A review of recent evaluation methods for image segmentation

Zhang, Yu Jin

doi:10.1109/isspa.2001.949797

Cited by 65 publications

(7 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We evaluate the segmentation methods by comparing the segmented MR images (S) to the corresponding gold standard manual segmentation (G) using spatial overlap-and distance-based metrics [69,70,71]. Those metrics are calculated using a slice-wise comparison and then averaged per patient;…”

Section: Evaluation Metricsmentioning

confidence: 99%

USE-Net: Incorporating Squeeze-and-Excitation blocks into U-Net for prostate zonal segmentation of multi-institutional MRI datasets

et al. 2019

View full text Add to dashboard Cite

Prostate cancer is the most common malignant tumors in men but prostate Magnetic Resonance Imaging (MRI) analysis remains challenging. Besides whole prostate gland segmentation, the capability to differentiate between the blurry boundary of the Central Gland (CG) and Peripheral Zone (PZ) can lead to differential diagnosis, since the frequency and severity of tumors differ in these regions. To tackle the prostate zonal segmentation task, we propose a novel Convolutional Neural Network (CNN), called USE-Net, which incorporates Squeeze-and-Excitation (SE) blocks into U-Net, i.e., one of the most effective CNNs in biomedical image segmentation. Especially, the SE blocks are added after every Encoder (Enc USE-Net) or Encoder-Decoder block (Enc-Dec USE-Net). This study evaluates the generalization ability of CNN-based architectures on three T2-weighted MRI datasets, each one consisting of a different number of patients and heterogeneous image characteristics, collected by different institutions. The following mixed scheme is used for training/testing: (i ) training on either each individual dataset or multiple prostate MRI datasets and (ii ) testing on all three datasets with all possible training/testing combinations. USE-Net is compared against three stateof-the-art CNN-based architectures (i.e., U-Net, pix2pix, and Mixed-Scale Dense Network), along with a semi-automatic continuous max-flow model. The results show that training on the union of the datasets generally outperforms training on each dataset separately, allowing for both intra-/crossdataset generalization. Enc USE-Net shows good overall generalization under any training condition, while Enc-Dec USE-Net remarkably outperforms the other methods when trained on all datasets. These findings reveal that the SE blocks' adaptive feature recalibration provides excellent cross-dataset generalization when testing is performed on samples of the datasets used during training. Therefore, we should consider multi-dataset training and SE

show abstract

Section: Evaluation Metricsmentioning

confidence: 99%

USE-Net: Incorporating Squeeze-and-Excitation blocks into U-Net for prostate zonal segmentation of multi-institutional MRI datasets

et al. 2019

View full text Add to dashboard Cite

show abstract

“…In order to objectively reflect the performance of the method, we evaluated the proposed model using various evaluation metrics. The evaluation metrics for segmentation results are Dice Similarity Coefficient (DSC), Intersection over Union (IoU) and Pixel Accuracy (PA) [ 40 ]. From the perspective of calculating the region similarity, DSC and IoU were used to evaluate the distance difference between the segmentation result and the ground truth.…”

Section: Methodsmentioning

confidence: 99%

Automatic Analysis of Transverse Musculoskeletal Ultrasound Images Based on the Multi-Task Learning Model

Zhou

Liu

Zheng

2023

Entropy

View full text Add to dashboard Cite

Musculoskeletal ultrasound imaging is an important basis for the early screening and accurate treatment of muscle disorders. It allows the observation of muscle status to screen for underlying neuromuscular diseases including myasthenia gravis, myotonic dystrophy, and ankylosing muscular dystrophy. Due to the complexity of skeletal muscle ultrasound image noise, it is a tedious and time-consuming process to analyze. Therefore, we proposed a multi-task learning-based approach to automatically segment and initially diagnose transverse musculoskeletal ultrasound images. The method implements muscle cross-sectional area (CSA) segmentation and abnormal muscle classification by constructing a multi-task model based on multi-scale fusion and attention mechanisms (MMA-Net). The model exploits the correlation between tasks by sharing a part of the shallow network and adding connections to exchange information in the deep network. The multi-scale feature fusion module and attention mechanism were added to MMA-Net to increase the receptive field and enhance the feature extraction ability. Experiments were conducted using a total of 1827 medial gastrocnemius ultrasound images from multiple subjects. Ten percent of the samples were randomly selected for testing, 10% as the validation set, and the remaining 80% as the training set. The results show that the proposed network structure and the added modules are effective. Compared with advanced single-task models and existing analysis methods, our method has a better performance at classification and segmentation. The mean Dice coefficients and IoU of muscle cross-sectional area segmentation were 96.74% and 94.10%, respectively. The accuracy and recall of abnormal muscle classification were 95.60% and 94.96%. The proposed method achieves convenient and accurate analysis of transverse musculoskeletal ultrasound images, which can assist physicians in the diagnosis and treatment of muscle diseases from multiple perspectives.

show abstract

“…First, the accuracy indicates the percentage of correctly predicted pixels for each class, while global accuracy indicates the percentage of correctly predicted pixels regardless of class. In addition, mean accuracy is the average accuracy obtained for all classes of all images used 37 . On the other hand, the intersection over union (IoU) value (TP/[TP + FP + FN]), also known as the Jaccard similarity coefficient is obtained by dividing the correctly classified pixels by the total number of predicted pixels and ground truth.…”

Section: Methodsmentioning

confidence: 99%

“…The confusion matrices for MobileNetV2-based DeepLab v3+ without augmentation and ResNet-18-based DeepLab v3+ with augmentation models accuracy obtained for all classes of all images used. 37 On the other hand, the intersection over union (IoU) value (TP/[TP + FP + FN]), also known as the Jaccard similarity coefficient is obtained by dividing the correctly classified pixels by the total number of predicted pixels and ground truth. Also, while the mean IoU is the average IoU score of all classes in the images, the average IoU of each class is weighted by the number of pixels in that class for the weighted IoU.…”

Section: Performance Metricsmentioning

confidence: 99%

Comparative parotid gland segmentation by using ResNet‐18 and MobileNetV2 based DeepLab v3+ architectures from magnetic resonance images

Sünnetci

Kaba

Çeliker

et al. 2022

Concurrency and Computation

View full text Add to dashboard Cite

Summary Nowadays, artificial intelligence‐based medicine plays an important role in determining correlations not comprehensible to humans. In addition, the segmentation of organs at risk is a tedious and time‐consuming procedure. Segmentation of these organs or tissues is widely used in early diagnosis, treatment planning, and diagnosis. In this study, we trained semantic segmentation networks to segment healthy parotid glands using deep learning. The dataset we used in the study was obtained from Recep Tayyip Erdogan University Training and Research Hospital, and there were 72 T2‐weighted magnetic resonance (MR) images in this dataset. After these images were manually segmented by experts, masks of these images were obtained according to them and all images were cropped. Afterward, these cropped images and masks were rotated 45°, 120°, and 210°, quadrupling the number of images. We trained ResNet‐18/MobileNetV2‐based DeepLab v3+ without augmentation and ResNet‐18/MobileNetV2‐based DeepLab v3+ with augmentation using these datasets. Here, we set the training set and testing set sizes for all architectures to be 80% and 20%, respectively. We designed two different graphical user interface (GUI) applications so that users can easily segment their parotid glands by utilizing all of these deep learning‐based semantic segmentation networks. From the results, mean‐weighted dice values of MobileNetV2‐based DeepLab v3+ without augmentation and ResNet‐18‐based DeepLab v3+ with augmentation were equal to 0.90845–0.93931 and 0.93237–0.96960, respectively. We also noted that the sensitivity (%), specificity (%), F1 score (%) values of these models were equal to 83.21, 96.65, 85.04 and 89.81, 97.84, 87.80, respectively. As a result, these designed models were found to be clinically successful, and the user‐friendly GUI applications of these proposed systems can be used by clinicians. This study is competitive as it uses MR images, can automatically segment both parotid glands, the results are meaningful according to the literature and have software application.

show abstract

A review of recent evaluation methods for image segmentation

Cited by 65 publications

References 11 publications

USE-Net: Incorporating Squeeze-and-Excitation blocks into U-Net for prostate zonal segmentation of multi-institutional MRI datasets

USE-Net: Incorporating Squeeze-and-Excitation blocks into U-Net for prostate zonal segmentation of multi-institutional MRI datasets

Automatic Analysis of Transverse Musculoskeletal Ultrasound Images Based on the Multi-Task Learning Model

Comparative parotid gland segmentation by using ResNet‐18 and MobileNetV2 based DeepLab v3+ architectures from magnetic resonance images

Contact Info

Product

Resources

About