Pub Date : 2025-02-01DOI: 10.1016/j.medengphy.2025.104298
Jui-Chung Ni , Shih-Hsiung Lee , Yen-Cheng Shen , Chu-Sing Yang
In previous tasks of glottis image segmentation, the position attention mechanism was rarely incorporated, neglecting the detailed information in glottis position detection. Aiming to improve the U-Net architecture, this study introduces the dual attention mechanism based on the squeeze and excitation (SE)-Net model. This mechanism can integrate traditional channel attention with position attention mechanisms to effectively adjust the weights of crucial features and significance of positions. Replacing the weight adjustment mechanism in SE-Net with the dual attention mechanism creates a broader perspective, enhancing the sensitivity to important features in the model. Furthermore, based on the characteristics of SE-Net, the skip-connection feature of U-Net can still be retained. The architecture proposed in this paper further replaces the convolutional layers in the U-Net encoder with the bottleneck to preserve the information on the features without significantly increasing the amount of computation. In addition, the decoder is replaced with residual blocks to reduce overfitting. The results of the experiment showed that models with retained features demonstrate better accuracy while reducing overfitting. The proposed model achieved positive results in predicting the scores on the public benchmark for automatic glottis segmentation (BAGLS) dataset.
{"title":"Improved U-Net based on ResNet and SE-Net with dual attention mechanism for glottis semantic segmentation","authors":"Jui-Chung Ni , Shih-Hsiung Lee , Yen-Cheng Shen , Chu-Sing Yang","doi":"10.1016/j.medengphy.2025.104298","DOIUrl":"10.1016/j.medengphy.2025.104298","url":null,"abstract":"<div><div>In previous tasks of glottis image segmentation, the position attention mechanism was rarely incorporated, neglecting the detailed information in glottis position detection. Aiming to improve the U-Net architecture, this study introduces the dual attention mechanism based on the squeeze and excitation (SE)-Net model. This mechanism can integrate traditional channel attention with position attention mechanisms to effectively adjust the weights of crucial features and significance of positions. Replacing the weight adjustment mechanism in SE-Net with the dual attention mechanism creates a broader perspective, enhancing the sensitivity to important features in the model. Furthermore, based on the characteristics of SE-Net, the skip-connection feature of U-Net can still be retained. The architecture proposed in this paper further replaces the convolutional layers in the U-Net encoder with the bottleneck to preserve the information on the features without significantly increasing the amount of computation. In addition, the decoder is replaced with residual blocks to reduce overfitting. The results of the experiment showed that models with retained features demonstrate better accuracy while reducing overfitting. The proposed model achieved positive results in predicting the scores on the public benchmark for automatic glottis segmentation (BAGLS) dataset.</div></div>","PeriodicalId":49836,"journal":{"name":"Medical Engineering & Physics","volume":"136 ","pages":"Article 104298"},"PeriodicalIF":1.7,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143350125","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01DOI: 10.1016/j.medengphy.2024.104265
Leila Ranjbar , Hossein Parsaei , Mohammad Mehdi Movahedi , Sam Sharifzadeh Javidi
This study explores the effectiveness of spectral clustering for spike sorting and proposes a Separability Index to measure the difficulty of spike sorting for a signal. The accuracy of spectral clustering is evaluated using different feature sets, including raw samples, first and second derivatives, and principal components analysis (PCA), and compared to two previously published methods. The results obtained over a dataset including 16 signals show that raw samples, with an average accuracy of 73.84 %, are effective for spectral clustering-based spike sorting. The analysis demonstrates that the proposed Separability Index can be utilized to classify signals beforehand, reducing the cost and processing time of large datasets. Furthermore, the proposed index can reveal spike sorting difficulty, making it a valuable tool for comparing the performance of various spike sorting methods in depth. The proposed method has higher accuracy (up to 23 %) compared to two previously published methods, and its accuracy is aligned with the Separability Index (correlation coefficient = 0.71). Overall, this study contributes to the field of spike sorting and offers insights into leveraging spectral clustering for this task.
{"title":"Improving spike sorting efficiency with separability index and spectral clustering","authors":"Leila Ranjbar , Hossein Parsaei , Mohammad Mehdi Movahedi , Sam Sharifzadeh Javidi","doi":"10.1016/j.medengphy.2024.104265","DOIUrl":"10.1016/j.medengphy.2024.104265","url":null,"abstract":"<div><div>This study explores the effectiveness of spectral clustering for spike sorting and proposes a Separability Index to measure the difficulty of spike sorting for a signal. The accuracy of spectral clustering is evaluated using different feature sets, including raw samples, first and second derivatives, and principal components analysis (PCA), and compared to two previously published methods. The results obtained over a dataset including 16 signals show that raw samples, with an average accuracy of 73.84 %, are effective for spectral clustering-based spike sorting. The analysis demonstrates that the proposed Separability Index can be utilized to classify signals beforehand, reducing the cost and processing time of large datasets. Furthermore, the proposed index can reveal spike sorting difficulty, making it a valuable tool for comparing the performance of various spike sorting methods in depth. The proposed method has higher accuracy (up to 23 %) compared to two previously published methods, and its accuracy is aligned with the Separability Index (correlation coefficient = 0.71). Overall, this study contributes to the field of spike sorting and offers insights into leveraging spectral clustering for this task.</div></div>","PeriodicalId":49836,"journal":{"name":"Medical Engineering & Physics","volume":"135 ","pages":"Article 104265"},"PeriodicalIF":1.7,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143096506","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Individuals with neuropsychiatric disorders experience both physical and mental difficulties, hindering their ability to live healthy lives and participate in daily activities. It is challenging to diagnose these disorders due to a lack of reliable diagnostic tests and the complex symptoms and treatments for various disorders. Generally, psychiatric disorders are identified manually by doctors using questionnaires, which may be prone to subjectivity and human errors. A few automated systems have recently been developed to identify these disorders using physiological signals, including electroencephalogram (EEG) and electrocardiogram (ECG) signals. Often, EEG signals are used to identify psychiatric disorders, but the EEG signals are nonlinear and non-stationary in nature and hence are relatively complex to analyze when compared to the ECG signals. The ECG signals in psychiatric patients are used due to the connection between the heart and brain. The proposed study is aimed at investigating the use of ECG signals for the automated identification of neuropsychiatric disorders, including bipolar disorder (BD), depression (DP), and schizophrenia (SZ). Generally, convolution neural networks (CNNs) have proven to be effective in accurately identifying psychological conditions. However, their application requires a large amount of data and technical expertise. The wavelet scattering network (WSN), a variant of CNNs, was introduced to overcome these limitations. The WSN is a network capable of accurately detecting unique patterns in the signal. The proposed research incorporated the WSN network and was conducted using a Psychiatric ECG Beat Dataset with a population of 233 subjects, of whom 198 were diagnosed with multiple psychiatric disorders, and 35 were control subjects. ECG signals from 3570 heartbeats were collected and analyzed using wavelet scattering-based feature extraction and machine learning techniques. The Fine K-Nearest Neighbor (FKNN) algorithm produced the best results with an average classification accuracy of 99.8% and a Kappa value of 0.996 using a ten-fold cross-validation. The model yielded an accuracy of 99.78%, 99.94%, 99.98%, and 100% for automated identification of BD, DP, SZ, and control subjects, respectively, with F1 scores and precision values close to 1. The proposed method could also help in the automated clinical detection of different psychiatric disorders.
{"title":"A novel ECG-based approach for classifying psychiatric disorders: Leveraging wavelet scattering networks","authors":"Hardik Telangore , Nishant Sharma , Manish Sharma , U. Rajendra Acharya","doi":"10.1016/j.medengphy.2024.104275","DOIUrl":"10.1016/j.medengphy.2024.104275","url":null,"abstract":"<div><div>Individuals with neuropsychiatric disorders experience both physical and mental difficulties, hindering their ability to live healthy lives and participate in daily activities. It is challenging to diagnose these disorders due to a lack of reliable diagnostic tests and the complex symptoms and treatments for various disorders. Generally, psychiatric disorders are identified manually by doctors using questionnaires, which may be prone to subjectivity and human errors. A few automated systems have recently been developed to identify these disorders using physiological signals, including electroencephalogram (EEG) and electrocardiogram (ECG) signals. Often, EEG signals are used to identify psychiatric disorders, but the EEG signals are nonlinear and non-stationary in nature and hence are relatively complex to analyze when compared to the ECG signals. The ECG signals in psychiatric patients are used due to the connection between the heart and brain. The proposed study is aimed at investigating the use of ECG signals for the automated identification of neuropsychiatric disorders, including bipolar disorder (BD), depression (DP), and schizophrenia (SZ). Generally, convolution neural networks (CNNs) have proven to be effective in accurately identifying psychological conditions. However, their application requires a large amount of data and technical expertise. The wavelet scattering network (WSN), a variant of CNNs, was introduced to overcome these limitations. The WSN is a network capable of accurately detecting unique patterns in the signal. The proposed research incorporated the WSN network and was conducted using a Psychiatric ECG Beat Dataset with a population of 233 subjects, of whom 198 were diagnosed with multiple psychiatric disorders, and 35 were control subjects. ECG signals from 3570 heartbeats were collected and analyzed using wavelet scattering-based feature extraction and machine learning techniques. The Fine K-Nearest Neighbor (FKNN) algorithm produced the best results with an average classification accuracy of 99.8% and a Kappa value of 0.996 using a ten-fold cross-validation. The model yielded an accuracy of 99.78%, 99.94%, 99.98%, and 100% for automated identification of BD, DP, SZ, and control subjects, respectively, with F1 scores and precision values close to 1. The proposed method could also help in the automated clinical detection of different psychiatric disorders.</div></div>","PeriodicalId":49836,"journal":{"name":"Medical Engineering & Physics","volume":"135 ","pages":"Article 104275"},"PeriodicalIF":1.7,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143095987","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01DOI: 10.1016/j.medengphy.2024.104271
Arthur Jourdan , Anthony Vegleur , Jeff Bodner , Pascal Rousset , Guillaume Passot , Anicet Le Ruyet
A ventral hernia involves tissue protrusion through the abdominal wall (AW). It is a common surgical issue with high recurrence rates. Primary stability of hernia meshes is essential to guarantee mesh integration, yet existing meshes often fail to match the AW's complex biomechanics. This study proposes a novel method aiming at understanding post-operative mesh-AW interactions. Three fresh frozen human specimens underwent an open Rives-Stoppa implantation of a synthetic hernia mesh coated with metallic micro-beads. Additional beads were placed into the AW muscle tissue. CT scans were conducted at increasing levels of intra-abdominal pressure to reproduce forced breathing. Beads 3D coordinates were exported from the CT-scans and motion and strain of both the hernia mesh and the AW were calculated. At 30 mmHg, the mesh-muscle motion (or sliding) was 2.3 ± 1.3 mm. Muscle exhibited significantly higher strains (12.9 ± 4.7 %) than the hernia mesh (4.7 ± 1.1 %), most likely due to difference in material properties between the mesh and the AW. A repeatability study was carried out to build confidence in the proposed method. This protocol can bring insights of the hernia mesh use-conditions to improve hernia mesh design requirements and develop safer implants to reduce hernia recurrence.
{"title":"A combined experimental and numerical approach to evaluate hernia mesh biomechanical stability in situ","authors":"Arthur Jourdan , Anthony Vegleur , Jeff Bodner , Pascal Rousset , Guillaume Passot , Anicet Le Ruyet","doi":"10.1016/j.medengphy.2024.104271","DOIUrl":"10.1016/j.medengphy.2024.104271","url":null,"abstract":"<div><div>A ventral hernia involves tissue protrusion through the abdominal wall (AW). It is a common surgical issue with high recurrence rates. Primary stability of hernia meshes is essential to guarantee mesh integration, yet existing meshes often fail to match the AW's complex biomechanics. This study proposes a novel method aiming at understanding post-operative mesh-AW interactions. Three fresh frozen human specimens underwent an open Rives-Stoppa implantation of a synthetic hernia mesh coated with metallic micro-beads. Additional beads were placed into the AW muscle tissue. CT scans were conducted at increasing levels of intra-abdominal pressure to reproduce forced breathing. Beads 3D coordinates were exported from the CT-scans and motion and strain of both the hernia mesh and the AW were calculated. At 30 mmHg, the mesh-muscle motion (or sliding) was 2.3 ± 1.3 mm. Muscle exhibited significantly higher strains (12.9 ± 4.7 %) than the hernia mesh (4.7 ± 1.1 %), most likely due to difference in material properties between the mesh and the AW. A repeatability study was carried out to build confidence in the proposed method. This protocol can bring insights of the hernia mesh use-conditions to improve hernia mesh design requirements and develop safer implants to reduce hernia recurrence.</div></div>","PeriodicalId":49836,"journal":{"name":"Medical Engineering & Physics","volume":"135 ","pages":"Article 104271"},"PeriodicalIF":1.7,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143095988","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01DOI: 10.1016/j.medengphy.2024.104276
Farnaz Garehdaghi, Yashar Sarbaz
Parkinson's disease (PD) is a neurodegenerative disease. Since the diagnosis of the PD is mainly made based on the symptoms and after the disease progression, early diagnosis can play a crucial role in delaying the passage of the PD. There have been many methods focusing on disease diagnosis using electroencephalography (EEG) signals, where most of the proposed methods are data-dependent. Here, the study aims to propose a technique that, despite its high accuracy, is robust. Various features including fractal dimension, approximate entropy, largest Lyapunov exponent, and the energy of different frequency sub-bands were extracted from EEG signals. Multi-layer perceptron neural networks were used for classification based on these features. Additionally, 2D phase space images reconstructed from EEG signals were classified using convolutional neural networks. Finally, a combination of these features and images was used for classification using ResNets. During 10 rounds of training and testing, the mean accuracies were calculated for three cases: using only features, only images, and a combination of both. The mean accuracies were 84.67 %, 76.5 %, and 90.2 % respectively. The variances for each case were 35.6 %, 19.5 %, and 13.97 %. The lower variance when using a combination of features and images indicates a more accurate and robust classification.
{"title":"A robust method for parkinson's disease diagnosis: Combining electroencephalography signal features with reconstructed phase space images","authors":"Farnaz Garehdaghi, Yashar Sarbaz","doi":"10.1016/j.medengphy.2024.104276","DOIUrl":"10.1016/j.medengphy.2024.104276","url":null,"abstract":"<div><div>Parkinson's disease (PD) is a neurodegenerative disease. Since the diagnosis of the PD is mainly made based on the symptoms and after the disease progression, early diagnosis can play a crucial role in delaying the passage of the PD. There have been many methods focusing on disease diagnosis using electroencephalography (EEG) signals, where most of the proposed methods are data-dependent. Here, the study aims to propose a technique that, despite its high accuracy, is robust. Various features including fractal dimension, approximate entropy, largest Lyapunov exponent, and the energy of different frequency sub-bands were extracted from EEG signals. Multi-layer perceptron neural networks were used for classification based on these features. Additionally, 2D phase space images reconstructed from EEG signals were classified using convolutional neural networks. Finally, a combination of these features and images was used for classification using ResNets. During 10 rounds of training and testing, the mean accuracies were calculated for three cases: using only features, only images, and a combination of both. The mean accuracies were 84.67 %, 76.5 %, and 90.2 % respectively. The variances for each case were 35.6 %, 19.5 %, and 13.97 %. The lower variance when using a combination of features and images indicates a more accurate and robust classification.</div></div>","PeriodicalId":49836,"journal":{"name":"Medical Engineering & Physics","volume":"135 ","pages":"Article 104276"},"PeriodicalIF":1.7,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143135687","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The stress shielding phenomenon causes undesirable changes in stress distribution in the bones in which implants are implanted. This phenomenon leads to a gradual decrease in bone density in the vicinity of the implant and ultimately loosening. To improve the endurance of the prosthesis and postpone the revision surgery the prosthesis should be designed and positioned in such a way that the post-implant stress distribution within the bones is as close to the natural distribution as possible.
We formulate the problem of achieving a near-normal stress distribution as a constrained optimization problem in which the difference between pre- and post-implant stress distributions constitutes the loss function and five positional and dimensional parameters, including Femoral Anteversion, Neck Shaft Angle, Femoral Head Offset, Cup Version, and Cup Inclination comprise the set of design variables. Finite Elements (FE) analysis is used to obtain the stress distribution in bones before and after the implantation. To create a FE model, first a three-dimensional model of the patient's hip is created using CT images. The model is then subjected to the loads obtained from the patient's gait analysis, and the Stress Shielding Index (SSI) is calculated at critical points of the bones before and after the implantation. The difference between pre- and post-implant SSI values defines the cost function. To reduce the computational cost of numerous cost function evaluations a surrogate model (a 5 × 1 MLP neural network) is employed to predict the value of the cost function. Design Of Experiments (DOE) is used to sample the hyperspace of the design variables and generate training, test and validation data. The optimization problem is solved using Genetic Algorithms and the results are compared with the results of the best set of initial samples.
Results from the proposed approach show a significant reduction in the difference between pre- and post-implant stress distributions and a 12 % reduction in the Stress Shielding Index.
{"title":"Surrogate-based positioning optimization of hip prostheses for minimal stress shielding","authors":"Mahmoud Mohammadizand, Massoud Shariat-Panahi, Morad Karimpour","doi":"10.1016/j.medengphy.2024.104273","DOIUrl":"10.1016/j.medengphy.2024.104273","url":null,"abstract":"<div><div>The stress shielding phenomenon causes undesirable changes in stress distribution in the bones in which implants are implanted. This phenomenon leads to a gradual decrease in bone density in the vicinity of the implant and ultimately loosening. To improve the endurance of the prosthesis and postpone the revision surgery the prosthesis should be designed and positioned in such a way that the post-implant stress distribution within the bones is as close to the natural distribution as possible.</div><div>We formulate the problem of achieving a near-normal stress distribution as a constrained optimization problem in which the difference between pre- and post-implant stress distributions constitutes the loss function and five positional and dimensional parameters, including Femoral Anteversion, Neck Shaft Angle, Femoral Head Offset, Cup Version, and Cup Inclination comprise the set of design variables. Finite Elements (FE) analysis is used to obtain the stress distribution in bones before and after the implantation. To create a FE model, first a three-dimensional model of the patient's hip is created using CT images. The model is then subjected to the loads obtained from the patient's gait analysis, and the Stress Shielding Index (SSI) is calculated at critical points of the bones before and after the implantation. The difference between pre- and post-implant SSI values defines the cost function. To reduce the computational cost of numerous cost function evaluations a surrogate model (a 5 × 1 MLP neural network) is employed to predict the value of the cost function. Design Of Experiments (DOE) is used to sample the hyperspace of the design variables and generate training, test and validation data. The optimization problem is solved using Genetic Algorithms and the results are compared with the results of the best set of initial samples.</div><div>Results from the proposed approach show a significant reduction in the difference between pre- and post-implant stress distributions and a 12 % reduction in the Stress Shielding Index.</div></div>","PeriodicalId":49836,"journal":{"name":"Medical Engineering & Physics","volume":"135 ","pages":"Article 104273"},"PeriodicalIF":1.7,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143095990","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01DOI: 10.1016/j.medengphy.2024.104277
Aswathy Elma Aby , S. Salaji , K.K. Anilkumar , Tintu Rajan
Acute Myeloid Leukemia(AML) is a rapidly progressing cancer affecting blood and bone marrow, marked by the swift proliferation of abnormal myeloid cells. Effective treatment requires precise classification of AML subtypes. Conventional classification methods rely on manual microscopic analysis, which is time-consuming and variable, while traditional machine learning approaches often struggle with feature extraction and generalization. This underscores the need for automated detection methods which enhance this process by minimizing manual effort. Activation functions are critical in neural networks, introducing non-linearity that influences training convergence, computational efficiency, and model performance. This study evaluates the effectiveness of CNN architectures, Sequential (VGG16), Directed Acyclic Graph (InceptionV3), and Residual (ResNet50v2) in distinguishing between AML subtypes: AML without maturation, Acute Monoblastic Leukemia, and Pure Erythroid Leukemia, using peripheral blood smear images, while also investigating the impact of different activation functions on model accuracy and training time. The results show that ResNet50v2 achieves the shortest training time, while InceptionV3 takes the longest due to its complex architecture. GELU delivers the highest accuracy, reaching 94.02 % in InceptionV3 and 92.53 % in ResNet50v2, while SELU achieves the highest accuracy for VGG16 at 92.83 %. Mish provides competitive accuracy with lower training time than GELU, while Softplus and Softsign consistently perform poorly. This research demonstrates the potential of CNNs for automating AML subtype classification and identifies GELU as the most effective activation function. Future work could explore data augmentation, optimized activation functions, and attention mechanisms to improve classification performance.
{"title":"Classification of acute myeloid leukemia by pre-trained deep neural networks: A comparison with different activation functions","authors":"Aswathy Elma Aby , S. Salaji , K.K. Anilkumar , Tintu Rajan","doi":"10.1016/j.medengphy.2024.104277","DOIUrl":"10.1016/j.medengphy.2024.104277","url":null,"abstract":"<div><div>Acute Myeloid Leukemia(AML) is a rapidly progressing cancer affecting blood and bone marrow, marked by the swift proliferation of abnormal myeloid cells. Effective treatment requires precise classification of AML subtypes. Conventional classification methods rely on manual microscopic analysis, which is time-consuming and variable, while traditional machine learning approaches often struggle with feature extraction and generalization. This underscores the need for automated detection methods which enhance this process by minimizing manual effort. Activation functions are critical in neural networks, introducing non-linearity that influences training convergence, computational efficiency, and model performance. This study evaluates the effectiveness of CNN architectures, Sequential (VGG16), Directed Acyclic Graph (InceptionV3), and Residual (ResNet50v2) in distinguishing between AML subtypes: AML without maturation, Acute Monoblastic Leukemia, and Pure Erythroid Leukemia, using peripheral blood smear images, while also investigating the impact of different activation functions on model accuracy and training time. The results show that ResNet50v2 achieves the shortest training time, while InceptionV3 takes the longest due to its complex architecture. GELU delivers the highest accuracy, reaching 94.02 % in InceptionV3 and 92.53 % in ResNet50v2, while SELU achieves the highest accuracy for VGG16 at 92.83 %. Mish provides competitive accuracy with lower training time than GELU, while Softplus and Softsign consistently perform poorly. This research demonstrates the potential of CNNs for automating AML subtype classification and identifies GELU as the most effective activation function. Future work could explore data augmentation, optimized activation functions, and attention mechanisms to improve classification performance.</div></div>","PeriodicalId":49836,"journal":{"name":"Medical Engineering & Physics","volume":"135 ","pages":"Article 104277"},"PeriodicalIF":1.7,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143135688","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01DOI: 10.1016/j.medengphy.2024.104278
Zhengxin Tu , Jinghua Xu , Zhenyu Dong , Shuyou Zhang , Jianrong Tan
A biomechanical evaluation method for bone arthrosis morphology based on reconstructed dynamic kinesiology (RDK) is proposed. The hip joint is a ball-and-socket joint, morphologically characterized by an acetabulum with a nearly spherical concavity and uniform curvatures, where Gaussian curvature exhibits negative characteristic. Subsequently, RDK of bone joint morphology is developed, offering detailed anatomical and kinematic insights. The hip joint is taken as a verification instance, where a precise biomechanical evaluation of bone arthrosis morphology is simulated through finite element analysis (FEA). Latin Hypercube sampling (LHS) with the criterion of maximizing the minimum distance enhances uniformity and representation. The response surface is subsequently constructed by Kriging interpolation, significantly enhancing computational efficiency and FEA accuracy. Innovatively, a stress contour statistical histogram of load transfer is presented to quantitatively analyze the stress lines, supplying support for biomechanical evaluation, which is essential for accurate hip replacement planning. The instance indicates that the proposed RDK facilitates accurate biomechanical evaluations for bone arthrosis morphology, providing a critical theoretical foundation for conceptual design of ergonomic wearable devices, as well as optimization of replacement surgeries.
{"title":"Biomechanical evaluation for bone arthrosis morphology based on reconstructed dynamic kinesiology","authors":"Zhengxin Tu , Jinghua Xu , Zhenyu Dong , Shuyou Zhang , Jianrong Tan","doi":"10.1016/j.medengphy.2024.104278","DOIUrl":"10.1016/j.medengphy.2024.104278","url":null,"abstract":"<div><div>A biomechanical evaluation method for bone arthrosis morphology based on reconstructed dynamic kinesiology (RDK) is proposed. The hip joint is a ball-and-socket joint, morphologically characterized by an acetabulum with a nearly spherical concavity and uniform curvatures, where Gaussian curvature exhibits negative characteristic. Subsequently, RDK of bone joint morphology is developed, offering detailed anatomical and kinematic insights. The hip joint is taken as a verification instance, where a precise biomechanical evaluation of bone arthrosis morphology is simulated through finite element analysis (FEA). Latin Hypercube sampling (LHS) with the criterion of maximizing the minimum distance enhances uniformity and representation. The response surface is subsequently constructed by Kriging interpolation, significantly enhancing computational efficiency and FEA accuracy. Innovatively, a stress contour statistical histogram of load transfer is presented to quantitatively analyze the stress lines, supplying support for biomechanical evaluation, which is essential for accurate hip replacement planning. The instance indicates that the proposed RDK facilitates accurate biomechanical evaluations for bone arthrosis morphology, providing a critical theoretical foundation for conceptual design of ergonomic wearable devices, as well as optimization of replacement surgeries.</div></div>","PeriodicalId":49836,"journal":{"name":"Medical Engineering & Physics","volume":"135 ","pages":"Article 104278"},"PeriodicalIF":1.7,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143135689","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01DOI: 10.1016/j.medengphy.2024.104274
Maila Petrucci , Antonino A. La Mattina , Cristina Curreli , Enrico Tassinari , Marco Viceconti
Intraoperative femur fractures are a complication of hip arthroplasty, strongly related to the cementless stem design; this kind of fracture is not always recognised during surgery, and revision surgery may be necessary. The present study aimed to simulate intraoperative crack propagation during stem implantation using subject-specific quasi-static finite element models. Eleven subject-specific finite element femur models were built starting from CT data, and the implant pose and size of a non-commercial cementless stem were identified. The model boundary conditions were set with a compressive load from 1000 N to 10 000 N, to simulate the surgeon's hammering, and element deactivation was used to model the crack propagation. Two damage quantifiers were analysed to identify a threshold value that would allow us to assess if a fracture occurred. A methodology to assess the primary stability of the stem during insertion was also proposed, based on a push-out test. Crack propagation up to the surface was obtained in six patients; in two cases there was no crack generation, while in three patients the crack did not reach the external surface. This study demonstrates the possibility to simulate the propagation of the fracture intraoperatively during hip replacement surgery and generate quantitative information about the bone damage using a virtual cohort of simulated patients with anatomical and physiological variability.
{"title":"A finite element model to simulate intraoperative fractures in cementless hip stem designs","authors":"Maila Petrucci , Antonino A. La Mattina , Cristina Curreli , Enrico Tassinari , Marco Viceconti","doi":"10.1016/j.medengphy.2024.104274","DOIUrl":"10.1016/j.medengphy.2024.104274","url":null,"abstract":"<div><div>Intraoperative femur fractures are a complication of hip arthroplasty, strongly related to the cementless stem design; this kind of fracture is not always recognised during surgery, and revision surgery may be necessary. The present study aimed to simulate intraoperative crack propagation during stem implantation using subject-specific quasi-static finite element models. Eleven subject-specific finite element femur models were built starting from CT data, and the implant pose and size of a non-commercial cementless stem were identified. The model boundary conditions were set with a compressive load from 1000 N to 10 000 N, to simulate the surgeon's hammering, and element deactivation was used to model the crack propagation. Two damage quantifiers were analysed to identify a threshold value that would allow us to assess if a fracture occurred. A methodology to assess the primary stability of the stem during insertion was also proposed, based on a push-out test. Crack propagation up to the surface was obtained in six patients; in two cases there was no crack generation, while in three patients the crack did not reach the external surface. This study demonstrates the possibility to simulate the propagation of the fracture intraoperatively during hip replacement surgery and generate quantitative information about the bone damage using a virtual cohort of simulated patients with anatomical and physiological variability.</div></div>","PeriodicalId":49836,"journal":{"name":"Medical Engineering & Physics","volume":"135 ","pages":"Article 104274"},"PeriodicalIF":1.7,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143135692","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Photoacoustic imaging (PAI) is a non-invasive diagnostic imaging technique that utilizes the photoacoustic effect by combining optical and ultrasound imaging systems. The development of PAI is mostly centered on the generation of a high-quality 3D reconstruction system for more optimal and accurate identification of tissue abnormalities. This literature study was conducted to analyze the 3D image reconstruction in PAI over 2017–2024. In this review, the collected articles in 3D photoacoustic imaging were categorized based on the approach, design, and purpose of each study. Firstly, the approaches of the studies were classified into three groups: experimental studies, numerical simulation, and numerical simulation with experimental validation. Secondly, the design of the study was assessed based on the photoacoustic modality, laser type, and sensing mechanism. Thirdly, the purpose of the collected studies was summarized into seven subsections, including image quality improvement, frame rate improvement, image segmentation, system integration, inter-systems comparisons, improving computational efficiency, and portable system development. The results of this review revealed that the 3D PAI systems have been developed by various research groups, suggesting the investigation of numerous biological objects. Therefore, 3D PAI has the potential to contribute a wide range of novel biological imaging systems that support real-time biomedical imaging in the future.
{"title":"Trends and developments in 3D photoacoustic imaging systems: A review of recent progress","authors":"Fikhri Astina Tasmara , Mitrayana Mitrayana , Andreas Setiawan , Takuro Ishii , Yoshifumi Saijo , Rini Widyaningrum","doi":"10.1016/j.medengphy.2024.104268","DOIUrl":"10.1016/j.medengphy.2024.104268","url":null,"abstract":"<div><div>Photoacoustic imaging (PAI) is a non-invasive diagnostic imaging technique that utilizes the photoacoustic effect by combining optical and ultrasound imaging systems. The development of PAI is mostly centered on the generation of a high-quality 3D reconstruction system for more optimal and accurate identification of tissue abnormalities. This literature study was conducted to analyze the 3D image reconstruction in PAI over 2017–2024. In this review, the collected articles in 3D photoacoustic imaging were categorized based on the approach, design, and purpose of each study. Firstly, the approaches of the studies were classified into three groups: experimental studies, numerical simulation, and numerical simulation with experimental validation. Secondly, the design of the study was assessed based on the photoacoustic modality, laser type, and sensing mechanism. Thirdly, the purpose of the collected studies was summarized into seven subsections, including image quality improvement, frame rate improvement, image segmentation, system integration, inter-systems comparisons, improving computational efficiency, and portable system development. The results of this review revealed that the 3D PAI systems have been developed by various research groups, suggesting the investigation of numerous biological objects. Therefore, 3D PAI has the potential to contribute a wide range of novel biological imaging systems that support real-time biomedical imaging in the future.</div></div>","PeriodicalId":49836,"journal":{"name":"Medical Engineering & Physics","volume":"135 ","pages":"Article 104268"},"PeriodicalIF":1.7,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143096507","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号