Medical image segmentation is crucial for medical image processing and the development of computer-aided diagnostics. In recent years, deep Convolutional Neural Networks (CNNs) have been widely adopted for medical image segmentation and have achieved significant success. UNet, which is based on CNNs, is the mainstream method used for medical image segmentation. However, its performance suffers owing to its inability to capture long-range dependencies. Transformers were initially designed for Natural Language Processing (NLP), and sequence-to-sequence applications have demonstrated the ability to capture long-range dependencies. However, their abilities to acquire local information are limited. Hybrid architectures of CNNs and Transformer, such as TransUNet, have been proposed to benefit from Transformer’s long-range dependencies and CNNs’ low-level details. Nevertheless, automatic medical image segmentation remains a challenging task due to factors such as blurred boundaries, the low-contrast tissue environment, and in the context of ultrasound, issues like speckle noise and attenuation. In this paper, we propose a new model that combines the strengths of both CNNs and Transformer, with network architectural improvements designed to enrich the feature representation captured by the skip connections and the decoder. To this end, we devised a new attention module called Three-Level Attention (TLA). This module is composed of an Attention Gate (AG), channel attention, and spatial normalization mechanism. The AG preserves structural information, whereas channel attention helps to model the interdependencies between channels. Spatial normalization employs the spatial coefficient of the Transformer to improve spatial attention akin to TransNorm. To further improve the skip connection and reduce the semantic gap, skip connections between the encoder and decoder were redesigned in a manner similar to that of the UNet++ dense connection. Moreover, deep supervision using a side-output channel was introduced, analogous to BASNet, which was originally used for saliency predictions. Two datasets from different modalities, a CT scan dataset and an ultrasound dataset, were used to evaluate the proposed UNet architecture. The experimental results showed that our model consistently improved the prediction performance of the UNet across different datasets.

Prostate cancer is a global health concern, and early diagnosis plays a vital role in improving the survival rate. Accurate segmentation is a key step in the automated diagnosis of prostate cancer; however, manual segmentation remains time-consuming and challenging. Micro-Ultrasound (US) is particularly well-suited for prostate cancer detection, offering real-time imaging with a resolution comparable to that of MRI. This enables improved spatial resolution and detailed visualization of small anatomical structures. With recent advances in deep learning for medical image segmentation, precise prostate segmentation has become critical for biopsy guidance, disease diagnosis, and follow-up. However, segmentation of the prostate in micro-US images remains challenging due to indistinct boundaries between the prostate and surrounding tissue. In this work, we propose a model for precise micro-ultrasound image segmentation. The model employs a dual-encoder architecture that integrates Convolutional Neural Networks (CNN) and Transformer-based encoders in parallel, combined with a fusion module to capture both global dependencies and low-level spatial details. More importantly, we introduce a decoder based on Mamba v2 to enhance segmentation accuracy. A Hypergraph Neural Network (HGNN) is employed as a bridge between the dual encoders and Mamba decoder to model correlations among non-pairwise connections. Experimental results on micro-US datasets demonstrated that our model achieved superior or comparable performance to state-of-the-art methods, with a Dice score of 0.9416 and an HD95 of 1.93.