Mol Syst Biol. 2025 Feb 17. doi: 10.1038/s44320-025-00090-9. Online ahead of print.

ABSTRACT

Algorithms that constrain metabolic network models with enzyme levels to predict metabolic activity assume that changes in enzyme levels are indicative of flux variations. However, metabolic flux can also be regulated by other mechanisms such as allostery and mass action. To systematically explore the relationship between fluctuations in enzyme expression and flux, we combine available yeast proteomic and fluxomic data to reveal that flux changes can be best predicted from changes in enzyme levels of pathways, rather than the whole network or only cognate reactions. We implement this principle in an 'enhanced flux potential analysis' (eFPA) algorithm that integrates enzyme expression data with metabolic network architecture to predict relative flux levels of reactions including those regulated by other mechanisms. Applied to human data, eFPA consistently predicts tissue metabolic function using either proteomic or transcriptomic data. Additionally, eFPA efficiently handles data sparsity and noisiness, generating robust flux predictions with single-cell gene expression data. Our approach outperforms alternatives by striking an optimal balance, evaluating enzyme expression at pathway level, rather than either single-reaction or whole-network levels.

PMID:39962320 | DOI:10.1038/s44320-025-00090-9

Nat Genet. 2025 Feb 17. doi: 10.1038/s41588-025-02098-1. Online ahead of print.

NO ABSTRACT

PMID:39962239 | DOI:10.1038/s41588-025-02098-1

Sci Rep. 2025 Feb 17;15(1):5737. doi: 10.1038/s41598-025-88417-2.

ABSTRACT

Triple-negative breast cancer (TNBC) is difficult to treat and has a low five-year survival rate. In South Africa, a large percentage of the population still relies on traditional plant-based medicine. To establish the utility of both methanol and water-soluble extracts from the leaves of Tulbaghia violacea, cytotoxicity assays were carried out to establish the IC50 values against a TNBC cell line. Cell cycle and apoptosis assays were carried out using the extracts. To identify the molecular compounds, present in water-soluble leaf extracts, NMR spectroscopy was performed. Compounds of interest were then used in computational docking studies with the anti-apoptotic protein COX-2. The IC50 values for the water- and methanol-soluble extracts were determined to be 400 and 820 µg/mL, respectively. The water-soluble extract induced apoptosis in the TNBC cell line to a greater extent than in the normal cell line. RNAseq indicated that there was an increase in the transcription of pro-apoptotic genes in the TNBC cell line. The crude extract also caused these cells to stall in the S phase. Of the 61 compounds identified in this extract, five demonstrated a high binding affinity for COX-2. Based on these findings, the compounds within the extract show significant potential for further investigation as candidates for the development of cancer therapeutics, particularly for TNBC.

PMID:39962120 | DOI:10.1038/s41598-025-88417-2

J Nutr Biochem. 2025 Feb 15:109869. doi: 10.1016/j.jnutbio.2025.109869. Online ahead of print.

ABSTRACT

Coffee brew is an integral part of the individual diet worldwide. Roasted coffee contains numerous bioactive substances whose significance for health is investigated in nutritional studies. Food biomarkers are recommended to correlate coffee consumption and health effects in the most unbiased way possible. Metabolites of atractyligenin derivatives from roasted coffee have been suggested as candidate analytes indicating coffee consumption. UHPLC-MS/MS analysis revealed that atractyligenin (1), 2-O-β-D-glucosylatractyligenin and 3'-O-β-D-glucosyl-2'-O-isovaleryl-2-O-β-D-glucosylatractyligenin were extracted into coffee brew. Their concentrations in filtered and unfiltered coffee did not differ significantly, suggesting independence from the preparation method. In a coffee intervention study (n=12, female/male 6/6), atractyligenin metabolites were not detectable in plasma after three days of coffee abstinence. After coffee, atractyligenin (1) and atractyligenin-19-O-D-glucuronide (M1) were the quantitatively dominant atractyligenin metabolites in plasma and showed two peaks each after 0.5 and 10 h, respectively. Half-lives after the first cmax in plasma were ∼0.31 h. 1 and M1 were detectable in plasma, indicating coffee consumption for up to 24 h after one serving. Within 10 h, ∼13.7% of the atractyligenin glycosides supplied by coffee brew were excreted in urine as metabolites 1 and M1. Metabolites 2β-hydroxy-15-oxoatractylan-4α-carboxy-19-O-β-d-glucuronide (M2) and 2β-hydroxy-15-oxoatractylan-4α-carboxylic acid-2-O-β-d-glucuronide (M3) were detected in only some samples and appeared unreliable as indicators for coffee consumption. No concentration differences between female and male study participants were observed in plasma and urine. In conclusion atractyligenin and its 19-O-β-D-glucuronide are promising markers of Arabica coffee consumption in plasma and urine for both men and women, independent of the brewing method.

PMID:39961551 | DOI:10.1016/j.jnutbio.2025.109869

Pharmacoepidemiol Drug Saf. 2025 Feb;34(2):e70100. doi: 10.1002/pds.70100.

ABSTRACT

BACKGROUND: The use of real-world data is increasing to examine immune-related adverse event (irAE) incidence and risk factors in immune checkpoint inhibitor (ICI) users. We aimed to validate five case definition algorithms for irAE in a Johns Hopkins lung cancer registry.

METHODS: We conducted a retrospective cohort study using linked electronic health record (EHR) and cancer registry data from a large academic healthcare system. The Lung Immunotherapy irAE Monitoring Registry assesses irAEs in a group of patients treated for lung cancer at Johns Hopkins Medicine from 2013 to 2020. We used data from inpatient, outpatient, and emergency department encounters, including International Classification of Disease (ICD)-10 codes and medication administration records to classify the presence or absence of irAEs using five distinct algorithms. These algorithms included three that used both diagnosis (Dx) and medication (Rx) codes, one that used Rx codes only, and one that used Dx codes only, ranging from most numerous criteria (most stringent) to least numerous criteria (least stringent). We compared all five algorithms' performances against chart review-ascertained irAE status and reported sensitivity (Se), specificity (Sp), positive predictive value (PPV), negative predictive value (NPV), and C-statistic (C-stat), with 95% confidence intervals (CI). We also explored algorithm performance by specific organ system toxicities and by Common Terminology Criteria for Adverse Events (CTCAE) severity.

RESULTS: The study cohort included 354 patients with ICI exposure for whom chart review-ascertained irAE status was available. A total of 89 (25.1%) experienced at least one irAE (38 pneumonitis, 12 arthritis, 12 colitis, 7 thyroiditis, and others). Across algorithm versions, Se ranged from 59.3% to 93.2% in descending order of algorithm stringency; Sp ranged from 21.0% to 77.6% in ascending order of algorithm stringency, and PPV ranged from 19.1% to 34.7%. The C-stat ranged from 0.57 (95% CI, 0.53-0.61) (Dx codes only) to 0.71 (0.64-0.77) (Rx codes only). For severe irAE (CTCAE Grade 3-5), all algorithms performed better than in the primary analysis, and four exceeded the threshold for usefulness as a measurement tool (maximum C-stat: 0.78 [0.71-0.85] [Rx codes only]). For severe tissue-specific toxicities, algorithmic detection of irAE pneumonitis, colitis, and hepatitis performed better than for the overall group of severe toxicities. Generally, the algorithm versions depicted a Se-Sp tradeoff depending on algorithm stringency.

CONCLUSION: In this validation study of five irAE case definition algorithms, a combination of ICD-10 codes and medication administration codes generally perform well to identify more severe irAE (CTCAE Grade 3-5), and severe pneumonitis, hepatitis, and colitis (common irAEs) among all possible irAE severity levels and sites. Medication codes alone perform well at identifying severe irAE, while the most stringent algorithm (mirroring guideline-recommended irAE treatment) has the highest Sp and PPV. Algorithms have utility for comparing the relative risk of irAE between regimens or patient subgroups.

PMID:39961795 | DOI:10.1002/pds.70100

Elife. 2025 Feb 17;13:RP95952. doi: 10.7554/eLife.95952.

ABSTRACT

Despite exciting developments in cancer immunotherapy, its broad application is limited by the paucity of targetable antigens on the tumor cell surface. As an intrinsic cellular pathway, nonsense-mediated decay (NMD) conceals neoantigens through the destruction of the RNA products from genes harboring truncating mutations. We developed and conducted a high-throughput screen, based on the ratiometric analysis of transcripts, to identify critical mediators of NMD in human cells. This screen implicated disruption of kinase SMG1's phosphorylation of UPF1 as a potential disruptor of NMD. This led us to design a novel SMG1 inhibitor, KVS0001, that elevates the expression of transcripts and proteins resulting from human and murine truncating mutations in vitro and murine cells in vivo. Most importantly, KVS0001 concomitantly increased the presentation of immune-targetable human leukocyte antigens (HLA) class I-associated peptides from NMD-downregulated proteins on the surface of human cancer cells. KVS0001 provides new opportunities for studying NMD and the diseases in which NMD plays a role, including cancer and inherited diseases.

PMID:39960487 | DOI:10.7554/eLife.95952

J Clin Invest. 2025 Feb 17;135(4):e174081. doi: 10.1172/JCI174081.

ABSTRACT

Dilated cardiomyopathy (DCM) due to genetic disorders results in decreased myocardial contractility, leading to high morbidity and mortality rates. There are several therapeutic challenges in treating DCM, including poor understanding of the underlying mechanism of impaired myocardial contractility and the difficulty of developing targeted therapies to reverse mutation-specific pathologies. In this report, we focused on K210del, a DCM-causing mutation, due to 3-nucleotide deletion of sarcomeric troponin T (TnnT), resulting in loss of Lysine210. We resolved the crystal structure of the troponin complex carrying the K210del mutation. K210del induced an allosteric shift in the troponin complex resulting in distortion of activation Ca2+-binding domain of troponin C (TnnC) at S69, resulting in calcium discoordination. Next, we adopted a structure-based drug repurposing approach to identify bisphosphonate risedronate as a potential structural corrector for the mutant troponin complex. Cocrystallization of risedronate with the mutant troponin complex restored the normal configuration of S69 and calcium coordination. Risedronate normalized force generation in K210del patient-induced pluripotent stem cell-derived (iPSC-derived) cardiomyocytes and improved calcium sensitivity in skinned papillary muscles isolated from K210del mice. Systemic administration of risedronate to K210del mice normalized left ventricular ejection fraction. Collectively, these results identify the structural basis for decreased calcium sensitivity in K210del and highlight structural and phenotypic correction as a potential therapeutic strategy in genetic cardiomyopathies.

PMID:39959972 | DOI:10.1172/JCI174081

J Eval Clin Pract. 2025 Feb;31(1):e70010. doi: 10.1111/jep.70010.

ABSTRACT

RATIONALE: Despite growing emphasis among healthcare decision-makers on patient perspectives and real-world outcomes to inform care and access decisions, understanding of patient journey experiences in rare diseases remains limited due to data collection and evaluation challenges.

AIMS AND OBJECTIVES: This systematic literature review (SLR) assessed study designs, methodologies, and outcomes reported in real-world investigations of rare disease patient journeys.

METHODS: Searches in PubMed and Google Scholar targeted English-language publications and congress proceedings from 1 January 2014, to 30 April 2024, including rare disease patients, caregivers, or healthcare providers. Keywords included 'Journey', 'Path', or 'Odyssey'. Two reviewers independently assessed eligibility and abstracted data. Descriptive analyses and quality assessments were conducted.

RESULTS: Thirty-one studies met inclusion criteria, with 296,548 participants spanning over 600 rare diseases. Most studies used prospective observational (61%) and cross-sectional (26%) designs and were conducted in Europe (45%). Interviews (39%) and surveys (29%) were common methodologies. Patients (87%) were the primary research focus, compared to caregivers (32%) or providers (10%). The most studied journey stages were 'Pre-diagnosis/Screening' (97%) and 'Diagnosis' (84%), while 'Disease Awareness' (16%) and 'Treatment Adherence' (6%) were less common. Across 164 outcomes reported, frequent outcomes included 'Healthcare Resource Utilization' (94%), 'Symptoms' (74%), and 'Time-to-Diagnosis' (71%). Fewer studies reported 'Costs' (19%), 'Caregiver/Family Burden' (16%), and 'Productivity' (13%). Time-to-diagnosis averaged 11.8 years and a median of 6.1 years. All but one study (97%) was rated low or very low quality due to observational designs.

CONCLUSION: Most rare disease patient journey evidence focuses on 'Pre-diagnosis/Screening' and 'Diagnosis' stages using qualitative methods and surveys. While symptoms, time-to-diagnosis, and resource utilization were commonly reported, evidence gaps included treatment adherence, caregiver burden and productivity. Longitudinal assessments to collect real-world care and treatment burden outcomes, including caregiver perspectives, can enhance both clinician and policy decision-making for individuals living with rare diseases.

PMID:39960234 | DOI:10.1111/jep.70010

Mo Med. 2025 Jan-Feb;122(1):53-59.

ABSTRACT

As the rate of diagnosis for rare disease increases, so does the need to develop scalable solutions to address patient community needs. Drawing upon our experiences in rare intellectual and developmental disability research, advocacy, and treatment, we present two examples of how collaboration between patient groups, clinicians, and investigators at Washington University in St. Louis have generated invaluable resources to accelerate toward treatments. These successful partnerships serve as models for building research and clinical infrastructure for rare diseases.

PMID:39958601 | PMC:PMC11827657

Pharmacogenet Genomics. 2025 Feb 12. doi: 10.1097/FPC.0000000000000562. Online ahead of print.

ABSTRACT

In Advancing Clinical Therapeutics Globally protocol A5372, a pharmacokinetic study of dolutegravir with 1-month of daily rifapentine/isoniazid, twice-daily dolutegravir offset the induction effects of rifapentine on plasma dolutegravir trough concentrations (Ctrough). Here, we characterize the impact on dolutegravir Ctrough of UGT1A1, AADAC, and NAT2 polymorphisms that affect dolutegravir, rifapentine, and isoniazid, respectively. People with HIV receiving dolutegravir-based antiretroviral therapy with an indication to treat latent tuberculosis underwent pharmacokinetic sampling during dolutegravir 50 mg once daily alone, and on day 28 of dolutegravir 50 mg twice daily with rifapentine/isoniazid. Multivariable linear regression models characterized genetic associations with dolutegravir Ctrough. Among 30 participants evaluable for genetic associations, median (Q1, Q3) day 0 dolutegravir Ctrough was 1745 (1099, 2694) ng/ml, and day 28 was 2146 (1412, 2484) ng/ml. Day 28 Ctrough was higher with UGT1A1 rs887829 TT [geometric mean ratio (GMR) = 1.65; 90% confidence interval (CI): 0.97-2.78] and CT (GMR = 1.38; 90% CI: 1.02-1.86) than with CC, and was higher with AADAC rs1803155 GG (GMR = 1.79; 90% CI: 1.09-2.93) and AG (GMR = 1.48; 90% CI: 1.14-1.90) than with AA. Median day 28 Ctrough ranged from 1205 (1063, 1897) ng/ml with 4 total UGT1A1 and AADAC risk alleles, to 3882 and 3717 ng/ml with only one risk allele. Individuals with concomitant AADAC slow metabolizer and UGT1A1 normal metabolizer genotypes may be at greater risk for clinically significant drug-drug interactions between rifapentine/isoniazid and dolutegravir.

PMID:39960813 | DOI:10.1097/FPC.0000000000000562

Clin Pharmacol Ther. 2025 Feb 17. doi: 10.1002/cpt.3599. Online ahead of print.

ABSTRACT

In the past decade, pharmacogenomic (PGx) testing to predict drug response have emerged into clinical care. Clinical decision support (CDS) has and continues to play a key role in educating prescribers and facilitating the integration of pharmacogenomic results into routine clinical practice. The Epic Genomics module, an add-on to Epic's base clinical software, allows for storage of structured genomic data and provides electronic heath record tools designed with PGx CDS implementation in mind. In early 2022, the University of Florida Health deployed the Genomics module. This tutorial outlines the steps taken by the University of Florida Health Precision Medicine Program to implement Epic's Genomic Module at University of Florida Health and identifies key factors for a successful implementation.

PMID:39960348 | DOI:10.1002/cpt.3599

Elife. 2025 Feb 17;13:RP95952. doi: 10.7554/eLife.95952.

ABSTRACT

Despite exciting developments in cancer immunotherapy, its broad application is limited by the paucity of targetable antigens on the tumor cell surface. As an intrinsic cellular pathway, nonsense-mediated decay (NMD) conceals neoantigens through the destruction of the RNA products from genes harboring truncating mutations. We developed and conducted a high-throughput screen, based on the ratiometric analysis of transcripts, to identify critical mediators of NMD in human cells. This screen implicated disruption of kinase SMG1's phosphorylation of UPF1 as a potential disruptor of NMD. This led us to design a novel SMG1 inhibitor, KVS0001, that elevates the expression of transcripts and proteins resulting from human and murine truncating mutations in vitro and murine cells in vivo. Most importantly, KVS0001 concomitantly increased the presentation of immune-targetable human leukocyte antigens (HLA) class I-associated peptides from NMD-downregulated proteins on the surface of human cancer cells. KVS0001 provides new opportunities for studying NMD and the diseases in which NMD plays a role, including cancer and inherited diseases.

PMID:39960487 | DOI:10.7554/eLife.95952

Pediatr Pulmonol. 2025 Feb;60(2):e70000. doi: 10.1002/ppul.70000.

ABSTRACT

BACKGROUND: The COVID-19 pandemic ushered widespread adoption of telehealth (TH) by cystic fibrosis (CF) centers in the USA. TH was initially described as well-accepted by both clinicians and patients. As we move past the unusual circumstances of the pandemic, the sustainability of TH remains untested. This study sought to test the durability of clinician perceptions of TH post-pandemic.

METHODS: This is a cross-sectional, survey study of clinicians at seven US CF centers. We refined a previously disseminated survey initially designed to assess clinician perceptions of TH in 2020. Survey results were analyzed using descriptive statistics and current responses were compared to prior results.

RESULTS: Clinician perceptions surrounding TH remain high but have changed over time with 75% now endorsing satisfaction (90% in 2020, p = 0.02). The most cited barriers were technology limitations (68%) and limited in-person assessments (66%). We found a significant decrease in concern over missing in-person assessments compared to 2020. Benefits of TH included convenience for patients and families (100%) and reduction in missed days of school or work (100%). In total, 83% of current respondents felt TH should remain part of routine CF care. A majority indicated certain patient characteristics increased their preference to conduct at least one TH visit per year.

CONCLUSIONS: Despite restoration of full access to in-person care, clinicians caring for pwCF continue to use TH across the surveyed CF centers post-pandemic. Respondents continue to view TH favorably. Further study is needed to understand for which patient and clinical scenarios TH is most appropriate.

PMID:39960328 | DOI:10.1002/ppul.70000

Invest Radiol. 2025 Feb 18. doi: 10.1097/RLI.0000000000001162. Online ahead of print.

ABSTRACT

OBJECTIVES: Deep learning for body composition analysis (BCA) is gaining traction in clinical research, offering rapid and automated ways to measure body features like muscle or fat volume. However, most current methods prioritize computed tomography (CT) over magnetic resonance imaging (MRI). This study presents a deep learning approach for automatic BCA using MR T2-weighted sequences.

METHODS: Initial BCA segmentations (10 body regions and 4 body parts) were generated by mapping CT segmentations from body and organ analysis (BOA) model to synthetic MR images created using an in-house trained CycleGAN. In total, 30 synthetic data pairs were used to train an initial nnU-Net V2 in 3D, and this preliminary model was then applied to segment 120 real T2-weighted MRI sequences from 120 patients (46% female) with a median age of 56 (interquartile range, 17.75), generating early segmentation proposals. These proposals were refined by human annotators, and nnU-Net V2 2D and 3D models were trained using 5-fold cross-validation on this optimized dataset of real MR images. Performance was evaluated using Sørensen-Dice, Surface Dice, and Hausdorff Distance metrics including 95% confidence intervals for cross-validation and ensemble models.

RESULTS: The 3D ensemble segmentation model achieved the highest Dice scores for the body region classes: bone 0.926 (95% confidence interval [CI], 0.914-0.937), muscle 0.968 (95% CI, 0.961-0.975), subcutaneous fat 0.98 (95% CI, 0.971-0.986), nervous system 0.973 (95% CI, 0.965-0.98), thoracic cavity 0.978 (95% CI, 0.969-0.984), abdominal cavity 0.989 (95% CI, 0.986-0.991), mediastinum 0.92 (95% CI, 0.901-0.936), pericardium 0.945 (95% CI, 0.924-0.96), brain 0.966 (95% CI, 0.927-0.989), and glands 0.905 (95% CI, 0.886-0.921). Furthermore, body part 2D ensemble model reached the highest Dice scores for all labels: arms 0.952 (95% CI, 0.937-0.965), head + neck 0.965 (95% CI, 0.953-0.976), legs 0.978 (95% CI, 0.968-0.988), and torso 0.99 (95% CI, 0.988-0.991). The overall average Dice across body parts (2D = 0.971, 3D = 0.969, P = ns) and body regions (2D = 0.935, 3D = 0.955, P < 0.001) ensemble models indicates stable performance across all classes.

CONCLUSIONS: The presented approach facilitates efficient and automated extraction of BCA parameters from T2-weighted MRI sequences, providing precise and detailed body composition information across various regions and body parts.

PMID:39961134 | DOI:10.1097/RLI.0000000000001162

Invest Radiol. 2025 Feb 18. doi: 10.1097/RLI.0000000000001166. Online ahead of print.

ABSTRACT

OBJECTIVES: Double-dose contrast-enhanced brain imaging improves tumor delineation and detection of occult metastases but is limited by concerns about gadolinium-based contrast agents' effects on patients and the environment. The purpose of this study was to test the benefit of a deep learning-based contrast signal amplification in true single-dose T1-weighted (T-SD) images creating artificial double-dose (A-DD) images for metastasis detection in brain magnetic resonance imaging.

MATERIALS AND METHODS: In this prospective, multicenter study, a deep learning-based method originally trained on noncontrast, low-dose, and T-SD brain images was applied to T-SD images of 30 participants (mean age ± SD, 58.5 ± 11.8 years; 23 women) acquired externally between November 2022 and June 2023. Four readers with different levels of experience independently reviewed T-SD and A-DD images for metastases with 4 weeks between readings. A reference reader reviewed additionally acquired true double-dose images to determine any metastases present. Performances were compared using Mid-p McNemar tests for sensitivity and Wilcoxon signed rank tests for false-positive findings.

RESULTS: All readers found more metastases using A-DD images. The 2 experienced neuroradiologists achieved the same level of sensitivity using T-SD images (62 of 91 metastases, 68.1%). While the increase in sensitivity using A-DD images was only descriptive for 1 of them (A-DD: 65 of 91 metastases, +3.3%, P = 0.424), the second neuroradiologist benefited significantly with a sensitivity increase of 12.1% (73 of 91 metastases, P = 0.008). The 2 less experienced readers (1 resident and 1 fellow) both found significantly more metastases on A-DD images (resident, T-SD: 61.5%, A-DD: 68.1%, P = 0.039; fellow, T-SD: 58.2%, A-DD: 70.3%, P = 0.008). They were therefore able to use A-DD images to increase their sensitivity to the neuroradiologists' initial level on regular T-SD images. False-positive findings did not differ significantly between sequences. However, readers showed descriptively more false-positive findings on A-DD images. The benefit in sensitivity particularly applied to metastases ≤5 mm (5.7%-17.3% increase in sensitivity).

CONCLUSIONS: A-DD images can improve the detectability of brain metastases without a significant loss of precision and could therefore represent a potentially valuable addition to regular single-dose brain imaging.

PMID:39961132 | DOI:10.1097/RLI.0000000000001166

Medicine (Baltimore). 2025 Feb 14;104(7):e41560. doi: 10.1097/MD.0000000000041560.

ABSTRACT

Acute allograft rejection in patients undergoing renal transplantation is diagnosed through histopathological analysis of renal graft biopsies, which can be used to quantify elementary lesions. However, quantification of elementary lesions requires considerable expertise, time, and effort. Using a 2-stage classification strategy, we sought to examine the effectiveness of deep learning in detecting and classifying glomeruli into 4 groups, namely normal, abnormal, sclerotic, and glomerulitis, as a potential biopsy triage system focused on transplant rejection. We used the U-Net model to build a glomeruli detection model using 137 kidney biopsy slides obtained from 80 kidney transplant patients. The median age of the patients was 52 (19-74) years, with 65% being men and 35% women. MobileNetV2 and VGG16 models were compared using a 2-stage classification strategy. In the first classification step, the models classified glomeruli into sclerotic and nonsclerotic glomeruli. In the second classification step, the nonsclerotic glomeruli from the first step were classified as normal, abnormal, or glomerulitis. The U-Net model achieved satisfactory detection (Dice coefficient = 0.90), and the MobileNetV2 model was the best for the 2 classification steps, with F1 scores of 0.85, 0.91, 0.98, and 0.92 for normal, abnormal, sclerotic, and glomerulitis, respectively. The 2-stage classification strategy identifies sclerotic glomeruli and abnormal glomeruli relative to permeable glomeruli and quantifies glomerulitis with significant accuracy while avoiding bias from abnormal glomeruli that do not have glomerulitis, providing valuable diagnostic information.

PMID:39960931 | DOI:10.1097/MD.0000000000041560

Medicine (Baltimore). 2025 Feb 14;104(7):e41530. doi: 10.1097/MD.0000000000041530.

ABSTRACT

This study reports the first steps toward establishing a computer vision system to help caregivers of bedridden patients detect pressure ulcers (PUs) early. While many previous studies have focused on using convolutional neural networks (CNNs) to elevate stages, hardware constraints have presented challenges related to model training and overreliance on medical opinions. This study aimed to develop a tool to classify PU stages using a Vision Transformer model to process actual PU photos. To do so, we used a retrospective observational design involving the analysis of 395 images of different PU stages that were accurately labeled by nursing specialists and doctors from 3 hospitals. In the pressure ulcer cluster vision transformer (PUC-ViT) model classifies the PU stage with a mean ROC curve value of 0.936, indicating a model accuracy of 97.76% and F1 score of 95.46%. We found that a PUC-ViT model showed higher accuracy than conventional models incorporating CNNs, and both effectively reduced computational complexity and achieved low floating point operations per second. Furthermore, we used internet of things technologies to propose a model that allows anyone to analyze input images even at low computing power. Based on the high accuracy of our proposed model, we confirm that it enables community caregivers to detect PUs early, facilitating medical referral.

PMID:39960905 | DOI:10.1097/MD.0000000000041530

Anal Chem. 2025 Feb 17. doi: 10.1021/acs.analchem.4c06001. Online ahead of print.

ABSTRACT

The integration of paper-based microfluidics with deep learning represents a pivotal trend in enhancing diagnostic capabilities. This paper introduces a new approach to improve the performance of a paper-based microfluidic enzyme-linked immunosorbent assay (ELISA) by training the temporal sequence colorimetric data rather than static data conventionally, using deep learning. Traditional deep learning-assisted ELISA analysis methods usually rely on a single snapshot of the reaction at its end, which limits the further improvement of sensitivity and specificity (or accuracy for combined evaluation), as it misses dynamic changes in the reaction over time. In this work, we developed a temporal sequence-enhanced paper analytical device (TimePAD) that captures continuous video data of the ELISA reaction, which contains the dynamic colorimetric changes. With the YOLOv8 deep learning alogrithm and the Rabbit IgG as the model for ELISA assay, we can use the initial 20 min signal instead of waiting for 30 min for full reaction, achieving a 33% reduction in the turnaround time. Moreover, the overall accuracy at 20 min is 94.1%, which is slightly improvement to the 93.5% using a traditional single snapshot method at 30 min. This method not only accelerates result interpretation but also enhances the overall efficiency of diagnostics, making it particularly valuable for time-sensitive point-of-care testing applications. Lastly, to demonstrate its real-world use, we expanded to the disease biomarker cTnI detection and obtained accuracy of 98.1% within only 10 min, compared to 25 min with 97.8% accuracy in traditional methods.

PMID:39960863 | DOI:10.1021/acs.analchem.4c06001

J Synchrotron Radiat. 2025 Mar 1. doi: 10.1107/S1600577525000359. Online ahead of print.

ABSTRACT

Computed tomography (CT) using synchrotron radiation is a powerful technique that, compared with laboratory CT techniques, boosts high spatial and temporal resolution while also providing access to a range of contrast-formation mechanisms. The acquired projection data are typically processed by a computational pipeline composed of multiple stages. Artifacts introduced during data acquisition can propagate through the pipeline and degrade image quality in the reconstructed images. Recently, deep learning has shown significant promise in enhancing image quality for images representing scientific data. This success has driven increasing adoption of deep learning techniques in CT imaging. Various approaches have been proposed to incorporate deep learning into computational pipelines, but each has limitations in addressing artifacts effectively and efficiently in synchrotron CT, either in properly addressing the specific artifacts or in computational efficiency. Recognizing these challenges, we introduce a novel method that incorporates separate deep learning models at each stage of the tomography pipeline - projection, sinogram and reconstruction - to address specific artifacts locally in a data-driven way. Our approach includes bypass connections that feed both the outputs from previous stages and raw data to subsequent stages, minimizing the risk of error propagation. Extensive evaluations on both simulated and real-world datasets illustrate that our approach effectively reduces artifacts and outperforms comparison methods.

PMID:39960472 | DOI:10.1107/S1600577525000359

J Craniofac Surg. 2025 Feb 17. doi: 10.1097/SCS.0000000000011143. Online ahead of print.

ABSTRACT

Accurate virtual orbital reconstruction is crucial for preoperative planning. Traditional methods, such as the mirroring technique, are unsuitable for orbital defects involving both sides of the midline and are time-consuming and labor-intensive. This study introduces a modified 3D U-Net+++ architecture for orbital defects reconstruction, aiming to enhance precision and automation. The model was trained and tested with 300 synthetic defects from cranial spiral CT scans. The method was validated in 15 clinical cases of orbital fractures and evaluated using quantitative metrics, visual assessments, and a 5-point Likert scale, by 3 surgeons. For synthetic defect reconstruction, the network achieved a 95% Hausdorff distance (HD95) of<2.0 mm, an average symmetric surface distance (ASSD) of ∼0.02 mm, a surface Dice similarity coefficient (Surface DSC)>0.94, a peak signal-to-noise ratio (PSNR)>35 dB, and a structural similarity index (SSIM)>0.98, outperforming the compared state-of-the-art networks. For clinical cases, the average 5-point Likert scale scores for structural integrity, edge consistency, and overall morphology were>4, with no significant difference between unilateral and bilateral/trans-midline defects. For clinical unilateral defect reconstruction, the HD95 was ∼2.5 mm, ASSD<0.02 mm, Surface DSC>0.91, PSNR>30 dB, and SSIM>0.99. The automatic reconstruction process took ∼10 seconds per case. In conclusion, this method offers a precise and highly automated solution for orbital defect reconstruction, particularly for bilateral and trans-midline defects. We anticipate that this method will significantly assist future clinical practice.

PMID:39960444 | DOI:10.1097/SCS.0000000000011143