Search

Fetal abdominal subcutaneous soft tissue thickness (FASSTT) is a key indicator in evaluating fetal growth, development, and nutritional status. Currently, manual measurement in FASSTT faces the problems of difficulty in positioning, time consumption, and inaccurate measurement. Therefore, this article proposes an automatic measurement scheme for FASSTT. Firstly, a U‐shaped attention network VGG‐SeUnet is proposed to automatically segment the subcutaneous soft tissue area of the fetal abdomen. Secondly, based on the segmentation results, a morphological method is proposed to obtain FASSTT. Specifically, the segmentation network uses VGG16 as the encoder and connects the decoder through jump connections for multi‐scale fusion. We introduce channel attention to jump connections, which enables the model to select key channels for feature fusion. At the same time, the model uses the proposed DF_Loss function for training to solve the problem of sample imbalance. Based on the segmentation results, a morphological distance transformation algorithm is proposed to obtain FASSTT by drawing the maximum inscribed circle and calculating its diameter. The method is evaluated on a fetal abdominal circumference ultrasound dataset containing 135 samples. The DSC, mIOU, mPA, Precision and Recall of the segmentation experiments reache 88.9%, 81.02%, 90.44%, 86.38%, and 90.44% respectively. The difference between FASSTT's result from that provided by radiologist is 0.0615 cm in Average Error (AVE) and 0.081 cm in Root Mean Square Error (RMSE). The overall performance of this algorithm is superior to existing methods with excellent performance, and the measurement error is less than 1 millimeter. This result proves that the proposed scheme in this paper can achieve automated measurement of FASSTT and assist doctors in subsequent evaluation of fetal development and nutritional status. [ABSTRACT FROM AUTHOR]

Channel and spatial attention mechanisms have proven to provide an evident performance boost of deep convolution neural networks. Most existing methods focus on one or run them parallel (series), neglecting the collaboration between the two attentions. In order to better establish the feature interaction between the two types of attentions, a plug‐and‐play attention module is proposed, which is termed as 'CAT'—activating the Collaboration between spatial and channel Attentions based on learned Traits. Specifically, traits are represented as trainable coefficients (i.e. colla‐factors) to adaptively combine contributions of different attention modules to fit different image hierarchies and tasks better. Moreover, the global entropy pooling is proposed apart from global average pooling and global maximum pooling (GMP) operators, which is an effective component in suppressing noise signals by measuring the information disorder of feature maps. A three‐way pooling operation is introduced into attention modules and the adaptive mechanism is applied to fuse their outcomes. Extensive experiments on MS COCO, Pascal‐VOC, Cifar‐100, and ImageNet show that our CAT outperforms the existing state‐of‐the‐art attention mechanisms in object detection, instance segmentation, and image classification. The model and code will be released soon. [ABSTRACT FROM AUTHOR]