External Set Research Articles

Headspace – Solid phase microextraction vs liquid injection GC-MS analysis of essential oils: Prediction of linear retention indices by multiple linear regression

AbstractDue to the relative independence from the operational parameters, the linear retention indices (LRIs) are useful tool in gas chromatography-mass spectrometry (GC-MS) qualitative analysis. The aim of the current study was to develop a multiple linear regression (MLR) model for the prediction of LRIs as a function of selected molecular descriptors. Liquid injection GC-MS was used for the analysis of Essential oils (Rose, Lavender and Peppermint) separating the ingredients by a semi-standard non-polar stationary phase. As a result, a sum of 103 compounds were identified and their experimental LRIs were derived relying on reference measurements of a standard mixture of n-alkanes (from C8 to C20). As a next step, a set of molecular descriptors was generated for the distinguished chemical structures. Further, a stepwise MLR was applied for the selection of the significant descriptors (variables) which can be used to predict the LRIs. From an exploit set of over 2000 molecular descriptors, it was found that only 16 can be regarded as significant and independent variables. At this point split validation was applied: the identified compounds were randomly divided into training (85%) and validation (15%) sets. The training set (87 compounds) was used to derive two MLR models by applying i) the ‘enter’ algorithm (R2 = 0.9960, RMSЕ = 17) and ii) the ‘stepwise’ one (R2 = 0.9958, RMSЕ = 17). The predictive power was assessed by the validation set (16 compounds) as follows i) q2F1 = 0.9896, RMSE = 25 and ii) q2F1 = 0.9886, RMSE = 26, respectively. The adequateness of both regression approaches was further evaluated. Newly developed headspace-solid phase microextraction (HS-SPME) procedures in combination with GC-MS were used for an alternative analysis of the studied Essential oils. Twelve additional compounds, not previously detected by the liquid sample introduction mode of analysis, were identified for which the values of the significant descriptors were within the working range of the developed MLRs. For the last compounds, the LRIs were calculated and the experimental data was used as an external set for assessment of the regression models. The predictive power for both regression approaches was assessed as follows: Enter RMSE = 41, q2F2 = 0.9503 and Stepwise RMSE = 40, q2F2 = 0.9521.

Clinical Evaluation of Deep Learning for Tumor Delineation on 18F-FDG PET/CT of Head and Neck Cancer.

Artificial intelligence (AI) may decrease 18F-FDG PET/CT-based gross tumor volume (GTV) delineation variability and automate tumor-volume-derived image biomarker extraction. Hence, we aimed to identify and evaluate promising state-of-the-art deep learning methods for head and neck cancer (HNC) PET GTV delineation. Methods: We trained and evaluated deep learning methods using retrospectively included scans of HNC patients referred for radiotherapy between January 2014 and December 2019 (ISRCTN16907234). We used 3 test datasets: an internal set to compare methods, another internal set to compare AI-to-expert variability and expert interobserver variability (IOV), and an external set to compare internal and external AI-to-expert variability. Expert PET GTVs were used as the reference standard. Our benchmark IOV was measured using the PET GTV of 6 experts. The primary outcome was the Dice similarity coefficient (DSC). ANOVA was used to compare methods, a paired t test was used to compare AI-to-expert variability and expert IOV, an unpaired t test was used to compare internal and external AI-to-expert variability, and post hoc Bland-Altman analysis was used to evaluate biomarker agreement. Results: In total, 1,220 18F-FDG PET/CT scans of 1,190 patients (mean age ± SD, 63 ± 10 y; 858 men) were included, and 5 deep learning methods were trained using 5-fold cross-validation (n = 805). The nnU-Net method achieved the highest similarity (DSC, 0.80 [95% CI, 0.77-0.86]; n = 196). We found no evidence of a difference between expert IOV and AI-to-expert variability (DSC, 0.78 for AI vs. 0.82 for experts; mean difference of 0.04 [95% CI, -0.01 to 0.09]; P = 0.12; n = 64). We found no evidence of a difference between the internal and external AI-to-expert variability (DSC, 0.80 internally vs. 0.81 externally; mean difference of 0.004 [95% CI, -0.05 to 0.04]; P = 0.87; n = 125). PET GTV-derived biomarkers of AI were in good agreement with experts. Conclusion: Deep learning can be used to automate 18F-FDG PET/CT tumor-volume-derived imaging biomarkers, and the deep-learning-based volumes have the potential to assist clinical tumor volume delineation in radiation oncology.

A case study in underspecification of UG: External Pair Merge of v and T

AbstractPrevious works have explored the options of Pair Merging R and v as well as T and C, respectively, yielding R‐v and T‐C with various consequences. This paper proposes that v and T can yield the complex head v‐T, an amalgam formed by external Pair Merge. Some empirical ramifications for the verb cluster and the middle field in German are explored. If tenable, the current approach supports the idea that heads can Merge freely, either by internal or external Set Merge, as well as by internal or by external Pair Merge.

A general method for assessment of glass transition temperature of polymeric materials only from various structural factors in their repeating unit structure

The glass transition temperature (Tg) of polymeric materials can be used to assess their stability because physical properties can be changed at this temperature. In contrast to previous general methods for predicting Tg of different polymers where they need computer codes, and expert users, the novel method can be used as a simple computer code because it requires only various structural factors in their repeating unit structure. The most extensive available reported data of Tg for 388 different organic polymers, such as polyacrylates, polystyrenes, polyvinyls, polycarbonates, polyxylylenes, polyimides, and polyquinolines, are used to derive and test the new model. Assessment of the high reliability of the novel method is done by considering diverse types of statistical factors and compared with the outputs of five models. The values of the coefficient of determination (r2)/Root Mean Square Error (RMSE) of the new model for 154, 52, and 38 data of training, test, and external sets are 0.846/0.049, 0.832/0.052, and 0.841/0.064, respectively, which are higher/lower (or close) than the five comparative models. For further 104 polymers where the computed outputs of comparative models were unavailable, the r2 value is relatively good, i.e., 0.851.

Development and multi-institutional validation of a deep learning model for grading of vesicoureteral reflux on voiding cystourethrogram: a retrospective multicenter study

Voiding cystourethrography (VCUG) is the gold standard for the diagnosis and grading of vesicoureteral reflux (VUR). However, VUR grading from voiding cystourethrograms is highly subjective with low reliability. This study aimed to develop a deep learning model to improve reliability for VUR grading on VCUG and compare its performance to that of clinicians. In this retrospective study in China, VCUG images were collected between January 2019 and September 2022 from our institution as an internal dataset for training and 4 external data sets as external testing set for validation. Samples were divided into training (N=1000) and validation sets (N=500), internal testing set (N=168), and external testing set (N=280). An ensemble learning-based model, Deep-VCUG, using Res-Net 101 and the voting methods was developed to predict VUR grade. The grading performance was assessed using heatmaps, area under the receiver operating characteristic curve (AUC), sensitivity, specificity, accuracy, and F1 score in the internal and external testing set. The performances of four clinicians (2 pediatric urologists and 2 radiologists) with and without the Deep-VCUG assisted to predict VUR grade were explored in external testing sets. A total of 1948 VCUG images were collected (Internal dataset=1668; multi-center external dataset=280). For assessing unilateral VUR grading, the Deep-VCUG achieved AUCs of 0.962 (95% confidence interval [CI]: 0.943-0.978) and 0.944 (95% [CI]: 0.921-0.964) in the internal and external testing sets, respectively, for bilateral VUR grading, the Deep-VCUG also achieved high AUCs of 0.960 (95% [CI]: 0.922-0.983) and 0.924 (95% [CI]: 0.887-0.957). The Deep-VCUG model using voting method outperformed single model and clinician in terms of classification based on VCUG image. Moreover, Under the Dee-VCUG assisted, the classification ability of junior and senior clinicians was significantly improved. The Deep-VCUG model is a generalizable, objective, and accurate tool for vesicoureteral reflux grading based on VCUG imaging and had good assistance with clinicians to VUR grading applicability. This study was supported by Natural Science Foundation of China, "Fuqing Scholar" Student Scientific Research Program of Shanghai Medical College, Fudan University, and the Program of Greater Bay Area Institute of Precision Medicine (Guangzhou).

Identification of a 2-aminothiazole framework using classical QSAR model targeting chloroquine-sensitive Plasmodium falciparum

Chloroquine-sensitive Plasmodium falciparum is the most deadly form of human malaria. It is associated with a number of mutations in P. falciparum. Chloroquine-resistant transporter is a protein that serves as a transporter in the parasite's digesting vacuole membrane. In order to combat chloroquine-sensitive P. falciparum strains (NF54), this study employs QSAR modelling to examine possible structural alterations of 2-amino-thiazole derivatives. The traditional QSAR model was built using the PaDEL descriptor via QSARINS software. The model was found to have an internal cross-validation value of Q2loo = 0.7890 and an external validation parameter of RMSE ext = 0.6938. The predicted pIC50 values from the QSAR techniques for the case study chemicals were compared and found to be well fitted to the model and well predicted for the external set of compounds. The outcome demonstrates the value of using the suggested method in the creation of new medication candidates could fill the critical gap in scientific knowledge and open up novel possibilities for pharmaceutical development.

An automatic parathyroid recognition and segmentation model based on deep learning of near-infrared autofluorescence imaging.

Near-infrared autofluorescence imaging (NIFI) can be used to identify parathyroid gland (PG) during surgery. The purpose of the study is to establish a new model, help surgeons better identify, and protect PGs. Five hundred and twenty three NIFI images were selected. The PGs were recorded by NIFI and marked with artificial intelligence (AI) model. The recognition rate for PGs was calculated. Analyze the differences between surgeons of different years of experience and AI recognition, and evaluate the diagnostic and therapeutic efficacy of AI model. Our model achieved 83.5% precision and 57.8% recall in the internal validation set. The visual recognition rate of AI model was 85.2% and 82.4% on internal and external sets. The PG recognition rate of AI model is higher than that of junior surgeons (p < 0.05). This AI model will help surgeons identify PGs, and develop their learning ability and self-confidence.

Pharmacophore & QSAR Guided Design, Synthesis, Pharmacokinetics, and in vitro Evaluation of Curcumin Analogs for Anticancer Activity.

As a part of our discovery of plant-based lead molecules, we provide a helpful tool, which helps in identification, designing, optimising, structural modifications, and prediction of curcumin, to discover novel analogs with enhanced bioavailability, pharmacologically safe, and anticancer potential. QSAR (Quantitative structure-activity relationship) and pharmacophore mapping models were developed and further used to design, synthesize, pharmacokinetics, and in vitro evaluation of curcumin analogs for anticancer activity. The QSAR model yielded a high activity-descriptors relationship accuracy (r2) of 84%, a high activity prediction accuracy (rcv2) of 81%, and external set prediction accuracy of 89%. The QSAR study indicates that the five chemical descriptors were significantly correlated with anticancer activity. The important pharmacophore features identified were a hydrogen bond acceptor, a hydrophobic centre, and a negative ionizable centre. The model&#039;s predictive ability was evaluated against a set of chemically synthesized curcumin analogs. Among the tested compounds, nine curcumin analogs were found with IC50 values of 0.10 to 1.86 µg/mL. The active analogs were assessed for pharmacokinetics compliance. EGFR was identified as a potential target of synthesized active curcumin analogs through docking studies. Integrating in silico design, QSAR-driven virtual screening, chemical synthesis, and experimental in vitro evaluation may lead to the early discovery of novel and promising anticancer compounds from natural sources. The developed QSAR model and common pharmacophore generation were used as a designing and predictive tool to develop novel curcumin analogs. This study may help optimize the therapeutic relationships of studied compounds for further drug development and their potential safety concerns. This study may guide compound selection and designing novel active chemical scaffolds or new combinatorial libraries of the curcumin series.

Development and validation of outcome prediction model for reperfusion therapy in acute ischemic stroke using nomogram and machine learning.

To develop logistic regression nomogram and machine learning (ML)-based models to predict 3-month unfavorable functional outcome for acute ischemic stroke (AIS) patients undergoing reperfusion therapy. Patients undergoing reperfusion therapy (intravenous thrombolysis and/or endovascular treatment) were prospectively recruited. Unfavorable outcome was defined as 3-month modified Rankin Scale (mRS) score 3-6. The independent risk factors associated with unfavorable outcome were obtained by regression analysis and included in the prediction model. The performance of nomogram was assessed by the area under the curve (AUC), calibration curve, and decision curve analysis (DCA). ML models were compared with nomogram using AUC; the generalizability of all models was ascertained in an external cohort. A total of 505 patients were enrolled, with 256 in the model construction, and 249 in the external validation. Five variables were identified as prognostic factors: baseline NIHSS, D-dimer level, random blood glucose (RBG), blood urea nitrogen (BUN), and systolic blood pressure (SBP) before reperfusion. The AUC values of nomogram were 0.865, 0.818, and 0.779 in the training set, test set, and external validation, respectively. The calibration curve and DCA indicated appreciable reliability and good net benefits. The best three ML models were extra trees (ET), CatBoost, and random forest (RF) models; all of them showed favorable discrimination in the training cohort, and confirmed in the test and external sets. Baseline NIHSS, D-dimer, RBG, BUN, and SBP before reperfusion were independent predictors for 3-month unfavorable outcome after reperfusion therapy in AIS patients. Both nomogram and ML models showed good discrimination and generalizability.

DNA-methylation variability in normal mucosa: a field cancerization marker in patients with adenomatous polyps.

The phenomenon of field cancerization reflects the transition of normal cells into those predisposed to cancer. Assessing the scope and intensity of this process in the colon may support risk prediction and colorectal cancer prevention. The SWEPIC study, encompassing 1,111 participants for DNA methylation analysis and a subset of 84 for RNA-seq, was employed to detect field cancerization in individuals with adenomatous polyps (AP). Methylation variations were evaluated for their discriminative capability, including in external cohorts, genomic localization, clinical correlations, and associated RNA expression patterns. Normal cecal tissue of individuals harboring an AP in the proximal colon manifested dysregulated DNA methylation compared to tissue from healthy individuals at 558 unique loci. Leveraging these adenoma-related differentially variable and methylated CpGs (aDVMCs), our classifier discerned between healthy and AP-adjacent tissues across SWEPIC datasets (cross-validated ROC AUC [0.63-0.81]), including within age-stratified cohorts. This discriminative capacity was validated in three external sets, differentiating healthy from cancer-adjacent tissue (ROC AUC: [0.82-0.88]). Notably, aDVMC dysregulation correlated with polyp multiplicity. More than 50% of aDVMCs were significantly associated with age. These aDVMCs were enriched in active regions of the genome (p < .001), and associated genes exhibited altered expression in AP-adjacent tissues. Our findings underscore the early onset of field cancerization in the right colon during the neoplastic transformation process. A more extensive validation of aDVMC dysregulation as a stratification tool could pave the way for enhanced surveillance approaches, especially given its linkage to adenoma emergence.

AUTOMATED DETECTION OF VITRITIS USING ULTRAWIDE-FIELD FUNDUS PHOTOGRAPHS AND DEEP LEARNING.

Evaluate the performance of a deep learning algorithm for the automated detection and grading of vitritis on ultrawide-field imaging. Cross-sectional noninterventional study. Ultrawide-field fundus retinophotographs of uveitis patients were used. Vitreous haze was defined according to the six steps of the Standardization of Uveitis Nomenclature classification. The deep learning framework TensorFlow and the DenseNet121 convolutional neural network were used to perform the classification task. The best fitted model was tested in a validation study. One thousand one hundred eighty-one images were included. The performance of the model for the detection of vitritis was good with a sensitivity of 91%, a specificity of 89%, an accuracy of 0.90, and an area under the receiver operating characteristics curve of 0.97. When used on an external set of images, the accuracy for the detection of vitritis was 0.78. The accuracy to classify vitritis in one of the six Standardization of Uveitis Nomenclature grades was limited (0.61) but improved to 0.75 when the grades were grouped into three categories. When accepting an error of one grade, the accuracy for the six-class classification increased to 0.90, suggesting the need for a larger sample to improve the model performances. A new deep learning model based on ultrawide-field fundus imaging that produces an efficient tool for the detection of vitritis was described. The performance of the model for the grading into three categories of increasing vitritis severity was acceptable. The performance for the six-class grading of vitritis was limited but can probably be improved with a larger set of images.

Application of near-infrared hyperspectral imaging for determination of cheese chemical composition

Cheeses vary in chemical composition as they have different applications according to the technological features; some hard cheeses are consumed fresh while melting is a desirable characteristic for others. The conventional reference methods for determination of cheese composition and quality are accurate, but include physical-chemical analyses that are time-consuming and require use of chemical reagents. The objective of this work was to apply near-infrared hyperspectral imaging (NIR-HSI) as a fast method for prediction and visualization of spatial distribution of chemical composition of different types of cheeses (mozzarela, requeijão do norte, minas meia cura, coalho, prato). Moisture, fat, protein, and free oil were measured, and used as reference for prediction models, while spectral information extracted from images in the range of 928–2524 nm were used as predictors. Partial Least Squares (PLS) regression models were built with full spectra and using only few selected regions with higher importance, which were applied to the images to build chemical maps and visualize the spatial variation of the cheese composition. The regression models were able to predict moisture (R2P = 0.92, RMSEP = 1.1%); fat (R2P = 0.84, RMSEP = 1.5%); protein (R2P = 0.80, RMSEP = 1.8%); and free oil (R2P =0.64, RMSEP = 0.5%) for an external set of samples, while requiring a small number of latent variables (LVs) (between 3 and 4). The chemical maps allowed visualization of the spatial distribution for each attribute, demonstrating that the composition may vary within sample. The study demonstrated the potential of NIR-HSI for the determination of cheese composition and possible application to cheese quality control.

Clinical prediction models for cervical lymph node metastasis of papillary thyroid carcinoma.

Accurate preoperative diagnosis of lymph node metastasis (LNM) in papillary thyroid carcinoma (PTC) remains an unsolved problem. This study aimed to construct a nomogram and scoring system for predicting LNM based on the clinical characteristics of patients with PTC. 1400 patients with PTC who underwent thyroidectomy and lymph node dissection at the First Affiliated Hospital of Sun Yat-sen University were retrospectively enrolled and randomly divided into training and internal testing sets. Furthermore, 692 patients with PTC from three other medical centers were collected as external testing sets. Least absolute shrinkage and selection operator (LASSO) was used to screen the predictors, and a nomogram was constructed. In addition, a scoring system was constructed using 10-fold cross-validation. The performances of the two models were verified among datasets and compared with preoperative ultrasound (US). Six independent predictors were included in the multivariate logistic model: age, sex, US diagnosis of LNM, tumor diameter, location, and thyroid peroxidase antibody level. The areas under the receiver operating characteristic curve (AUROC) (95% confidence interval) of this nomogram in the training, internal testing, and three external testing sets were 0.816 (0.791-0.840), 0.782 (0.727-0.837), 0.759 (0.699-0.819), 0.749 (0.667-0.831), and 0.777 (0.726-0.828), respectively. The AUROC of the scoring system were 0.810 (0.785-0.835), 0.772 (0.718-0.826), 0.736 (0.675-0.798), 0.717 (0.635-0.799) and 0.756 (0.704-0.808), respectively. The prediction performances were both significantly superior to those of preoperative US (P < 0.001). The nomogram and scoring system performed well in different datasets and significantly improved the preoperative prediction of LNM than US alone.

Real-World Application of Artificial Intelligence for Detecting Pathologic Gastric Atypia and Neoplastic Lesions.

Results of initial endoscopic biopsy of gastric lesions often differ from those of the final pathological diagnosis. We evaluated whether an artificial intelligence-based gastric lesion detection and diagnostic system, ENdoscopy as AI-powered Device Computer Aided Diagnosis for Gastroscopy (ENAD CAD-G), could reduce this discrepancy. We retrospectively collected 24,948 endoscopic images of early gastric cancers (EGCs), dysplasia, and benign lesions from 9,892 patients who underwent esophagogastroduodenoscopy between 2011 and 2021. The diagnostic performance of ENAD CAD-G was evaluated using the following real-world datasets: patients referred from community clinics with initial biopsy results of atypia (n=154), participants who underwent endoscopic resection for neoplasms (Internal video set, n=140), and participants who underwent endoscopy for screening or suspicion of gastric neoplasm referred from community clinics (External video set, n=296). ENAD CAD-G classified the referred gastric lesions of atypia into EGC (accuracy, 82.47%; 95% confidence interval [CI], 76.46%-88.47%), dysplasia (88.31%; 83.24%-93.39%), and benign lesions (83.12%; 77.20%-89.03%). In the Internal video set, ENAD CAD-G identified dysplasia and EGC with diagnostic accuracies of 88.57% (95% CI, 83.30%-93.84%) and 91.43% (86.79%-96.07%), respectively, compared with an accuracy of 60.71% (52.62%-68.80%) for the initial biopsy results (P<0.001). In the External video set, ENAD CAD-G classified EGC, dysplasia, and benign lesions with diagnostic accuracies of 87.50% (83.73%-91.27%), 90.54% (87.21%-93.87%), and 88.85% (85.27%-92.44%), respectively. ENAD CAD-G is superior to initial biopsy for the detection and diagnosis of gastric lesions that require endoscopic resection. ENAD CAD-G can assist community endoscopists in identifying gastric lesions that require endoscopic resection.

Refined image quality assessment for color fundus photography based on deep learning.

Color fundus photography is widely used in clinical and screening settings for eye diseases. Poor image quality greatly affects the reliability of further evaluation and diagnosis. In this study, we developed an automated assessment module for color fundus photography image quality assessment using deep learning. A total of 55,931 color fundus photography images from multiple centers in Shanghai and the public database were collected and annotated as training, validation, and testing data sets. The pre-diagnosis image quality assessment module based on the multi-task deep neural network was designed. The detailed criterion of color fundus photography image quality including three subcategories with three levels of grading was applied to improve precision and objectivity. The auxiliary tasks such as the localization of the optic nerve head and macula, the classification of laterality, and the field of view were also included to assist the quality assessment. Finally, we validated our module internally and externally by evaluating the area under the receiver operating characteristic curve, sensitivity, specificity, accuracy, and quadratic weighted Kappa. The "Location" subcategory achieved area under the receiver operating characteristic curves of 0.991, 0.920, and 0.946 for the three grades, respectively. The "Clarity" subcategory achieved area under the receiver operating characteristic curves of 0.980, 0.917, and 0.954 for the three grades, respectively. The "Artifact" subcategory achieved area under the receiver operating characteristic curves of 0.976, 0.952, and 0.986 for the three grades, respectively. The accuracy and Kappa of overall quality reach 88.15% and 89.70%, respectively, on the internal set. These two indicators on the external set were 86.63% and 88.55%, respectively, which were very close to that of the internal set. This work showed that our deep module was able to evaluate the color fundus photography image quality using more detailed three subcategories with three grade criteria. The promising results on both internal and external validation indicated the strength and generalizability of our module.

Ligand based pharmacophore modelling and integrated computational approaches in the quest for small molecule inhibitors against hCA IX.

Carbonic anhydrase IX is an important biomarker to fight hypoxic tumours in both initial and metastatic stages of many forms of cancer. Overexpression of hCA IX in the hypoxic environment, has an active role in pH maintenance and makes the hCA IX a better target for the inhibitors targeting specific types of cancer stages. Being a member of the carbonic anhydrase family and having sixteen isoforms, it is important to have a selective inhibition of hCA IX to limit the disruption in the biological and metabolic pathways where other isoforms of hCA are localised and to avoid the other toxicity and adverse effects we try to find selective hCA IX inhibitors from a natural derivative. In the process of finding selective hCA inhibitors we developed a pharmacophore model based on existing inhibitors with IC50 values of less than 50 nm, which is then validated with the external decoy set and used for database searching followed by virtual screening to identify the hits based on the pharmacophore fit score and RMSD. Molecular docking studies were performed to identify protein ligand interaction and molecular dynamics simulation studies to analyse the stability of the complex and DFT studies were carried out. The initial screening yielded 43 hits with the RMSD value less than 1, which when subjected to docking exhibited very good interaction with key residues ZN301, HIS94, HIS96 and HIS119. The top 4 compounds in the molecular dynamics simulation studies for 100 ns provided useful insights on the stability of the complex and the DFT studies confirmed the energy variation between HOMO and LUMO is within an acceptable range. An average binding score of -7.8 Kcal mol-1 for the lead compounds and high stability margin in the dynamics study concludes that these lead compounds demonstrated outstanding potential for hCA IX inhibitory action theoretically and that further experimental studies for selective inhibition are inevitable.

Joint Energy-based Model for Semi-supervised Respiratory Sound Classification: A Method of Insensitive to Distribution Mismatch.

Semi-supervised learning effectively mitigates the lack of labeled data by introducing extensive unlabeled data. Despite achieving success in respiratory sound classification, in practice, it usually takes years to acquire a sufficiently sizeable unlabeled set, which consequently results in an extension of the research timeline. Considering that there are also respiratory sounds available in other related tasks, like breath phase detection and COVID-19 detection, it might be an alternative manner to treat these external samples as unlabeled data for respiratory sound classification. However, since these external samples are collected in different scenarios via different devices, there inevitably exists a distribution mismatch between the labeled and external unlabeled data. For existing methods, they usually assume that the labeled and unlabeled data follow the same data distribution. Therefore, they cannot benefit from external samples. To utilize external unlabeled data, we propose a semi-supervised method based on Joint Energy-based Model (JEM) in this paper. During training, the method attempts to use only the essential semantic components within the samples to model the data distribution. When non-semantic components like recording environments and devices vary, as these non-semantic components have a small impact on the model training, a relatively accurate distribution estimation is obtained. Therefore, the method exhibits insensitivity to the distribution mismatch, enabling the model to leverage external unlabeled data to mitigate the lack of labeled data. Taking ICBHI 2017 as the labeled set, HF_Lung_V1 and COVID-19 Sounds as the external unlabeled sets, the proposed method exceeds the baseline by 12.86.

Distinct MRI characteristics of spinal cord diffuse midline glioma, H3 K27-altered in comparison to spinal cord glioma without H3 K27-alteration and demyelination disorder.

An applicable magnetic resonance imaging (MRI) biomarker for diffuse midline glioma (DMG), H3 K27-altered of the spinal cord is important for non-invasive diagnosis. To evaluate the efficacy of conventional MRI (cMRI) in distinguishing between DMGs, H3 K27-altered, gliomas without H3 K27-alteration, and demyelinating lesions in the spinal cord. Between January 2017 and February 2023, patients with pathology-confirmed spinal cord gliomas (including ependymomas) with definite H3 K27 status and demyelinating diseases diagnosed by recognized criteria were recruited as the training set for this retrospective study. Morphologic parameter assessment was performed by two neuroradiologists on T1-weighted, T2-weighted, and contrast-enhanced T1-weighted imaging. Variables with high inter- and intra-observer agreement were included in univariable correlation analysis and multivariable logistic regression. The performance of the final model was verified by internal and external testing sets. The training cohort included 21 patients with DMGs (13 men; mean age = 34.57 ± 13.489 years), 21 with wild-type gliomas (10 men; mean age = 46.76 ± 17.017 years), and 20 with demyelinating diseases (5 men; mean age = 49.50 ± 18.872 years). A significant difference was observed in MRI features, including cyst(s), hemorrhage, pial thickening with enhancement, and the maximum anteroposterior diameter of the spinal cord. The prediction model, integrating age, age2, and morphological characteristics, demonstrated good performance in the internal and external testing cohort (accuracy: 0.810 and 0.800, specificity: 0.810 and 0.720, sensitivity: 0.872 and 0.849, respectively). Based on cMRI, we developed a model with good performance for differentiating among DMGs, H3 K27-altered, wild-type glioma, and demyelinating lesions in the spinal cord.

SaB-Net: Self-attention backward network for gastric tumor segmentation in CT images

Gastric cancer is a significant contributor to cancer-related fatalities globally. The automated segmentation of gastric tumors has the potential to analyze the medical condition of patients and enhance the likelihood of surgical treatment success. However, the development of an automatic solution is challenged by the heterogeneous intensity distribution of gastric tumors in computed tomography (CT) images, the low-intensity contrast between organs, and the high variability in the stomach shapes and gastric tumors in different patients. To address these challenges, we propose a self-attention backward network (SaB-Net) for gastric tumor segmentation (GTS) in CT images by introducing a self-attention backward layer (SaB-Layer) to feed the self-attention information learned at the deep layer back to the shallow layers. The SaB-Layer efficiently extracts tumor information from CT images and integrates the information into the network, thereby enhancing the network’s tumor segmentation ability. We employed datasets from two centers, one for model training and testing and the other for external validation. The model achieved dice scores of 0.8456 on the test set and 0.8068 on the external verification set. Moreover, we validated the model’s transfer learning ability on a publicly available liver cancer dataset, achieving results comparable to state-of-the-art liver cancer segmentation models recently developed. SaB-Net has strong potential for assisting in the clinical diagnosis of and therapy for gastric cancer. Our implementation is available at https://github.com/TyrionJ/SaB-Net.

Detection and subtyping of basal cell carcinoma in whole-slide histopathology using weakly-supervised learning

The frequency of basal cell carcinoma (BCC) cases is putting an increasing strain on dermatopathologists. BCC is the most common type of skin cancer, and its incidence is increasing rapidly worldwide. AI can play a significant role in reducing the time and effort required for BCC diagnostics and thus improve the overall efficiency of the process. To train such an AI system in a fully-supervised fashion however, would require a large amount of pixel-level annotation by already strained dermatopathologists. Therefore, in this study, our primary objective was to develop a weakly-supervised for the identification of basal cell carcinoma (BCC) and the stratification of BCC into low-risk and high-risk categories within histopathology whole-slide images (WSI). We compared Clustering-constrained Attention Multiple instance learning (CLAM) with StreamingCLAM and hypothesized that the latter would be the superior approach. A total of 5147 images were used to train and validate the models, which were subsequently tested on an internal set of 949 images and an external set of 183 images. The labels for training were automatically extracted from free-text pathology reports using a rule-based approach. All data has been made available through the COBRA dataset. The results showed that both the CLAM and StreamingCLAM models achieved high performance for the detection of BCC, with an area under the ROC curve (AUC) of 0.994 and 0.997, respectively, on the internal test set and 0.983 and 0.993 on the external dataset. Furthermore, the models performed well on risk stratification, with AUC values of 0.912 and 0.931, respectively, on the internal set, and 0.851 and 0.883 on the external set. In every single metric the StreamingCLAM model outperformed the CLAM model or is on par. The performance of both models was comparable to that of two pathologists who scored 240 BCC positive slides. Additionally, in the public test set, StreamingCLAM demonstrated a comparable AUC of 0.958, markedly superior to CLAM’s 0.803. This difference was statistically significant and emphasized the strength and better adaptability of the StreamingCLAM approach.