Nutritional composition analysis in food images: an innovative Swin Transformer approach.

Accurate recognition of nutritional components in food is crucial for dietary management and health monitoring. Current methods often rely on traditional chemical analysis techniques, which are time-consuming, require destructive sampling, and are not suitable for large-scale or real-time applications. Therefore, there is a pressing need for efficient, non-destructive, and accurate methods to identify and quantify nutrients in food. In this study, we propose a novel deep learning model that integrates EfficientNet, Swin Transformer, and Feature Pyramid Network (FPN) to enhance the accuracy and efficiency of food nutrient recognition. Our model combines the strengths of EfficientNet for feature extraction, Swin Transformer for capturing long-range dependencies, and FPN for multi-scale feature fusion. Experimental results demonstrate that our model significantly outperforms existing methods. On the Nutrition5k dataset, it achieves a Top-1 accuracy of 79.50% and a Mean Absolute Percentage Error (MAPE) for calorie prediction of 14.72%. On the ChinaMartFood109 dataset, the model achieves a Top-1 accuracy of 80.25% and a calorie MAPE of 15.21%. These results highlight the model's robustness and adaptability across diverse food images, providing a reliable and efficient tool for rapid, non-destructive nutrient detection. This advancement supports better dietary management and enhances the understanding of food nutrition, potentially leading to more effective health monitoring applications.

DECNet: Left Atrial Pulmonary Vein Class Imbalance Classification Network.

In clinical practice, the anatomical classification of pulmonary veins plays a crucial role in the preoperative assessment of atrial fibrillation radiofrequency ablation surgery. Accurate classification of pulmonary vein anatomy assists physicians in selecting appropriate mapping electrodes and avoids causing pulmonary arterial hypertension. Due to the diverse and subtly different anatomical classifications of pulmonary veins, as well as the imbalance in data distribution, deep learning models often exhibit poor expression capability in extracting deep features, leading to misjudgments and affecting classification accuracy. Therefore, in order to solve the problem of unbalanced classification of left atrial pulmonary veins, this paper proposes a network integrating multi-scale feature-enhanced attention and dual-feature extraction classifiers, called DECNet. The multi-scale feature-enhanced attention utilizes multi-scale information to guide the reinforcement of deep features, generating channel weights and spatial weights to enhance the expression capability of deep features. The dual-feature extraction classifier assigns a fixed number of channels to each category, equally evaluating all categories, thus alleviating the learning bias and overfitting caused by data imbalance. By combining the two, the expression capability of deep features is strengthened, achieving accurate classification of left atrial pulmonary vein morphology and providing support for subsequent clinical treatment. The proposed method is evaluated on datasets provided by the People's Hospital of Liaoning Province and the publicly available DermaMNIST dataset, achieving average accuracies of 78.81% and 83.44%, respectively, demonstrating the effectiveness of the proposed approach.

TransDiffSeg: Transformer-Based Conditional Diffusion Segmentation Model for Abdominal Multi-Objective.

In the domain of medical image segmentation, traditional diffusion probabilistic models are hindered by local inductive biases stemming from convolutional operations, constraining their ability to model long-term dependencies and leading to inaccurate mask generation. Conversely, Transformer offers a remedy by obviating the local inductive biases inherent in convolutional operations, thereby enhancing segmentation precision. Currently, the integration of Transformer and convolution operations mainly occurs in two forms: nesting and stacking. However, both methods address the bias elimination at a relatively large granularity, failing to fully leverage the advantages of both approaches. To address this, this paper proposes a conditional diffusion segmentation model named TransDiffSeg, which combines Transformer with convolution operations from traditional diffusion models in a parallel manner. This approach eliminates the accumulated local inductive bias of convolution operations at a finer granularity within each layer. Additionally, an adaptive feature fusion block is employed to merge conditional semantic features and noise features, enhancing global semantic information and reducing the Transformer's sensitivity to noise features. To validate the impact of granularity in bias elimination on performance and the impact of Transformer in alleviating the accumulated local inductive biases of convolutional operations in diffusion probabilistic models, experiments are conducted on the AMOS22 dataset and BTCV dataset. Experimental results demonstrate that eliminating local inductive bias at a finer granularity significantly improves the segmentation performance of diffusion probabilistic models. Furthermore, the results confirm that the finer the granularity of bias elimination, the better the segmentation performance.

Deep Learning Model for Predicting Proliferative Hepatocellular Carcinoma Using Dynamic Contrast-Enhanced MRI: Implications for Early Recurrence Prediction Following Radical Resection.

RATIONALE AND OBJECTIVES: The proliferative nature of hepatocellular carcinoma (HCC) is closely related to early recurrence following radical resection. This study develops and validates a deep learning (DL) prediction model to distinguish between proliferative and non-proliferative HCCs using dynamic contrast-enhanced MRI (DCE-MRI), aiming to refine preoperative assessments and optimize treatment strategies by assessing early recurrence risk. MATERIALS AND METHODS: In this retrospective study, 355 HCC patients from two Chinese medical centers (April 2018-February 2023) who underwent radical resection were included. Patient data were collected from medical records, imaging databases, and pathology reports. The cohort was divided into a training set (n = 251), an internal test set (n = 62), and external test sets (n = 42). A DL model was developed using DCE-MRI images of primary tumors. Clinical and radiological models were generated from their respective features, and fusion strategies were employed for combined model development. The discriminative abilities of the clinical, radiological, DL, and combined models were extensively analyzed. The performances of these models were evaluated against pathological diagnoses, with independent and fusion DL-based models validated for clinical utility in predicting early recurrence. RESULTS: The DL model, using DCE-MRI, outperformed clinical and radiological feature-based models in predicting proliferative HCC. The area under the curve (AUC) for the DL model was 0.98, 0.89, and 0.83 in the training, internal validation, and external validation sets, respectively. The AUCs for the combined DL and clinical feature models were 0.99, 0.86, and 0.83 in these sets, while the AUCs for the combined DL, clinical, and radiological model were 0.99, 0.87, and 0.8, respectively. Among models predicting early recurrence, the DL plus clinical features model showed superior performance. CONCLUSION: The DL-based DCE-MRI model demonstrated robust performance in predicting proliferative HCC and stratifying patient risk for early postoperative recurrence. As a non-invasive tool, it shows promise in enhancing decision-making for individualized HCC management strategies.

Segmentation of the left atrial appendage based on fusion attention.

In clinical practice, the morphology of the left atrial appendage (LAA) plays an important role in the selection of LAA closure devices for LAA closure procedures. The morphology determination is influenced by the segmentation results. The LAA occupies only a small part of the entire 3D medical image, and the segmentation results are more likely to be biased towards the background region, making the segmentation of the LAA challenging. In this paper, we propose a lightweight attention mechanism called fusion attention, which imitates human visual behavior. We process the 3D image of the LAA using a method that involves overview observation followed by detailed observation. In the overview observation stage, the image features are pooled along the three dimensions of length, width, and height. The obtained features from the three dimensions are then separately input into the spatial attention and channel attention modules to learn the regions of interest. In the detailed observation stage, the attention results from the previous stage are fused using element-wise multiplication and combined with the original feature map to enhance feature learning. The fusion attention mechanism was evaluated on a left atrial appendage dataset provided by Liaoning Provincial People's Hospital, resulting in an average Dice coefficient of 0.8855. The results indicate that the fusion attention mechanism achieves better segmentation results on 3D images compared to existing lightweight attention mechanisms.

MOTC: Abdominal Multi-objective Segmentation Model with Parallel Fusion of Global and Local Information.

Convolutional Neural Networks have been widely applied in medical image segmentation. However, the existence of local inductive bias in convolutional operations restricts the modeling of long-term dependencies. The introduction of Transformer enables the modeling of long-term dependencies and partially eliminates the local inductive bias in convolutional operations, thereby improving the accuracy of tasks such as segmentation and classification. Researchers have proposed various hybrid structures combining Transformer and Convolutional Neural Networks. One strategy is to stack Transformer blocks and convolutional blocks to concentrate on eliminating the accumulated local bias of convolutional operations. Another strategy is to nest convolutional blocks and Transformer blocks to eliminate bias within each nested block. However, due to the granularity of bias elimination operations, these two strategies cannot fully exploit the potential of Transformer. In this paper, a parallel hybrid model is proposed for segmentation, which includes a Transformer branch and a Convolutional Neural Network branch in encoder. After parallel feature extraction, inter-layer information fusion and exchange of complementary information are performed between the two branches, simultaneously extracting local and global features while eliminating the local bias generated by convolutional operations within the current layer. A pure convolutional operation is used in decoder to obtain final segmentation results. To validate the impact of the granularity of bias elimination operations on the effectiveness of local bias elimination, the experiments in this paper were conducted on Flare21 dataset and Amos22 dataset. The average Dice coefficient reached 92.65% on Flare21 dataset, and 91.61% on Amos22 dataset, surpassing comparative methods. The experimental results demonstrate that smaller granularity of bias elimination operations leads to better performance.

A new predictor of coronary artery disease in acute ischemic stroke or transient ischemic attack patients: pericarotid fat density.

OBJECTIVES: The study aims to evaluate the incremental predictive value of pericarotid fat density (PFD) on head and neck computed tomography angiography (CTA) for the obstructive coronary artery disease (CAD) (≥ 50% stenosis) relative to a clinical risk model (Framingham risk score (FRS)) and the degree of carotid artery stenosis and plaque type in acute ischemic stroke (AIS) or transient ischemic attack (TIA) patients without a known history of CAD. METHODS: In a cohort of 134 consecutive stable patients diagnosed with AIS or TIA undergoing head and neck CTA between January 2010 and December 2021, pericarotid adipose tissue density (PFD) was quantified using a dedicated software. We collected demographic and clinical data, assessed the risk of CAD using the FRS, and analyzed coronary and carotid artery CTA images. Univariate and multivariate logistic regression analyses were performed to assess associations between FRS, PFD, CTA variables, and obstructive CAD risk. Four prediction models were established to evaluate the incremental predictive value of PFD relative to FRS, stenosis degree, and plaque types. Receiver operating characteristic (ROC) curves were generated, and the areas under the curves (AUC) were compared. RESULTS: Increasing FRS, stenosis degree, and PFD values were positively correlated with obstructive CAD (all p < 0.05). In the predictive models for obstructive CAD, the model incorporating carotid stenosis exhibited superior predictive performance compared to FRS alone (p < 0.05). Moreover, the predictive model integrating PFD demonstrated enhanced performance and yielded the highest AUC of the receiver operator characteristic curve (AUC = 0.783), with sensitivity and specificity values of 86.89% and 65.75%, respectively. CONCLUSION: CTA-derived PFD measurements offer supplementary predictive value for obstructive CAD beyond FRS and stenosis, thereby facilitating improved risk stratification of TIA or stroke patients without a history of CAD history. CLINICAL RELEVANCE STATEMENT: CTA-derived PFD provides incremental predictive value for obstructive coronary artery disease in acute ischemic stroke or transient ischemic attack patients without CAD history, beyond Framingham risk score and carotid artery stenosis degree, improving risk stratification. KEY POINTS: • Pericarotid fat density is associated with obstructive coronary artery disease in acute ischemic stroke or transient ischemic attack patients. • Higher pericarotid fat density corresponds to an increased risk of obstructive coronary artery disease. • Estimation of pericarotid fat density using computed tomography angiography imparts additional predictive value for obstructive CAD in risk stratification of acute ischemic stroke or transient ischemic attack patients.

MSA-YOLOv5: Multi-scale attention-based YOLOv5 for automatic detection of acute ischemic stroke from multi-modality MRI images.

BACKGROUND AND OBJECTIVE: Acute ischemic stroke (AIS) is a common neurological disorder characterized by the sudden onset of cerebral ischemia, leading to functional impairments. Swift and precise detection of AIS lesions is crucial for stroke diagnosis and treatment but poses a significant challenge. This study aims to leverage multimodal fusion technology to combine complementary information from various modalities, thereby enhancing the detection performance of AIS target detection models. METHODS: In this retrospective study of AIS, we collected data from 316 AIS patients and created a multi-modality magnetic resonance imaging (MRI) dataset. We propose a Multi-Scale Attention-based YOLOv5 (MSA-YOLOv5), targeting challenges such as small lesion size and blurred borders at low resolutions. Specifically, we augment YOLOv5 with a prediction head to detect objects at various scales. Next, we replace the original prediction head with a Multi-Scale Swin Transformer Prediction Head (MS-STPH), which reduces computational complexity to linear levels and enhances the ability to detect small lesions. We incorporate a Second-Order channel attention (SOCA) module to adaptively rescale channel features by employing second-order feature statistics for more discriminative representations. Finally, we further validate the effectiveness of our method using the ISLES 2022 dataset. RESULTS: On our in-house AIS dataset, MSA-YOLOv5 achieves a 79.0% mAP0.5, substantially surpassing other single-stage models. Compared to two-stage models, it maintains a comparable performance level while significantly reducing the number of parameters and resolution. On the ISLES 2022 dataset, MSA-YOLOv5 attains an 80.0% mAP0.5, outperforming other network models by a considerable margin. MS-STPH and SOCA modules can significantly increase mAP0.5 by 2.7% and 1.9%, respectively. Visualization interpretability results show that the proposed MSA-YOLOv5 restricts high attention in the small regions of AIS lesions. CONCLUSIONS: The proposed MSA-YOLOv5 is capable of automatically and effectively detecting acute ischemic stroke lesions in multimodal images, particularly for small lesions and artifacts. Our enhanced model reduces the number of parameters while improving detection accuracy. This model can potentially assist radiologists in providing more accurate diagnosis, and enable clinicians to develop better treatment plans.

Deep medullary veins as an important imaging indicator of poor prognosis in acute ischemic stroke: a retrospective cohort survey.

Background: The deep medullary veins (DMVs), which constitute a component of the intracerebral venous circulation system and are part of intracerebral reperfusion mechanisms, have been suggested as a novel imaging marker for cerebral white matter hypersignal and cerebral small vessel disease based on their discontinuous and reduced visual representation. However, the correlation between the number and continuity of visible DMVs and the poor prognosis of acute ischemic stroke (AIS) remains undefined. Magnetic susceptibility-weighted imaging was applied in this study to assess the distribution and structural characteristics of DMVs in patients with AIS and to investigate its relationship with the poor prognosis of those with AIS. Methods: This retrospective study included 90 patients diagnosed with AIS in the middle cerebral artery region by the Neurology Department of Liaoning Provincial People's Hospital. Clinical, laboratory, and cranial magnetic resonance imaging data were collected. After the 3-month follow-up visit, patients were dichotomized into good (0-2 points) and poor (≥3 points) prognosis groups based on the modified Rankin Scale score, and the DMV imaging characteristics were evaluated using a 3-level visual rating scale. The association between DMV and AIS prognosis was determined through Mann-Whitney test and multivariate logistic regression analysis. Results: In univariate analysis, factors that were statistically significant between the different prognostic groups were DMV score (P=0.007), DMV symmetry (P=0.016), infarct size (P=0.029), and admission National Institutes of Health Stroke Scale (NIHSS) score (P<0.001). DMV score had a positive correlation with NIHSS score, (rs=0.209; P=0.048). Logistic regression analysis showed that the DMV score [odds ratio (OR), 1.356; 95% confidence interval (CI): 1.114-1.650; P=0.002], NIHSS score (OR, 1.280; 95% CI: 1.117-1.466; P<0.001), and fasting glucose (OR, 1.220; 95% CI: 1.023-1.456; P=0.027) were risk factors for poor prognosis in those with AIS. Conclusions: Discontinuity in DMV visualization was found to be associated with an unfavorable prognosis for patients AIS. The visual assessment of DMV through susceptibility-weighted imaging has the potential to predict AIS prognosis and furnish valuable insights for clinical treatment.

Value of DWI Combined with Magnetic Resonance Spectroscopy in the Differential Diagnosis between Recurrent Glioma and Radiation Injury: A Meta-Analysis.

To analyse the value of the apparent diffusion coefficient (ADC) in diffusion-weighted imaging (DWI) and the choline (Cho)/creatine (Cr) ratio and Cho/N-acetyl-aspartate (NAA) ratio in magnetic resonance spectroscopy (MRS) in the differential diagnosis between recurrent glioma and radiation injury. Chinese and English studies related to the diagnosis of recurrent glioma and radiation injury using DWI and MRS and published before 15 October 2022 were retrieved from PubMed, Embase, the Cochrane Library, China National Knowledge Infrastructure, China Biomedical Literature Database, VIP Journal Database, and Wanfang Database for a meta-analysis. A total of 11 articles were included in this study. ADC was lower in the recurrent glioma group than in the radiation injury group (standardized mean difference = -1.29, 95% confidence interval (CI) (-1.87, -0.71), P < 0.001). The Cho/Cr ratio was higher in the recurrent glioma group than in the radiation injury group (weighted mean difference = 0.65, 95% CI (0.40, 0.90), and P < 0.001). The Cho/NAA ratio was higher in the recurrent glioma group than in the radiation injury group, as evidenced by the sensitivity analysis. The sensitivity and specificity of the Cho/Cr ratio were 0.85 (0.73-0.92) and 0.82 (0.67-0.91), respectively, and the area under the curve was 0.86. The sensitivity and specificity of the Cho/NAA ratio were 0.82 (0.66-0.91) and 0.94 (0.69-0.99), respectively, and the area under the curve was 0.93. This meta-analysis showed that ADC, Cho/Cr, and Cho/NAA ratios all had high sensitivity and specificity. Therefore, DWI combined with MRS can effectively improve the diagnosis of recurrent glioma and radiation injury.

Abdomen tissues segmentation from computed tomography images using deep learning and level set methods.

Accurate abdomen tissues segmentation is one of the crucial tasks in radiation therapy planning of related diseases. However, abdomen tissues segmentation (liver, kidney) is difficult because the low contrast between abdomen tissues and their surrounding organs. In this paper, an attention-based deep learning method for automated abdomen tissues segmentation is proposed. In our method, image cropping is first applied to the original images. U-net model with attention mechanism is then constructed to obtain the initial abdomen tissues. Finally, level set evolution which consists of three energy terms is used for optimize the initial abdomen segmentation. The proposed model is evaluated across 470 subsets. For liver segmentation, the mean dice are 96.2 and 95.1% for the FLARE21 datasets and the LiTS datasets, respectively. For kidney segmentation, the mean dice are 96.6 and 95.7% for the FLARE21 datasets and the LiTS datasets, respectively. Experimental evaluation exhibits that the proposed method can obtain better segmentation results than other methods.

Optimization extraction and purification of biological activity curcumin from Curcuma longa L by high-performance counter-current chromatography.

The extraction condition of curcumin from Curcuma longa L was optimized through four factors and three levels orthogonal experiment based on the results of single factor tests. Under the optimal conditions: the concentration of ethanol  80%, extraction temperature 70°C, the ratio of liquid to material 20, and extraction time 3 h, a crude extract with the yield of curcumin 56.8 mg/g could be obtained. The isolation and purification of curcuminoids from the crude extract was performed on high performance counter current chromatography employing an optimized solvent system n-hexane/ethyl acetate/methanol/water (2/3/3/1, v/v/v/v). From 97 mg crude sample (in which the purity of curmumin was 68.56%), 67 mg curmumin, 18 mg demethoxycurcumin, and 9.7 mg bisdemethoxycurcumin with a high-performance liquid chromatography purity of 98.26, 97.39, and 98.67%, respectively, were obtained within 70 min. The antioxidant activities and cytotoxicity of purified curcumin was comparable to that of the commercial product, indicating that the biological activity of curcumin could be maintained by this method.

Early Diagnosis of Alzheimer's Disease Based on Resting-State Brain Networks and Deep Learning.

Computerized healthcare has undergone rapid development thanks to the advances in medical imaging and machine learning technologies. Especially, recent progress on deep learning opens a new era for multimedia based clinical decision support. In this paper, we use deep learning with brain network and clinical relevant text information to make early diagnosis of Alzheimer's Disease (AD). The clinical relevant text information includes age, gender, and ApoE gene of the subject. The brain network is constructed by computing the functional connectivity of brain regions using resting-state functional magnetic resonance imaging (R-fMRI) data. A targeted autoencoder network is built to distinguish normal aging from mild cognitive impairment, an early stage of AD. The proposed method reveals discriminative brain network features effectively and provides a reliable classifier for AD detection. Compared to traditional classifiers based on R-fMRI time series data, about 31.21 percent improvement of the prediction accuracy is achieved by the proposed deep learning method, and the standard deviation reduces by 51.23 percent in the best case that means our prediction model is more stable and reliable compared to the traditional methods. Our work excavates deep learning's advantages of classifying high-dimensional multimedia data in medical services, and could help predict and prevent AD at an early stage.

Purple Sweet Potato Attenuate Weight Gain in High Fat Diet Induced Obese Mice.

Purple sweet potato (PSP) is widely grown in Asia and considered as a healthy vegetable. The objective of the current study was to determine the anti-obesity effect of the PSP on high fat diet induced obese C57BL/6J mice. The mice were administrated with high fat diet supplemented with the sweet potato (SP) or PSP at the concentration of 15% and 30% for 12 wk, respectively. The results showed that the supplementation of SP or PSP at 30% significantly ameliorated high fat diet induced obesity and its associated risk factors, including reduction of body weight and fat accumulation, improvement of lipid profile and modulation of energy expenditure. Moreover, PSP also posed beneficial effect on the liver and kidney functions. These results indicate that PSP and SP have anti-obesity effect and are effective to reduce the metabolic risk.