large subject appearance variations and limited training data. This paper proposes a novel physically motivated Color/Illuminance-Aware data-augmentation (CIADA) scheme and a style-adaptive fusion approach to address these issues. The CIADA scheme estimates the color/illuminance distribution from the training data via manifold learning and generates new samples under different color/illuminance perturbations to better capture objects’ appearance for mitigating the small-sample-size and color variation problems. A Color/Illuminance Aware Feature Augmentation (CIAFA) approach, which is applicable to state-of-the-art features and metric learning algorithms, is then proposed to integrate the features generated by the augmented samples for metric learning. A new Color/Illuminance-Aware Style Fusion (CIASF) scheme, which allows the learning and matching process to be performed independently on each pair of datasets generated for estimating a set of ‘local’ distance functions, is also proposed. A canonical correlation analysis-based weighting scheme is developed to fuse these local distances to an overall distance for recognition. This reduces the memory requirement and complexity over the original CIAFA. Experiments on common datasets show that the proposed methodologies substantially improve the performance of state-of-the-art subspace learning algorithms. It is applicable to both small and large datasets with hand-craft and deep features.