Generative AI in Medical Imaging: Applications, Challenges, and Ethics.

Medical imaging is playing an important role in diagnosis and treatment of diseases. Generative artificial intelligence (AI) have shown great potential in enhancing medical imaging tasks such as data augmentation, image synthesis, image-to-image translation, and radiology report generation. This commentary aims to provide an overview of generative AI in medical imaging, discussing applications, challenges, and ethical considerations, while highlighting future research directions in this rapidly evolving field.

Construction of machine learning-based models for cancer outcomes in low and lower-middle income countries: A scoping review.

Background: The impact and utility of machine learning (ML)-based prediction tools for cancer outcomes including assistive diagnosis, risk stratification, and adjunctive decision-making have been largely described and realized in the high income and upper-middle-income countries. However, statistical projections have estimated higher cancer incidence and mortality risks in low and lower-middle-income countries (LLMICs). Therefore, this review aimed to evaluate the utilization, model construction methods, and degree of implementation of ML-based models for cancer outcomes in LLMICs. Methods: PubMed/Medline, Scopus, and Web of Science databases were searched and articles describing the use of ML-based models for cancer among local populations in LLMICs between 2002 and 2022 were included. A total of 140 articles from 22,516 citations that met the eligibility criteria were included in this study. Results: ML-based models from LLMICs were often based on traditional ML algorithms than deep or deep hybrid learning. We found that the construction of ML-based models was skewed to particular LLMICs such as India, Iran, Pakistan, and Egypt with a paucity of applications in sub-Saharan Africa. Moreover, models for breast, head and neck, and brain cancer outcomes were frequently explored. Many models were deemed suboptimal according to the Prediction model Risk of Bias Assessment tool (PROBAST) due to sample size constraints and technical flaws in ML modeling even though their performance accuracy ranged from 0.65 to 1.00. While the development and internal validation were described for all models included (n=137), only 4.4% (6/137) have been validated in independent cohorts and 0.7% (1/137) have been assessed for clinical impact and efficacy. Conclusion: Overall, the application of ML for modeling cancer outcomes in LLMICs is increasing. However, model development is largely unsatisfactory. We recommend model retraining using larger sample sizes, intensified external validation practices, and increased impact assessment studies using randomized controlled trial designs. Systematic review registration: https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=308345, identifier CRD42022308345.

A Multimodal Deep Learning Approach to Predicting Systemic Diseases from Oral Conditions.

Background: It is known that oral diseases such as periodontal (gum) disease are closely linked to various systemic diseases and disorders. Deep learning advances have the potential to make major contributions to healthcare, particularly in the domains that rely on medical imaging. Incorporating non-imaging information based on clinical and laboratory data may allow clinicians to make more comprehensive and accurate decisions. Methods: Here, we developed a multimodal deep learning method to predict systemic diseases and disorders from oral health conditions. A dual-loss autoencoder was used in the first phase to extract periodontal disease-related features from 1188 panoramic radiographs. Then, in the second phase, we fused the image features with the demographic data and clinical information taken from electronic health records (EHR) to predict systemic diseases. We used receiver operation characteristics (ROC) and accuracy to evaluate our model. The model was further validated by an unseen test dataset. Findings: According to our findings, the top three most accurately predicted chapters, in order, are the Chapters III, VI and IX. The results indicated that the proposed model could predict systemic diseases belonging to Chapters III, VI and IX, with AUC values of 0.92 (95% CI, 0.90-94), 0.87 (95% CI, 0.84-89) and 0.78 (95% CI, 0.75-81), respectively. To assess the robustness of the models, we performed the evaluation on the unseen test dataset for these chapters and the results showed an accuracy of 0.88, 0.82 and 0.72 for Chapters III, VI and IX, respectively. Interpretation: The present study shows that the combination of panoramic radiograph and clinical oral features could be considered to train a fusion deep learning model for predicting systemic diseases and disorders.

Early stage NSCLS patients' prognostic prediction with multi-information using transformer and graph neural network model.

Background: We proposed a population graph with Transformer-generated and clinical features for the purpose of predicting overall survival (OS) and recurrence-free survival (RFS) for patients with early stage non-small cell lung carcinomas and to compare this model with traditional models. Methods: The study included 1705 patients with lung cancer (stages I and II), and a public data set for external validation (n=127). We proposed a graph with edges representing non-imaging patient characteristics and nodes representing imaging tumour region characteristics generated by a pretrained Vision Transformer. The model was compared with a TNM model and a ResNet-Graph model. To evaluate the models' performance, the area under the receiver operator characteristic curve (ROC-AUC) was calculated for both OS and RFS prediction. The Kaplan-Meier method was used to generate prognostic and survival estimates for low- and high-risk groups, along with net reclassification improvement (NRI), integrated discrimination improvement (IDI), and decision curve analysis. An additional subanalysis was conducted to examine the relationship between clinical data and imaging features associated with risk prediction. Results: Our model achieved AUC values of 0.785 (95% confidence interval [CI]: 0.716-0.855) and 0.695 (95% CI: 0.603-0.787) on the testing and external data sets for OS prediction, and 0.726 (95% CI: 0.653-0.800) and 0.700 (95% CI: 0.615-0.785) for RFS prediction. Additional survival analyses indicated that our model outperformed the present TNM and ResNet-Graph models in terms of net benefit for survival prediction. Conclusions: Our Transformer-Graph model was effective at predicting survival in patients with early stage lung cancer, which was constructed using both imaging and non-imaging clinical features. Some high-risk patients were distinguishable by using a similarity score function defined by non-imaging characteristics such as age, gender, histology type, and tumour location, while Transformer-generated features demonstrated additional benefits for patients whose non-imaging characteristics were non-discriminatory for survival outcomes. Funding: The study was supported by the National Natural Science Foundation of China (91959126, 8210071009), and Science and Technology Commission of Shanghai Municipality (20XD1403000, 21YF1438200).

Captopril Alleviates Chondrocyte Senescence in DOCA-Salt Hypertensive Rats Associated with Gut Microbiome Alteration.

Gut microbiota is the key controller of healthy aging. Hypertension and osteoarthritis (OA) are two frequently co-existing age-related pathologies in older adults. Both are associated with gut microbiota dysbiosis. Hereby, we explore gut microbiome alteration in the Deoxycorticosterone acetate (DOCA)-induced hypertensive rat model. Captopril, an anti-hypertensive medicine, was chosen to attenuate joint damage. Knee joints were harvested for radiological and histological examination; meanwhile, fecal samples were collected for 16S rRNA and shotgun sequencing. The 16S rRNA data was annotated using Qiime 2 v2019.10, while metagenomic data was functionally profiled with HUMAnN 2.0 database. Differential abundance analyses were adopted to identify the significant bacterial genera and pathways from the gut microbiota. DOCA-induced hypertension induced p16INK4a+ senescent cells (SnCs) accumulation not only in the aorta and kidney (p < 0.05) but also knee joint, which contributed to articular cartilage degradation and subchondral bone disturbance. Captopril removed the p16INK4a + SnCs from different organs, partially lowered blood pressure, and mitigated cartilage damage. Meanwhile, these alterations were found to associate with the reduction of Escherichia-Shigella levels in the gut microbiome. As such, gut microbiota dysbiosis might emerge as a metabolic link in chondrocyte senescence induced by DOCA-triggered hypertension. The underlying molecular mechanism warrants further investigation.

Self-supervised region-aware segmentation of COVID-19 CT images using 3D GAN and contrastive learning.

Medical image segmentation is a key initial step in several therapeutic applications. While most of the automatic segmentation models are supervised, which require a well-annotated paired dataset, we introduce a novel annotation-free pipeline to perform segmentation of COVID-19 CT images. Our pipeline consists of three main subtasks: automatically generating a 3D pseudo-mask in self-supervised mode using a generative adversarial network (GAN), leveraging the quality of the pseudo-mask, and building a multi-objective segmentation model to predict lesions. Our proposed 3D GAN architecture removes infected regions from COVID-19 images and generates synthesized healthy images while keeping the 3D structure of the lung the same. Then, a 3D pseudo-mask is generated by subtracting the synthesized healthy images from the original COVID-19 CT images. We enhanced pseudo-masks using a contrastive learning approach to build a region-aware segmentation model to focus more on the infected area. The final segmentation model can be used to predict lesions in COVID-19 CT images without any manual annotation at the pixel level. We show that our approach outperforms the existing state-of-the-art unsupervised and weakly-supervised segmentation techniques on three datasets by a reasonable margin. Specifically, our method improves the segmentation results for the CT images with low infection by increasing sensitivity by 20% and the dice score up to 4%. The proposed pipeline overcomes some of the major limitations of existing unsupervised segmentation approaches and opens up a novel horizon for different applications of medical image segmentation.

Artificial intelligence-designed single molar dental prostheses: A protocol of prospective experimental study.

BACKGROUND: Dental prostheses, which aim to replace missing teeth and to restore patients' appearance and oral functions, should be biomimetic and thus adopt the occlusal morphology and three-dimensional (3D) position of healthy natural teeth. Since the teeth of an individual subject are controlled by the same set of genes (genotype) and are exposed to mostly identical oral environment (phenotype), the occlusal morphology and 3D position of teeth of an individual patient are inter-related. It is hypothesized that artificial intelligence (AI) can automate the design of single-tooth dental prostheses after learning the features of the remaining dentition. MATERIALS AND METHODS: This article describes the protocol of a prospective experimental study, which aims to train and to validate the AI system for design of single molar dental prostheses. Maxillary and mandibular dentate teeth models will be collected and digitized from at least 250 volunteers. The (original) digitized maxillary teeth models will be duplicated and processed by removal of right maxillary first molars (FDI tooth 16). Teeth models will be randomly divided into training and validation sets. At least 200 training sets of the original and the processed digitalized teeth models will be input into 3D Generative Adversarial Network (GAN) for training. Among the validation sets, tooth 16 will be generated by AI on 50 processed models and the morphology and 3D position of AI-generated tooth will be compared to that of the natural tooth in the original maxillary teeth model. The use of different GAN algorithms and the need of antagonist mandibular teeth model will be investigated. Results will be reported following the CONSORT-AI.

Re-sensitization of mcr carrying multidrug resistant bacteria to colistin by silver.

Colistin is considered the last-line antimicrobial for the treatment of multidrug-resistant gram-negative bacterial infections. The emergence and spread of superbugs carrying the mobile colistin resistance gene (mcr) have become the most serious and urgent threat to healthcare. Here, we discover that silver (Ag+), including silver nanoparticles, could restore colistin efficacy against mcr-positive bacteria. We show that Ag+ inhibits the activity of the MCR-1 enzyme via substitution of Zn2+ in the active site. Unexpectedly, a tetra-silver center was found in the active-site pocket of MCR-1 as revealed by the X-ray structure of the Ag-bound MCR-1, resulting in the prevention of substrate binding. Moreover, Ag+effectively slows down the development of higher-level resistance and reduces mutation frequency. Importantly, the combined use of Ag+ at a low concentration with colistin could relieve dermonecrotic lesions and reduce the bacterial load of mice infected with mcr-1­carrying pathogens. This study depicts a mechanism of Ag+ inhibition of MCR enzymes and demonstrates the potentials of Ag+ as broad-spectrum inhibitors for the treatment of mcr-positive bacterial infection in combination with colistin.

Comparison of time-to-event machine learning models in predicting oral cavity cancer prognosis.

BACKGROUND: Applying machine learning to predicting oral cavity cancer prognosis is important in selecting candidates for aggressive treatment following diagnosis. However, models proposed so far have only considered cancer survival as discrete rather than dynamic outcomes. OBJECTIVES: To compare the model performance of different machine learning-based algorithms that incorporate time-to-event data. These algorithms included DeepSurv, DeepHit, neural net-extended time-dependent cox model (Cox-Time), and random survival forest (RSF). MATERIALS AND METHODS: Retrospective cohort of 313 oral cavity cancer patients were obtained from electronic health records. Models were trained on patient data following preprocessing. Predictors were based on demographic, clinicopathologic, and treatment information of the cases. Outcomes were the disease-specific and overall survival. Multivariable analyses were conducted to select significant prognostic features associated with tumor prognosis. Two models were generated per algorithm based on all-prognostic features and significant-prognostic features following statistical analysis. Concordance index (c-index) and integrated Brier scores were used as performance evaluators and model stability was assessed using intraclass correlation coefficients (ICC) calculated from these measures obtained from the cross-validation folds. RESULTS: While all models were satisfactory, better discriminatory performance and calibration was observed for disease-specific than overall survival (mean c-index: 0.85 vs 0.74; mean integrated Brier score: 0.12 vs 0.17). DeepSurv performed best in terms of discrimination for both outcomes (c-indices: 0.76 -0.89) while RSF produced better calibrated survival estimates (integrated Brier score: 0.06 -0.09). Model stability of the algorithms varied with the outcomes as Cox-Time had the best intraclass correlation coefficient (mean ICC: 1.00) for disease-specific survival while DeepSurv was most stable for overall survival prediction (mean ICC: 0.99). CONCLUSIONS: Machine learning algorithms based on time-to-event outcomes are successful in predicting oral cavity cancer prognosis with DeepSurv and RSF producing the best discriminative performance and calibration.

Deep Learning Predicts the Malignant-Transformation-Free Survival of Oral Potentially Malignant Disorders.

Machine-intelligence platforms for the prediction of the probability of malignant transformation of oral potentially malignant disorders are required as adjunctive decision-making platforms in contemporary clinical practice. This study utilized time-to-event learning models to predict malignant transformation in oral leukoplakia and oral lichenoid lesions. A total of 1098 patients with oral white lesions from two institutions were included in this study. In all, 26 features available from electronic health records were used to train four learning algorithms-Cox-Time, DeepHit, DeepSurv, random survival forest (RSF)-and one standard statistical method-Cox proportional hazards model. Discriminatory performance, calibration of survival estimates, and model stability were assessed using a concordance index (c-index), integrated Brier score (IBS), and standard deviation of the averaged c-index and IBS following training cross-validation. This study found that DeepSurv (c-index: 0.95, IBS: 0.04) and RSF (c-index: 0.91, IBS: 0.03) were the two outperforming models based on discrimination and calibration following internal validation. However, DeepSurv was more stable than RSF upon cross-validation. External validation confirmed the utility of DeepSurv for discrimination (c-index-0.82 vs. 0.73) and RSF for individual survival estimates (0.18 vs. 0.03). We deployed the DeepSurv model to encourage incipient application in clinical practice. Overall, time-to-event models are successful in predicting the malignant transformation of oral leukoplakia and oral lichenoid lesions.

Arsenic trioxide targets Hsp60, triggering degradation of p53 and survivin.

The mechanisms of action of arsenic trioxide (ATO), a clinically used drug for the treatment of acute promyelocytic leukemia (APL), have been actively studied mainly through characterization of individual putative protein targets. There appear to be no studies at a system level. Herein, we integrate metalloproteomics through a newly developed organoarsenic probe, As-AC (C20H17AsN4O3S2) with quantitative proteomics, allowing 37 arsenic binding and 250 arsenic regulated proteins to be identified in NB4, a human APL cell line. Bioinformatics analysis reveals that ATO disrupts multiple physiological processes, in particular, chaperone-related protein folding and cellular response to stress. Furthermore, we discover heat shock protein 60 (Hsp60) as a vital target of ATO. Through biophysical and cell-based assays, we demonstrate that ATO binds to Hsp60, leading to abolishment of Hsp60 refolding capability. Significantly, the binding of ATO to Hsp60 disrupts the formation of Hsp60-p53 and Hsp60-survivin complexes, resulting in degradation of p53 and survivin. This study provides significant insights into the mechanism of action of ATO at a systemic perspective, and serves as guidance for the rational design of metal-based anticancer drugs.

BING, a novel antimicrobial peptide isolated from Japanese medaka plasma, targets bacterial envelope stress response by suppressing cpxR expression.

Antimicrobial peptides (AMPs) have emerged as a promising alternative to small molecule antibiotics. Although AMPs have previously been isolated in many organisms, efforts on the systematic identification of AMPs in fish have been lagging. Here, we collected peptides from the plasma of medaka (Oryzias latipes) fish. By using mass spectrometry, 6399 unique sequences were identified from the isolated peptides, among which 430 peptides were bioinformatically predicted to be potential AMPs. One of them, a thermostable 13-residue peptide named BING, shows a broad-spectrum toxicity against pathogenic bacteria including drug-resistant strains, at concentrations that presented relatively low toxicity to mammalian cell lines and medaka. Proteomic analysis indicated that BING treatment induced a deregulation of periplasmic peptidyl-prolyl isomerases in gram-negative bacteria. We observed that BING reduced the RNA level of cpxR, an upstream regulator of envelope stress responses. cpxR is known to play a crucial role in the development of antimicrobial resistance, including the regulation of genes involved in drug efflux. BING downregulated the expression of efflux pump components mexB, mexY and oprM in P. aeruginosa and significantly synergised the toxicity of antibiotics towards these bacteria. In addition, exposure to sublethal doses of BING delayed the development of antibiotic resistance. To our knowledge, BING is the first AMP shown to suppress cpxR expression in Gram-negative bacteria. This discovery highlights the cpxR pathway as a potential antimicrobial target.

Prognostic value of non-smoking, non-alcohol drinking status in oral cavity cancer.

OBJECTIVES: To compare the treatment response and prognosis of oral cavity cancer between non-smoking and non-alcohol-drinking (NSND) patients and smoking and alcohol-drinking (SD) patients. METHODS: A total of 313 consecutively treated patients from 2000 to 2019 were included. Demographic, clinicopathologic, treatment, and prognosis information were obtained. Relapse-free survival (RFS), disease-specific survival (DSS), and overall survival (OS) were compared between NSND and SD groups using Kaplan-Meier plots, log-rank test, and multivariate Cox regression analysis. RESULTS: Sample prevalence of NSND patients was 54.6%. These patients were predominantly females in their eighth decade with lower prevalence of floor of the mouth cancers compared to SD patients (1.8% vs 14.8%). No difference in the RFS and DSS between both groups was found following multivariable analysis; however, NSND patients had better OS (HR (95% CI) - 0.47 (0.29-0.75); p = 0.002). Extracapsular extension was associated with significantly poorer OS, DSS, and RFS in this oral cavity cancer cohort. CONCLUSION: Treatment response and disease-specific prognosis are comparable between NSND and SD patients with oral cavity cancer. However, NSND patients have better OS. CLINICAL RELEVANCE: This study shows that oral cavity cancer in NSND is not less or more aggressive compared to SD patients. Although better survival is expected for NSND than SD patients, this is likely due to the reduced incidence of other chronic diseases in the NSND group.

Deciphering molecular mechanism of silver by integrated omic approaches enables enhancing its antimicrobial efficacy in E. coli.

Despite the broad-spectrum antimicrobial activities of silver, its internal usage is restricted, owing to the toxicity. Strategies to enhance its efficacy are highly desirable but rely heavily on the understanding of its molecular mechanism of action. However, up to now, no direct silver-targeting proteins have been mined at a proteome-wide scale, which hinders systemic studies on the biological pathways interrupted by silver. Herein, we build up a unique system, namely liquid chromatography gel electrophoresis inductively coupled plasma mass spectrometry (LC-GE-ICP-MS), allowing 34 proteins directly bound by silver ions to be identified in Escherichia coli. By using integrated omic approaches, including metalloproteomics, metabolomics, bioinformatics, and systemic biology, we delineated the first dynamic antimicrobial actions of silver (Ag+) in E. coli, i.e., it primarily damages multiple enzymes in glycolysis and tricarboxylic acid (TCA) cycle, leading to the stalling of the oxidative branch of the TCA cycle and an adaptive metabolic divergence to the reductive glyoxylate pathway. It then further damages the adaptive glyoxylate pathway and suppresses the cellular oxidative stress responses, causing systemic damages and death of the bacterium. To harness these novel findings, we coadministrated metabolites involved in the Krebs cycles with Ag+ and found that they can significantly potentiate the efficacy of silver both in vitro and in an animal model. Our study reveals the comprehensive and dynamic mechanisms of Ag+ toxicity in E. coli cells and offers a novel and general approach for deciphering molecular mechanisms of metallodrugs in various pathogens and cells to facilitate the development of new therapeutics.

MetaMarker: a pipeline for de novo discovery of novel metagenomic biomarkers.

SUMMARY: We present MetaMarker, a pipeline for discovering metagenomic biomarkers from whole-metagenome sequencing samples. Different from existing methods, MetaMarker is based on a de novo approach that does not require mapping raw reads to a reference database. We applied MetaMarker on whole-metagenome sequencing of colorectal cancer (CRC) stool samples from France to discover CRC specific metagenomic biomarkers. We showed robustness of the discovered biomarkers by validating in independent samples from Hong Kong, Austria, Germany and Denmark. We further demonstrated these biomarkers could be used to build a machine learning classifier for CRC prediction. AVAILABILITY AND IMPLEMENTATION: MetaMarker is freely available at https://bitbucket.org/mkoohim/metamarker under GPLv3 license. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Novel Neural Network Approach to Predict Drug-Target Interactions Based on Drug Side Effects and Genome-Wide Association Studies.

AIMS: We propose a novel machine learning approach to expand the knowledge about drug-target interactions. Our method may help to develop effective, less harmful treatment strategies and to enable the detection of novel indications for existing drugs. METHODS: We developed a novel machine learning strategy to predict drug-target interactions based on drug side effects and traits from genome-wide association studies. We integrated data from the databases SIDER and GWASdb and utilized them in a unique way by a neural network approach. RESULTS: We validate our method using drug-target interactions from the STITCH database. In addition, we compare the chemical similarity of the predicted target to known targets of the drug under consideration and present literature-based evidence for predicted interactions. We find drug combination warnings for drugs we predict to target the same protein, hinting to synergistic effects aggravating harmful events. This substantiates the translational value of our approach, because we are able to detect drugs that should be taken together with care due to common mechanisms of action. CONCLUSION: Taken together, we conclude that our approach is able to generate a novel and clinically applicable insight into the molecular determinants of drug action.

Metallochaperone UreG serves as a new target for design of urease inhibitor: A novel strategy for development of antimicrobials.

Urease as a potential target of antimicrobial drugs has received considerable attention given its versatile roles in microbial infection. Development of effective urease inhibitors, however, is a significant challenge due to the deeply buried active site and highly specific substrate of a bacterial urease. Conventionally, urease inhibitors are designed by either targeting the active site or mimicking substrate of urease, which is not efficient. Up to now, only one effective inhibitor-acetohydroxamic acid (AHA)-is clinically available, but it has adverse side effects. Herein, we demonstrate that a clinically used drug, colloidal bismuth subcitrate, utilizes an unusual way to inhibit urease activity, i.e., disruption of urease maturation process via functional perturbation of a metallochaperone, UreG. Similar phenomena were also observed in various pathogenic bacteria, suggesting that UreG may serve as a general target for design of new types of urease inhibitors. Using Helicobacter pylori UreG as a showcase, by virtual screening combined with experimental validation, we show that two compounds targeting UreG also efficiently inhibited urease activity with inhibitory concentration (IC)50 values of micromolar level, resulting in attenuated virulence of the pathogen. We further demonstrate the efficacy of the compounds in a mammalian cell infection model. This study opens up a new opportunity for the design of more effective urease inhibitors and clearly indicates that metallochaperones involved in the maturation of important microbial metalloenzymes serve as new targets for devising a new type of antimicrobial drugs.

Integration of fluorescence imaging with proteomics enables visualization and identification of metallo-proteomes in living cells.

Metalloproteins account for nearly one-third of proteins in proteomes. To date, the identification of metalloproteins relies mainly on protein purification and the subsequent characterization of bound metals, which often leads to losses of metal ions bound weakly and transiently. Herein, we developed a strategy to visualize and subsequently identify endogenous metalloproteins and metal-binding proteins in living cells via integration of fluorescence imaging with proteomics. We synthesized a "metal-tunable" fluorescent probe (denoted as Mn+-TRACER) that rapidly enters cells to target proteins with 4-40 fold fluorescence enhancements. By using Ni2+-TRACER as an example, we demonstrate the feasibility of tracking Ni2+-binding proteins in vitro, while cellular small molecules exhibit negligible interference on the labeling. We identified 44 Ni2+-binding proteins from microbes using Helicobacter pylori as a showcase. We further applied Cu2+-TRACER to mammalian cells and found 54 Cu2+-binding proteins. The strategy we report here provides a great opportunity to track various endogenous metallo-proteomes and to mine potential targets of metallodrugs.