Dataset Information

New hybrid segmentation algorithm: UNet-GOA.

ABSTRACT: The U-Net architecture is a prominent technique for image segmentation. However, a significant challenge in utilizing this algorithm is the selection of appropriate hyperparameters. In this study, we aimed to address this issue using an evolutionary approach. We conducted experiments on four different geometric datasets (triangle, kite, parallelogram, and square), with 1,000 training samples and 200 test samples. Initially, we performed image segmentation without the evolutionary approach, manually adjusting the U-Net hyperparameters. The average accuracy rates for the geometric images were 0.94463, 0.96289, 0.96962, and 0.93971, respectively. Subsequently, we proposed a hybrid version of the U-Net architecture, incorporating the Grasshopper Optimization Algorithm (GOA) for an evolutionary approach. This method automatically discovered the optimal hyperparameters, resulting in improved image segmentation performance. The average accuracy rates achieved by the proposed method were 0.99418, 0.99673, 0.99143, and 0.99946, respectively, for the geometric images. Comparative analysis revealed that the proposed UNet-GOA approach outperformed the traditional U-Net architecture, yielding higher accuracy rates.

SUBMITTER: Yousefi T

PROVIDER: S-EPMC10496000 | biostudies-literature | 2023

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

New hybrid segmentation algorithm: UNet-GOA.

Yousefi Tohid T Aktaş Özlem Ö

PeerJ. Computer science 20230808

The U-Net architecture is a prominent technique for image segmentation. However, a significant challenge in utilizing this algorithm is the selection of appropriate hyperparameters. In this study, we aimed to address this issue using an evolutionary approach. We conducted experiments on four different geometric datasets (triangle, kite, parallelogram, and square), with 1,000 training samples and 200 test samples. Initially, we performed image segmentation without the evolutionary approach, manua ...[more]

PMID: 37705637

Similar Datasets

Project description:Plant phenotypic traits play an important role in understanding plant growth dynamics and complex genetic traits. In phenotyping, the segmentation of plant organs, such as leaves and stems, helps in automatically monitoring growth and improving screening efficiency for large-scale genetic breeding. In this paper, we propose an AC-UNet stem and leaf segmentation algorithm based on an improved UNet. This algorithm aims to address the issues of feature edge information loss and sample breakage in the segmentation of plant organs, specifically in Betula luminifera. The method replaces the backbone feature extraction network of UNet with VGG16 to reduce the redundancy of network information. It adds a multi-scale mechanism in the splicing part, an optimized hollow space pyramid pooling module, and a cross-attention mechanism in the expanding network part at the output end to obtain deeper feature information. Additionally, Dice_Boundary is introduced as a loss function in the back-end of the algorithm to circumvent the sample distribution imbalance problem. The PSPNet model achieves mIoU of 58.76%, mPA of 73.24%, and Precision of 66.90%, the DeepLabV3 model achieves mIoU of 82.13%, mPA of 91.47%, and Precision of 87.73%, on the data set. The traditional UNet model achieves mIoU of 84.45%, mPA of 91.11%, and Precision of 90.63%, and the Swin-UNet model achieves . The mIoU is 79.02%, mPA is 85.99%, and Precision is 88.73%. The AC-UNet proposed in this article achieved excellent performance on the Swin-UNet dataset, with mIoU, mPA, and Precision of 87.50%, 92.71%, and 93.69% respectively, which are better than the selected PSPNet, DeepLabV3, traditional UNet, and Swin-UNet. Commonly used semantic segmentation algorithms. Experiments show that the algorithm in this paper can not only achieve efficient segmentation of the stem and leaves of Betula luminifera but also outperforms the existing state-of-the-art algorithms in terms of both speed. This can provide more accurate auxiliary support for the subsequent acquisition of plant phenotypic traits.

Project description:BackgroundMedical image segmentation is crucial for improving healthcare outcomes. Convolutional neural networks (CNNs) have been widely applied in medical image analysis; however, their inherent inductive biases limit their ability to capture global contextual information. Vision transformer (ViT) architectures address this limitation by leveraging attention mechanisms to model global relationships; however, they typically require large-scale datasets for effective training, which is challenging in the field of medical imaging due to limited data availability. This study aimed to integrate the advantages of CNN and ViT architectures to improve segmentation performance on small-scale medical image datasets.MethodsIn this study, we established a U-shaped network architecture based on a Transformer-assisted convolutional neural network (TAC-UNet). The TAC-UNet is primarily composed of a hybrid structure integrating CNN and Transformer components. Specifically, the hybrid architecture follows a dual-path design in which the Transformer branch continuously conveys global contextual information to the CNN backbone. This allows the CNN backbone to enhance its global perception while building on the local features it extracts, thereby improving its ability to comprehend complex image structures. A channel cross-attention (CCA) module is also incorporated as a bridge between the encoder and decoder to better reconcile the semantic discrepancies between them.ResultsDetailed experiments on three public datasets were conducted. Specifically, our model was trained on 30 images from the Multi-organ Nucleus Segmentation (MoNuSeg) training dataset, 85 images from the Gland Segmentation (GlaS) training dataset, and 551 images from the Computer Vision Center Colorectal Cancer-Clinic Database (CVC-ClinicDB) dataset. We evaluated the performance of our model on the corresponding test sets. Our TAC-UNet achieved the best Dice scores (80.36%, 90.70%, and 91.81% on the MoNuSeg, GlaS, and CVC-ClinicDB datasets, respectively) of all the models. Compared to other CNN-based, Transformer-based, and hybrid methods, the TAC-UNet demonstrated significantly superior segmentation performance.ConclusionsOur TAC-UNet model showed advanced segmentation performance on small-scale medical image datasets. The detailed experimental results showed the effectiveness of the method. Our model's code is available at: https://github.com/hejlhello/TAC-UNet.

Dataset Information

New hybrid segmentation algorithm: UNet-GOA.

Publications

New hybrid segmentation algorithm: UNet-GOA.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets