Dietary Assessment with Multimodal ChatGPT: A Systematic Analysis.

Conventional approaches to dietary assessment are primarily grounded in self-reporting methods or structured interviews conducted under the supervision of dietitians. These methods, however, are often subjective, potentially inaccurate, and time-intensive. Although artificial intelligence (AI)-based solutions have been devised to automate the dietary assessment process, prior AI methodologies tackle dietary assessment in a fragmented landscape (e.g., merely recognizing food types or estimating portion size), and encounter challenges in their ability to generalize across a diverse range of food categories, dietary behaviors, and cultural contexts. Recently, the emergence of multimodal foundation models, such as GPT-4V, has exhibited transformative potential across a wide range of tasks (e.g., scene understanding and image captioning) in various research domains. These models have demonstrated remarkable generalist intelligence and accuracy, owing to their large-scale pre-training on broad datasets and substantially scaled model size. In this study, we explore the application of GPT-4V powering multimodal ChatGPT for dietary assessment, along with prompt engineering and passive monitoring techniques. We evaluated the proposed pipeline using a self-collected, semi free-living dietary intake dataset comprising 16 real-life eating episodes, captured through wearable cameras. Our findings reveal that GPT-4V excels in food detection under challenging conditions without any fine-tuning or adaptation using food-specific datasets. By guiding the model with specific language prompts (e.g., African cuisine), it shifts from recognizing common staples like rice and bread to accurately identifying regional dishes like banku and ugali. Another GPT-4V's standout feature is its contextual awareness. GPT-4V can leverage surrounding objects as scale references to deduce the portion sizes of food items, further facilitating the process of dietary assessment.

In-tube micro-pyramidal silicon nanopore for inertial-kinetic sensing of single molecules.

Electrokinetic force has been the major choice for driving the translocation of molecules through a nanopore. However, the use of this approach is limited by an uncontrollable translocation speed, resulting in non-uniform conductance signals with low conformational sensitivity, which hinders the accurate discrimination of the molecules. Here, we show the use of inertial-kinetic translocation induced by spinning an in-tube micro-pyramidal silicon nanopore fabricated using photovoltaic electrochemical etch-stop technique for biomolecular sensing. By adjusting the kinetic properties of a funnel-shaped centrifugal force field while maintaining a counter-balanced state of electrophoretic and electroosmotic effect in the nanopore, we achieved regulated translocation of proteins and obtained stable signals of long and adjustable dwell times and high conformational sensitivity. Moreover, we demonstrated instantaneous sensing and discrimination of molecular conformations and longitudinal monitoring of molecular reactions and conformation changes by wirelessly measuring characteristic features in current blockade readouts using the in-tube nanopore device.

Hypo-connectivity of the primary somatosensory cortex in Parkinson's disease: a resting-state functional MRI study.

Background: Parkinson's disease (PD) is characterized by a range of motor symptoms as well as documented sensory dysfunction. This sensory dysfunction can present itself either as a "pure" sensory disturbance or as a consequence of sensory-motor integration within the central nervous system. This study aims to investigate changes in the functional connectivity of the primary somatosensory cortex (S1) and its clinical significance in Parkinson's disease (PD), an area that has received limited attention in previous neuroimaging studies. Methods: This study included thirty-three patients with PD and thirty-four healthy controls (HCs). Clinical evaluations were conducted to assess the clinical manifestations, severity, and functional capacity of all the patients. Resting-state functional MRI (fMRI) was employed to evaluate the functional connectivity of six paired S1 subregions in the participants. Seed-based correlation (SBC) analysis was utilized to construct the correlation matrix among the subregions and to generate connectivity maps between the subregions and the remaining brain voxels. Finally, the study employed partial least-squares (PLS) correlation analysis to investigate the association between modified functional connectivity and clinical characteristics in PD patients. Results: In the correlation matrix, patients with PD demonstrated a notable decrease in functional connectivity across various S1 subregions in comparison to HCs (p < 0.001, corrected using network-based methods). In connectivity maps, hypo-connectivity was primarily observed in the sensorimotor network as common patterns (p < 0.001, corrected for false discovery rate) and in the default mode network (DMN) as distinct patterns. Moreover, this study identified a negative association between the correlation matrix within S1 subregions and the scores for axial symptoms and postural instability/gait difficulty (PIGD) in PD patients. Nevertheless, a direct relationship between the connectivity maps of S1 subregions and clinical assessment scales was not established. Conclusion: This study offers novel insights into the neurobiological mechanisms that contribute to S1 dysfunction in PD, highlighting the significant involvement of S1 hypo-connectivity in the motor disturbances observed in PD patients.

CauDR: A causality-inspired domain generalization framework for fundus-based diabetic retinopathy grading.

Diabetic retinopathy (DR) is the most common diabetic complication, which usually leads to retinal damage, vision loss, and even blindness. A computer-aided DR grading system has a significant impact on helping ophthalmologists with rapid screening and diagnosis. Recent advances in fundus photography have precipitated the development of novel retinal imaging cameras and their subsequent implementation in clinical practice. However, most deep learning-based algorithms for DR grading demonstrate limited generalization across domains. This inferior performance stems from variance in imaging protocols and devices inducing domain shifts. We posit that declining model performance between domains arises from learning spurious correlations in the data. Incorporating do-operations from causality analysis into model architectures may mitigate this issue and improve generalizability. Specifically, a novel universal structural causal model (SCM) was proposed to analyze spurious correlations in fundus imaging. Building on this, a causality-inspired diabetic retinopathy grading framework named CauDR was developed to eliminate spurious correlations and achieve more generalizable DR diagnostics. Furthermore, existing datasets were reorganized into 4DR benchmark for DG scenario. Results demonstrate the effectiveness and the state-of-the-art (SOTA) performance of CauDR. Diabetic retinopathy (DR) is the most common diabetic complication, which usually leads to retinal damage, vision loss, and even blindness. A computer-aided DR grading system has a significant impact on helping ophthalmologists with rapid screening and diagnosis. Recent advances in fundus photography have precipitated the development of novel retinal imaging cameras and their subsequent implementation in clinical practice. However, most deep learning-based algorithms for DR grading demonstrate limited generalization across domains. This inferior performance stems from variance in imaging protocols and devices inducing domain shifts. We posit that declining model performance between domains arises from learning spurious correlations in the data. Incorporating do-operations from causality analysis into model architectures may mitigate this issue and improve generalizability. Specifically, a novel universal structural causal model (SCM) was proposed to analyze spurious correlations in fundus imaging. Building on this, a causality-inspired diabetic retinopathy grading framework named CauDR was developed to eliminate spurious correlations and achieve more generalizable DR diagnostics. Furthermore, existing datasets were reorganized into 4DR benchmark for DG scenario. Results demonstrate the effectiveness and the state-of-the-art (SOTA) performance of CauDR.

Advancing MicroRNA Detection: Enhanced Biotin-Streptavidin Dual-Mode Phase Imaging Surface Plasmon Resonance Aptasensor.

MicroRNAs (miRNAs) are novel tumor biomarkers owing to their important physiological functions in cell communication and the progression of multiple diseases. Due to the small molecular weight, short sequence length, and low concentration levels of miRNA, miRNA detection presents substantial challenges, requiring the advancement of more refined and sensitive techniques. There is an urgent demand for the development of a rapid, user-friendly, and sensitive miRNA analysis method. Here, we developed an enhanced biotin-streptavidin dual-mode phase imaging surface plasmon resonance (PI-SPR) aptasensor for sensitive and rapid detection of miRNA. Initially, we evaluated the linear sensing range for miRNA detection across two distinct sensing modalities and investigated the physical factors that influence the sensing signal in the aptamer-miRNA interaction within the PI-SPR aptasensor. Then, an enhanced biotin-streptavidin amplification strategy was introduced in the PI-SPR aptasensor, which effectively reduced the nonspecific adsorption by 20% and improved the limit of detection by 548 times. Furthermore, we have produced three types of tumor marker chips, which utilize the rapid sensing mode (less than 2 min) of PI-SPR aptasensor to achieve simultaneous detection of multiple miRNA markers in the serum from clinical cancer patients. This work not only developed a new approach to detect miRNA in different application scenarios but also provided a new reference for the application of the biotin-streptavidin amplification system in the detection of other small biomolecules.

The effects of pterygomaxillary disjunction in surgically assisted rapid maxillary expansion: A systematic review and meta-analysis.

OBJECTIVE: This study aims to assess the expansive effects of pterygomaxillary disjunction (PMD) in surgically assisted rapid maxillary expansion (SARME) surgery using a meta-analysis approach. MATERIALS AND METHODS: The study conducted a comprehensive literature search across five databases: PubMed, Scopus, Medline, Embase, and Cochrane, adhering to the PRISMA 2020 guidelines. Dental alterations were assessed using either cone-beam computed tomography (CBCT) or dental casts, while skeletal changes were exclusively measured from CBCT scans. We analysed the dentoskeletal changes between PMD +/- groups and conducted a within-group comparison. The primary focus of the results was on the mean differences observed in pre- and post-operative measurements. RESULTS: Dental expansion was larger in the PMD+ group but not statistically significant. Skeletal expansion showed a significantly larger expansion in the posterior region in the PMD+ group (P = .033). Without PMD, anterior palatal expansion was significantly larger (P = .03), and the buccal tipping of posterior teeth was also significantly larger (P = .011) to achieve acceptable dental expansion outcomes. CONCLUSIONS: Both PMD +/- groups of SARME surgery can achieve satisfactory dental expansion outcomes. However, bone expansion and tooth inclination are also important factors that influence orthodontic treatment and post-expansion stability. By reducing the bony resistance with PMD, larger posterior palatal expansion and more parallel bony expansion are observed. In contrast, without PMD, there is smaller palatal expansion and greater tooth inclination in the posterior region. This could potentially lead to compromised periodontal conditions following expansion.

Sustained AhR activity programs memory fate of early effector CD8+ T cells.

Identification of mechanisms that program early effector T cells to either terminal effector T (Teff) or memory T (Tm) cells has important implications for protective immunity against infections and cancers. Here, we show that the cytosolic transcription factor aryl hydrocarbon receptor (AhR) is used by early Teff cells to program memory fate. Upon antigen engagement, AhR is rapidly up-regulated via reactive oxygen species signaling in early CD8+ Teff cells, which does not affect the effector response, but is required for memory formation. Mechanistically, activated CD8+ T cells up-regulate HIF-1α to compete with AhR for HIF-1ß, leading to the loss of AhR activity in HIF-1αhigh short-lived effector cells, but sustained in HIF-1αlow memory precursor effector cells (MPECs) with the help of autocrine IL-2. AhR then licenses CD8+ MPECs in a quiescent state for memory formation. These findings partially resolve the long-standing issue of how Teff cells are regulated to differentiate into memory cells.

GlanceSeg: Real-time microaneurysm lesion segmentation with gaze-map-guided foundation model for early detection of diabetic retinopathy.

Early-stage diabetic retinopathy (DR) presents challenges in clinical diagnosis due to inconspicuous and minute microaneurysms (MAs), resulting in limited research in this area. Additionally, the potential of emerging foundation models, such as the segment anything model (SAM), in medical scenarios remains rarely explored. In this work, we propose a human-in-the-loop, label-free early DR diagnosis framework called GlanceSeg, based on SAM. GlanceSeg enables real-time segmentation of MA lesions as ophthalmologists review fundus images. Our human-in-the-loop framework integrates the ophthalmologist's gaze maps, allowing for rough localization of minute lesions in fundus images. Subsequently, a saliency map is generated based on the located region of interest, which provides prompt points to assist the foundation model in efficiently segmenting MAs. Finally, a domain knowledge filtering (DKF) module refines the segmentation of minute lesions. We conducted experiments on two newly-built public datasets, i.e., IDRiD and Retinal-Lesions, and validated the feasibility and superiority of GlanceSeg through visualized illustrations and quantitative measures. Additionally, we demonstrated that GlanceSeg improves annotation efficiency for clinicians and further enhances segmentation performance through fine-tuning using annotations. The clinician-friendly GlanceSeg is able to segment small lesions in real-time, showing potential for clinical applications.

Deep-qGFP: A Generalist Deep Learning Assisted Pipeline for Accurate Quantification of Green Fluorescent Protein Labeled Biological Samples in Microreactors.

Absolute quantification of biological samples provides precise numerical expression levels, enhancing accuracy, and performance for rare templates. Current methodologies, however, face challenges-flow cytometers are costly and complex, whereas fluorescence imaging, relying on software or manual counting, is time-consuming and error-prone. It is presented that Deep-qGFP, a deep learning-aided pipeline for the automated detection and classification of green fluorescent protein (GFP) labeled microreactors, enables real-time absolute quantification. This approach achieves an accuracy of 96.23% and accurately measures the sizes and occupancy status of microreactors using standard laboratory fluorescence microscopes, providing precise template concentrations. Deep-qGFP demonstrates remarkable speed, quantifying over 2000 microreactors across ten images in just 2.5 seconds, with a dynamic range of 56.52-1569.43 copies µL-1 . The method demonstrates impressive generalization capabilities, successfully applied to various GFP-labeling scenarios, including droplet-based, microwell-based, and agarose-based applications. Notably, Deep-qGFP is the first all-in-one image analysis algorithm successfully implemented in droplet digital polymerase chain reaction (PCR), microwell digital PCR, droplet single-cell sequencing, agarose digital PCR, and bacterial quantification, without requiring transfer learning, modifications, or retraining. This makes Deep-qGFP readily applicable in biomedical laboratories and holds potential for broader clinical applications.

Oxygen affinity and light intensity induced robust phosphorus removal and fragile ammonia removal in a non-aerated bacteria-algae system.

Non-aerated bacteria-algae system gaining O2 through photosynthesis presents an alternative for costly mechanical aeration. This study investigated oxygen supply and performance of nutrients removal at low and high light intensity (LL and HL). The results showed that P removal was high and robust (LL 97 ± 1.8 %, HL 95 % ± 2.9 %), while NH4+-N removal fluctuated dramatically (LL 66 ± 14.7 %, HL 84 ± 8.6 %). Oxygen generated at illumination of 200 µmol m-2 s-1, 6 h was sufficient to sustain aerobic phase for 2.25 g/L MLSS. However, O2 produced by algae was preferentially captured in the order of heterotrophic bacteria (HB), ammonia oxidizing bacteria (AOB), nitrite oxidizing bacteria (NOB). Oxygen affinity coupled with light intensity led to NOB suppression with stable nitrite accumulation ratio of 57 %. Free nitrous acid (FNA) and light stimulated the abundance of denitrifying polyphosphate accumulating organism (DPAO) of Flavobacterium, but with declined P-accumulating metabolism (PAM) of P release, P/C, K/P and Mg/P ratios. Flavobacterium and cyanobacteria Leptolyngbya, along with biologically induced CaP in extracellular polymeric substances was the key to robust P removal. AOB of Ellin6067 and DPAO of Flavobacteria offer a promising scenario for partial nitrification-denitrifying phosphorus removal.

Label-free biosensing with singular-phase-enhanced lateral position shift based on atomically thin plasmonic nanomaterials.

Rapid plasmonic biosensing has attracted wide attention in early disease diagnosis and molecular biology research. However, it was still challenging for conventional angle-interrogating plasmonic sensors to obtain higher sensitivity without secondary amplifying labels such as plasmonic nanoparticles. To address this issue, we developed a plasmonic biosensor based on the enhanced lateral position shift by phase singularity. Such singularity presents as a sudden phase retardation at the dark point of reflection from resonating plasmonic substrate, leading to a giant position shift on reflected beam. Herein, for the first time, the atomically thin layer of Ge2Sb2Te5 (GST) on silver nanofilm was demonstrated as a novel phase-response-enhancing plasmonic material. The GST layer was not only precisely engineered to singularize phase change but also served as a protective layer for active silver nanofilm. This new configuration has achieved a record-breaking largest position shift of 439.3 µm measured in calibration experiments with an ultra-high sensitivity of 1.72 × 108 nm RIU-1 (refractive index unit). The detection limit was determined to be 6.97 × 10-7 RIU with a 0.12 µm position resolution. Besides, a large figure of merit (FOM) of 4.54 × 1011 µm (RIU∙°)-1 was evaluated for such position shift interrogation, enabling the labelfree detection of trace amounts of biomolecules. In targeted biosensing experiments, the optimized sensor has successfully detected small cytokine biomarkers (TNF-α and IL-6) with the lowest concentration of 1 × 10-16 M. These two molecules are the key proinflammatory cancer markers in clinical diagnosis, which cannot be directly screened by current clinical techniques. To further validate the selectivity of our sensing systems, we also measured the affinity of integrin binding to arginylglycylaspartic acid (RGD) peptide (a key protein interaction in cell adhesion) with different Mn2+ ion concentrations, ranging from 1 nM to 1 mM.

CRISPR-powered optothermal nanotweezers: Diverse bio-nanoparticle manipulation and single nucleotide identification.

Optothermal nanotweezers have emerged as an innovative optical manipulation technique in the past decade, which revolutionized classical optical manipulation by efficiently capturing a broader range of nanoparticles. However, the optothermal temperature field was merely employed for in-situ manipulation of nanoparticles, its potential for identifying bio-nanoparticles remains largely untapped. Hence, based on the synergistic effect of optothermal manipulation and CRIPSR-based bio-detection, we developed CRISPR-powered optothermal nanotweezers (CRONT). Specifically, by harnessing diffusiophoresis and thermo-osmotic flows near the substrate upon optothermal excitation, we successfully trapped and enriched DNA functionalized gold nanoparticles, CRISPR-associated proteins, as well as DNA strands. Remarkably, we built an optothermal scheme for enhancing CRISPR-based single-nucleotide polymorphism (SNP) detection at single molecule level, while also introducing a novel CRISPR methodology for observing nucleotide cleavage. Therefore, this innovative approach has endowed optical tweezers with DNA identification ability in aqueous solution which was unattainable before. With its high specificity and feasibility for in-situ bio-nanoparticle manipulation and identification, CRONT will become a universal tool in point-of-care diagnosis, biophotonics, and bio-nanotechnology.

Allyl Isothiocyanate Suppresses the Proliferation in Oral Squamous Cell Carcinoma via Mediating the KDM8/CCNA1 Axis.

The dysregulated expression of cyclin genes can lead to the uncontrolled proliferation of cancer cells. Histone demethylase Jumonji-C domain-containing protein 5 (KDM8, JMJD5) and cyclin A1 (CCNA1) are pivotal in cell cycle progression. A promising candidate for augmenting cancer treatment is Allyl isothiocyanate (AITC), a natural dietary chemotherapeutic and epigenetic modulator. This study aimed to investigate AITC's impact on the KDM8/CCNA1 axis to elucidate its role in oral squamous cell carcinoma (OSCC) tumorigenesis. The expression of KDM8 and CCNA1 was assessed using a tissue microarray (TMA) immunohistochemistry (IHC) assay. In vitro experiments with OSCC cell lines and in vivo experiments with patient-derived tumor xenograft (PDTX) and SAS subcutaneous xenograft tumor models were conducted to explore AITC's effects on their expression and cell proliferation. The results showed elevated KDM8 and CCNA1 levels in the OSCC patient samples. AITC exhibited inhibitory effects on OSCC tumor growth in vitro and in vivo. Additionally, AITC downregulated KDM8 and CCNA1 expression while inducing histone H3K36me2 expression in oral cancer cells. These findings underscore AITC's remarkable anticancer properties against oral cancer, highlighting its potential as a therapeutic option for oral cancer treatment by disrupting the cell cycle by targeting the KDM8/CCNA1 axis.

Large AI Models in Health Informatics: Applications, Challenges, and the Future.

Large AI models, or foundation models, are models recently emerging with massive scales both parameter-wise and data-wise, the magnitudes of which can reach beyond billions. Once pretrained, large AI models demonstrate impressive performance in various downstream tasks. A prime example is ChatGPT, whose capability has compelled people's imagination about the far-reaching influence that large AI models can have and their potential to transform different domains of our lives. In health informatics, the advent of large AI models has brought new paradigms for the design of methodologies. The scale of multi-modal data in the biomedical and health domain has been ever-expanding especially since the community embraced the era of deep learning, which provides the ground to develop, validate, and advance large AI models for breakthroughs in health-related areas. This article presents a comprehensive review of large AI models, from background to their applications. We identify seven key sectors in which large AI models are applicable and might have substantial influence, including: 1) bioinformatics; 2) medical diagnosis; 3) medical imaging; 4) medical informatics; 5) medical education; 6) public health; and 7) medical robotics. We examine their challenges, followed by a critical discussion about potential future directions and pitfalls of large AI models in transforming the field of health informatics.

ER stress aggravates diaphragm weakness through activating PERK/JNK signaling in obesity hypoventilation syndrome.

OBJECTIVE: Obesity hypoventilation syndrome is associated with diaphragmatic dysfunction. This study aimed to explore the role of endoplasmic reticulum (ER) stress in mediating obesity-induced diaphragmatic dysfunction. METHODS: A pulmonary function test and ultrasound were applied to evaluate diaphragmatic function and magnetic resonance imaging was applied to measure diaphragmatic lipid deposition in human patients. For the mechanistic study, obese mice were introduced to a high-fat diet for 24 weeks, followed by diaphragmatic ultrasound measurement, transcriptomic sequencing, and respective biochemical analysis. Automatic force mapping was applied to measure the mechanical properties of C2C12 myotubes. RESULTS: People with obesity showed significant diaphragm weakness and lipid accumulation, which was further confirmed in obese mice. Consistently, diaphragms from obese mice showed altered gene expression profile in lipid metabolism and activation of ER stress response, indicated by elevated protein kinase R-like ER kinase (PERK) and c-Jun NH2 -terminal kinase (JNK) activation. In C2C12 myotubes, inhibition of PERK or JNK signaling abrogated lipotoxicity-induced intracellular lipid deposition and insulin resistance. Inhibition of JNK signaling reversed lipotoxicity-induced impairment of elasticity in C2C12 myotubes. CONCLUSIONS: These data suggest that ectopic lipid deposition impairs the diaphragmatic function of people with obesity. Activation of PERK/JNK signaling is involved in the pathogenesis of lipotoxicity-induced diaphragm weakness in obesity hypoventilation syndrome.

Unraveling the complexity of Optical Coherence Tomography image segmentation using machine and deep learning techniques: A review.

Optical Coherence Tomography (OCT) is an emerging technology that provides three-dimensional images of the microanatomy of biological tissue in-vivo and at micrometer-scale resolution. OCT imaging has been widely used to diagnose and manage various medical diseases, such as macular degeneration, glaucoma, and coronary artery disease. Despite its wide range of applications, the segmentation of OCT images remains difficult due to the complexity of tissue structures and the presence of artifacts. In recent years, different approaches have been used for OCT image segmentation, such as intensity-based, region-based, and deep learning-based methods. This paper reviews the major advances in state-of-the-art OCT image segmentation techniques. It provides an overview of the advantages and limitations of each method and presents the most relevant research works related to OCT image segmentation. It also provides an overview of existing datasets and discusses potential clinical applications. Additionally, this review gives an in-depth analysis of machine learning and deep learning approaches for OCT image segmentation. It outlines challenges and opportunities for further research in this field.

Ultrasensitive and Specific Clustered Regularly Interspaced Short Palindromic Repeats Empowered a Plasmonic Fiber Tip System for Amplification-Free Monkeypox Virus Detection and Genotyping.

The urgent necessity for highly sensitive diagnostic tools has been accentuated by the ongoing mpox (monkeypox) virus pandemic due to the complexity in identifying asymptomatic and presymptomatic carriers. Traditional polymerase chain reaction-based tests, despite their effectiveness, are hampered by limited specificity, expensive and bulky equipment, labor-intensive operations, and time-consuming procedures. In this study, we present a clustered regularly interspaced short palindromic repeats (CRISPR)/Cas12a-based diagnostic platform with a surface plasmon resonance-based fiber tip (CRISPR-SPR-FT) biosensor. The compact CRISPR-SPR-FT biosensor, with a 125 µm diameter, offers high stability and portability, enabling exceptional specificity for mpox diagnosis and precise identification of samples with a fatal mutation site (L108F) in the F8L gene. The CRISPR-SPR-FT system can analyze viral double-stranded DNA from mpox virus without amplification in under 1.5 h with a limit of detection below 5 aM in plasmids and about 59.5 copies/µL when in pseudovirus-spiked blood samples. Our CRISPR-SPR-FT biosensor thus offers fast, sensitive, portable, and accurate target nucleic acid sequence detection.

Generalist Vision Foundation Models for Medical Imaging: A Case Study of Segment Anything Model on Zero-Shot Medical Segmentation.

Medical image analysis plays an important role in clinical diagnosis. In this paper, we examine the recent Segment Anything Model (SAM) on medical images, and report both quantitative and qualitative zero-shot segmentation results on nine medical image segmentation benchmarks, covering various imaging modalities, such as optical coherence tomography (OCT), magnetic resonance imaging (MRI), and computed tomography (CT), as well as different applications including dermatology, ophthalmology, and radiology. Those benchmarks are representative and commonly used in model development. Our experimental results indicate that while SAM presents remarkable segmentation performance on images from the general domain, its zero-shot segmentation ability remains restricted for out-of-distribution images, e.g., medical images. In addition, SAM exhibits inconsistent zero-shot segmentation performance across different unseen medical domains. For certain structured targets, e.g., blood vessels, the zero-shot segmentation of SAM completely failed. In contrast, a simple fine-tuning of it with a small amount of data could lead to remarkable improvement of the segmentation quality, showing the great potential and feasibility of using fine-tuned SAM to achieve accurate medical image segmentation for a precision diagnostics. Our study indicates the versatility of generalist vision foundation models on medical imaging, and their great potential to achieve desired performance through fine-turning and eventually address the challenges associated with accessing large and diverse medical datasets in support of clinical diagnostics.

Chemo-drugs in cell microparticles reset antitumor activity of macrophages by activating lysosomal P450 and nuclear hnRNPA2B1.

Macrophages in tumors (tumor-associated macrophages, TAMs), a major population within most tumors, play key homeostatic functions by stimulating angiogenesis, enhancing tumor cell growth, and suppressing antitumor immunity. Resetting TAMs by simple, efficacious and safe approach(s) is highly desirable to enhance antitumor immunity and attenuate tumor cell malignancy. Previously, we used tumor cell-derived microparticles to package chemotherapeutic drugs (drug-MPs), which resulted in a significant treatment outcome in human malignant pleural effusions via neutrophil recruitments, implicating that drug-MPs might reset TAMs, considering the inhibitory effects of M2 macrophages on neutrophil recruitment and activation. Here, we show that drug-MPs can function as an antitumor immunomodulator by resetting TAMs with M1 phenotype and IFN-ß release. Mechanistically, drug molecules in tumor MPs activate macrophage lysosomal P450 monooxygenases, resulting in superoxide anion formation, which further amplifies lysosomal ROS production and pH value by activating lysosomal NOX2. Consequently, lysosomal Ca2+ signaling is activated, thus polarizing macrophages towards M1. Meanwhile, the drug molecules are delivered from lysosomes into the nucleus where they activate DNA sensor hnRNPA2B1 for IFN-ß production. This lysosomal-nuclear machinery fully arouses the antitumor activity of macrophages by targeting both lysosomal pH and the nuclear innate immunity. These findings highlight that drug-MPs can act as a new immunotherapeutic approach by revitalizing antitumor activity of macrophages. This mechanistic elucidation can be translated to treat malignant ascites by drug-MPs combined with PD-1 blockade.