Medical & Biological Engineering & Computing最新文献

Hepatocellular carcinoma is among the leading causes of cancer-related mortality, and accurate survival prediction is crucial for personalized treatment. However, conventional approaches such as the Cox proportional hazards model often struggle with nonlinear relationships and high-dimensional data, resulting in suboptimal predictive performance. In this study, we utilized HCC patient data from the SEER and TCGA databases to investigate the potential of machine learning methods in HCC survival prediction. Specifically, we introduced a self-attention mechanism into DeepSurv and DeepHit to better capture feature dependencies and incorporated residual network modules to enhance the training stability of the deep architectures. Furthermore, we developed an ensemble model based on a Cox neural network, combining the predictions from our improved deep learning models, the Cox proportional hazards model, and random survival forest. Both the model improvements and the ensemble approach described here are being applied for the first time in survival analysis. Experimental results demonstrate that the ensemble model achieves superior predictive accuracy (C-index = 0.872) and reliability (Brier score at 9 months = 0.149) compared to individual models. These findings indicate that an ensemble-learning-based model offers promising prospects for more precise individualized treatment of HCC.

The accurate diagnosis of lymph node metastasis in esophageal squamous cell carcinoma is crucial in the treatment workflow, and the process is often time-consuming for clinicians. Recent deep learning models predicting whether lymph nodes are affected by cancer in esophageal cancer cases suffer from challenging node delineation and hence gain poor diagnosis accuracy. This paper proposes an innovative multi-task and multi-scale attention network (M $_{}^2$ ANet) to predict lymph node metastasis precisely. The network softly expands the regions of the node mask and subsequently utilizes the expanded mask to aggregate image features, thereby amplifying the node contexts. It additionally proposes a two-branch training strategy that compels the model to simultaneously predict metastasis probability and node masks, fostering a more comprehensive learning process. The node metastasis prediction performance has been evaluated on a self-collected dataset with 177 patients. Our model finally achieves a competitive accuracy of 83.7% on the test set comprising 577 nodes. With the adaptability to intricate patterns and ability to handle data variations, M $_{}^2$ ANet emerges as a promising tool for robust and comprehensive lymph node metastasis prediction in medical image analysis.

Concussions present a significant risk to athletes, with females exhibiting higher rates and prolonged recovery times than males. Current sideline concussion detection methods, such as the King-Devick test commonly used as a rapid screening tool designed to evaluate eye movement, attention, language, and cognitive processing abilities suffer from validity issues. This is especially true among young athletes highlighting the need for more accurate and objective assessment tools. This study investigates the ability of the Microsoft Kinect V2 pose estimation depth sensor to reliably measure subtle postural stability differences between athletes with a history of concussion and healthy controls. Traditional methods make use of expensive force plates which require trained personnel and controlled environments, limiting their use in resource-limited settings. Inspired by previous research utilizing force plates, our study analyzes video recordings of athletes performing specific exercises to detect dynamic balance deficits. A machine learning approach is employed to predict ground reaction forces from pose estimation video recordings, which are then analyzed to measure time to stabilization. Results reveal significant differences in movement mechanics between concussed and control groups, with the drop vertical jump (DVJ) exercise demonstrating the highest discriminatory power. Notably, concussed individuals exhibit longer time to stabilization (mean difference $=$ 0.089 s, p = 0.046) during DVJ, indicating potential lingering balance impairments. While single-leg squat (SLS) and single-leg hop (SLH) exercises showed fewer discriminatory metrics than DVJ, they still provide valuable insights into balance capabilities. The DVJ yielded the largest statistical difference between injured and healthy male athletes, while the SLH was more effective for females and the SLS, while effective for ACL rehab progress assessment, was equally ineffective for both males and females.

The integration of brain-computer interface (BCI) and virtual reality (VR) systems offers transformative potential for spatial cognition training and assessment. By leveraging artificial intelligence (AI) to analyze electroencephalogram (EEG) data, brain activity patterns during spatial tasks can be decoded with high precision. In this context, a hybrid neural network named MSFHNet is proposed, optimized for extracting spatiotemporal features from spatial cognitive EEG signals. The model employs a hierarchical architecture where its temporal module uses multi-scale dilated convolutions to capture dynamic EEG variations, while its spatial module integrates channel-spatial attention mechanisms to model inter-channel dependencies and spatial distributions. Cross-stacked modules further refine discriminative features through deep-level fusion. Evaluations demonstrate the superiority of MSFHNet in the beta2 frequency band, achieving 98.58% classification accuracy and outperforming existing models. This innovation enhances EEG signal representation, advancing AI-powered BCI-VR systems for robust spatial cognitive training.

Magnetic resonance images (MRI) denoising aims to obtain clean image for further treatment by doctors. Recently, low-rank tensor methods have achieved amazing results in MRI denoising. Nevertheless, imbalanced matricization from Tucker decomposition and nuclear norm penalty mechanism are incapable of fully characterizing the internal structure information of 3D MR image. To mitigate these matters, a novel framework, which combines non-local self-similarity technique and low-rank tensor regularization from tensor train decomposition with balanced matricization, is proposed to noise removal. The constructed fourth-order tensor from non-local self-similarity technique is conducted by tensor train regularization with weighted Schatten-p norm function. The designed method not only considers structural correlation across different dimensions for 3D MR images, but also takes the importance of various singular values into account. Experimental results over synthetic and real images demonstrate that our proposal achieves competitive performance with respect to the state-of-the-art MR images denoising filters (ANLM3D, BM4D, WNNM3D, NLM-tSVD and HOSVD-R) both visually and quantitatively.

Parkinson's disease (PD) is a prevalent neurodegenerative disorder worldwide, often progressing to mild cognitive impairment (MCI) and dementia. Clinical diagnosis of PD mainly depends on characteristic motor symptoms, which can lead to misdiagnosis, underscoring the need for reliable biomarkers. Early detection of PD and effective monitoring of disease progression are crucial for enhancing patient outcomes. Electroencephalogram (EEG) signals, as non-invasive neural recordings, show great promise as diagnostic biomarkers. In this study, we present a novel approach for PD diagnosis through the analysis of EEG signals from distinct brain regions. We used two publicly available EEG datasets and constructed three-dimensional (3D) time-frequency spectrograms for each brain region using the continuous wavelet transform (CWT). To improve feature representation, these spectrograms were encoded in the red-green-blue (RGB) color space. A ResNet18 model was trained separately on the spectrograms of each brain region, and its performance was assessed using the leave-one-subject-out cross-validation (LOSOCV) method. The proposed method achieved classification accuracies of 92.86% and 90.32% on the two datasets, respectively. The experimental results confirm the efficacy of our approach, highlighting its potential as a valuable tool to aid clinical diagnosis of PD.

Mutagenicity is concerning due to its link to genetic mutations, which can lead to cancer and other adverse effects. Early identification of mutagenic compounds in drug development is crucial to prevent unsafe candidates and reduce costs. While computational techniques, especially machine learning (ML) models, have become prevalent for mutagenicity prediction, they typically rely on a single modality. Our work introduces a novel stacked ensemble mutagenicity prediction model that integrates multiple modalities, including SMILES and molecular graphs. These modalities capture diverse molecular information such as substructural, physicochemical, geometrical, and topological features. We use SMILES for deriving substructural, geometrical, and physicochemical data, while a graph attention network (GAT) extracts topological information from molecular graphs. Our model employs a stacked ensemble of ML classifiers and SHAP (Shapley Additive Explanations) to identify the significance of classifiers and key features. Our method outperforms state-of-the-art techniques on two standard datasets, achieving an area under the curve of 95.21% on the Hansen benchmark dataset. This research is expected to interest clinicians and computational biologists in translational research.

Current methods for obtaining accurate joint loading data lack simplicity, efficiency, and cost-effectiveness. This study aims to generate joint loading prediction models using anthropometric parameters and walking speed in overweight or obese females with flexible flatfoot. Sixteen participants' motion capture data from walking trails and anthropometric parameters were collected. The lower limb joint contact forces and the walking speed were calculated via a musculoskeletal model. Regression analysis was used to generate the prediction model. The second peak of knee joint contact force revealed a strong negative correlation with hip circumference and a weak positive correlation with age (p < 0.001 and adjusted R² = 0.720). The peak ankle joint contact force exhibited a strong positive correlation with walking speed while strong negative correlations with waist circumference and lower limb length (p < 0.001 and adjusted R² = 0.782). The first peak of vertical GRF displayed a medium negative correlation with walking speed (p < 0.001 and adjusted R² = 0.750). Anthropometric parameters and walking speed are effective predictors of joint loading. This rapid, low-cost estimation method can be applied to areas such as flexible flatfoot that require assessment of joint stress, thereby saving costs and time.

Biomechanical analyses of the sacroiliac joint (SIJ) are limited. We hypothesize that influence of ligament pre-tension on strain and relative joint movement is morphologically sex-specific and more pronounced than effects of body weight. Finite element models were developed from CTs of a larger cohort (N = 818) for typical male (TMJ) and typical female joint (TFJ) geometries. For different loading scenarios, stresses were higher in TFJ than TMJ for same pre-tension, only considering sex-specific morphology. Loading in antero-posterior direction caused highest stresses and relative movement. Ligament pre-tension was most sensitive with mean sensitivity factor (change output [%]/change input [%]): 71.04/33.64 for translation, 43.09/4.02 for rotation, 2.11/ - 8.97 for stress for TFJ/TMJ respectively. Mean sensitivity factor of ligament stiffness was - 1.14/ - 1.06 for translation, - 0.90/ - 0.89 for rotation and 0.17/0.13 for stress, while mean sensitivity of load intensity was 1.09/1.10 for translation, 0.91/0.88 for rotation and 0.54/0.58 for stress for TFJ/TMJ respectively. Relative motion was more sensitive to parameter variations than stress. The hypothesis was confirmed: influence of ligament pre-tension on stress but especially relative joint movement of SIJ is morphologically sex-specific and larger than body weight effects. As this may play a crucial role in pain development, ligament pre-tension must be verified in situ in the future.