Abstract: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.

MLDF-Net: Metadata Based Multi-level Dynamic Fusion Network.

M-DFNet

Self-contrastive Feature Guidance Based Multidimensional Collaborative Network of Metadata and Image Features for Skin Disease Classification

MDFNet: application of multimodal fusion method based on skin image and clinical data to skin cancer classification

MLFF-Net: a multi-model late feature fusion network for skin disease classification

Mmy-net: a Multimodal Network Exploiting Image and Patient Metadata for Simultaneous Segmentation and Diagnosis

M4FNet: Multimodal Medical Image Fusion Network Via Multi-Receptive-field and Multi-Scale Feature Integration.

Fusing Metadata and Dermoscopy Images for Skin Disease Diagnosis

MCFNet: Multi-Layer Concatenation Fusion Network for Medical Images Fusion

MDFF-Net: A multi-dimensional feature fusion network for breast histopathology image classification

A novel multi-task learning network for skin lesion classification based on multi-modal clues and label-level fusion

MDC-RHT: Multi-Modal Medical Image Fusion via Multi-Dimensional Dynamic Convolution and Residual Hybrid Transformer

FusionM4Net: A Multi-Stage Multi-Modal Learning Algorithm for Multi-Label Skin Lesion Classification

AMFF-NET: Adaptive Multi-Layer Feature Fusion Network for Medical Image Segmentation

MLDNet: Multi-level Dense Network for Multi-Focus Image Fusion

Metadata-P2T-Net: a Skin Lesion Classification Network Integrating Dermoscopic Metadata and Image Data

MASDF-Net: A Multi-Attention Codec Network with Selective and Dynamic Fusion for Skin Lesion Segmentation

MM-Net: A MixFormer-Based Multi-Scale Network for Anatomical and Functional Image Fusion

Multi-Modal Fusion Learning for Cervical Dysplasia Diagnosis.

MDF-Net: A Multi-Scale Dynamic Fusion Network for Breast Tumor Segmentation of Ultrasound Images

DDIFN: A Dual-discriminator Multi-modal Medical Image Fusion Network