Internal Dataset Research Articles

GMILT: A Novel Transformer Network That Can Noninvasively Predict EGFR Mutation Status.

Noninvasively and accurately predicting the epidermal growth factor receptor (EGFR) mutation status is a clinically vital problem. Moreover, further identifying the most suspicious area related to the EGFR mutation status can guide the biopsy to avoid false negatives. Deep learning methods based on computed tomography (CT) images may improve the noninvasive prediction of EGFR mutation status and potentially help clinicians guide biopsies by visual methods. Inspired by the potential inherent links between EGFR mutation status and invasiveness information, we hypothesized that the predictive performance of a deep learning network can be improved through extra utilization of the invasiveness information. Here, we created a novel explainable transformer network for EGFR classification named gated multiple instance learning transformer (GMILT) by integrating multi-instance learning and discriminative weakly supervised feature learning. Pathological invasiveness information was first introduced into the multitask model as embeddings. GMILT was trained and validated on a total of 512 patients with adenocarcinoma and tested on three datasets (the internal test dataset, the external test dataset, and The Cancer Imaging Archive (TCIA) public dataset). The performance (area under the curve (AUC) =0.772 on the internal test dataset) of GMILT exceeded that of previously published methods and radiomics-based methods (i.e., random forest and support vector machine) and attained a preferable generalization ability (AUC =0.856 in the TCIA test dataset and AUC =0.756 in the external dataset). A diameter-based subgroup analysis further verified the efficiency of our model (most of the AUCs exceeded 0.772) to noninvasively predict EGFR mutation status from computed tomography (CT) images. In addition, because our method also identified the "core area" of the most suspicious area related to the EGFR mutation status, it has the potential ability to guide biopsies.

Advancing Glaucoma Diagnosis: Employing Confidence-Calibrated Label Smoothing Loss for Model Calibration

ObjectiveThe aim of our research is to enhance the calibration of machine learning models for glaucoma classification through a specialized loss function named Confidence-Calibrated Label Smoothing Loss (CC-LS). This approach is specifically designed to refine model calibration without compromising accuracy by integrating label smoothing and confidence penalty techniques, tailored to the specifics of glaucoma detection. DesignThis study focuses on the development and evaluation of a calibrated deep learning model. ParticipantsThe study employs fundus images from both external datasets-ORIGA (482 normal, 168 glaucoma) and REFUGE (720 normal, 80 glaucoma)-and an extensive internal dataset (4,639 images per category), aiming to bolster the model's generalizability. The model's clinical performance is validated using a comprehensive test set (47,913 normal, 1,629 glaucoma) from the internal dataset. MethodsThe CC-LS loss function seamlessly integrates label smoothing, which tempers extreme predictions to avoid overfitting, with confidence-based penalties. These penalties deter the model from expressing undue confidence in incorrect classifications. Our study aims at training models using the CC-LS and comparing their performance with those trained using conventional loss functions. Main Outcome MeasuresThe model's precision is evaluated using metrics like the Brier score, sensitivity, specificity, and the false positive rate, alongside qualitative heatmap analyses for a holistic accuracy assessment. ResultsPreliminary findings reveal that models employing the CC-LS mechanism exhibit superior calibration metrics, as evidenced by a Brier score of 0.098, along with notable accuracy measures: sensitivity of 81%, specificity of 80%, and weighted accuracy of 80%. Importantly, these enhancements in calibration are achieved without sacrificing classification accuracy. ConclusionsThe Confidence-Calibrated Label Smoothing Loss presents a significant advancement in the pursuit of deploying machine learning models for glaucoma diagnosis. By improving calibration, the CC-LS ensures that clinicians can interpret and trust the predictive probabilities, making AI-driven diagnostic tools more clinically viable. From a clinical standpoint, this heightened trust and interpretability can potentially lead to more timely and appropriate interventions, thereby optimizing patient outcomes and safety.

A Multiscale Connected UNet for the Segmentation of Lung Cancer Cells in Pathology Sections Stained Using Rapid On-Site Cytopathological Evaluation

Lung cancer is an increasingly serious health problem worldwide, and early detection and diagnosis are crucial for successful treatment. With the development of artificial intelligence and the growth of data volume, machine learning techniques can play a significant role in improving the accuracy of early detection in lung cancer.This study proposes a deep learning-based segmentation algorithm for rapid on-site cytopathological evaluation (ROSE) to enhance the diagnostic efficiency of endobronchial ultrasoundguided transbronchial needle aspiration biopsy (EBUS-TBNA) during surgery. By utilizing the CUNet3+ network model, cell clusters, including cancer cell clusters, can be accurately segmented in ROSE-stained pathological sections. The model demonstrated high accuracy, with an F1-score of 0.9604, recall of 0.9609, precision of 0.9654, and accuracy of 0.9834 on the internal testing dataset. It also achieved an AUC of 0.9972 for cancer identification. The proposed algorithm provides time savings for on-site diagnosis, improves EBUS-TBNA efficiency, and outperforms classical segmentation algorithms in accurately identifying lung cancer cell clusters in ROSE-stained images. It effectively reduces over-segmentation, decreases network parameters, and enhances computational efficiency, making it suitable for real-time patient evaluation during surgical procedures.

Identification of HER2 ultra-low based on an artificial intelligence (AI)-powered HER2 subcellular quantification from HER2 immunohistochemistry images.

1115 Background: HER2-targeted antibody-drug conjugates (ADCs) can effectively target tumor cells even in HER2-low breast cancers (1+ and 2+/no amp). We developed an AI-based whole-slide image (WSI) analyzer for immunohistochemistry (IHC)-stained slides, exploring a cutoff differentiating between true HER2 negative versus HER2 ultra-low. Methods: An AI-based IHC WSI analyzer was developed using 6,188 WSIs stained by 19 types of IHC including HER2 and from various cancers, annotated by 153 pathologists. The model predicts cells as tumor cells or non-tumor cells and generates HER2 expression continuous score (HER2ecs, 0-100%), quantifying IHC staining intensity of membrane, nucleus, cytoplasm, and membrane+cytoplasm separately for each cell. A total of 401 HER2 IHC WSIs were acquired from Samsung Medical Center (n=275), Kyung Hee University Hospital (n=78), and Commercial Biobank (n=48), scored as 0 (n=347) and 1+ (n=54) by pathologists. Results: In total, 67,169,114 tumor cells and 119,113,886 non-tumor cells from 401 WSIs were successfully evaluated. Comparing HER2 score 0 versus 1+ by pathologists, mean HER2ecs of nucleus (1.7% vs 4.1%), cytoplasm (6.5% vs 12.9%), and membrane (3.6% vs 8.2%) were comparable. In HER2 score 0 WSIs, membrane HER2ecs of tumor cell was 3.6 ± 7.8% (mean ± standard deviation), while that of non-tumor cell was 0.9 ± 2.9% (p <0.001 vs. tumor cell), suggesting, even in HER2 score 0 cases, tumor cell exhibits HER2 expression beyond non-specific HER2 staining intensity. Based on our internal validation dataset, membrane-specific HER2ecs of 6.0% was the optimal threshold discriminating HER2 no-staining cells versus HER2 incomplete and faint/barely perceptible cells (AI-HER2-weak) per ASCO/CAP guideline. Among HER2 score 0 by pathologists, when applying this cutoff, the proportion of WSIs showing AI-HER2-weak or higher membrane-specific HER2ecs in over 10% of all tumor cells was 23.6% (82/347). Interestingly, that of WSIs among HER2 score 1+ by pathologists was 51.9% (28/54), which is comparable to 52.3%, objective response rate of trastuzumab deruxtecan in HER2-low patients in DESTINY-Breast04. Conclusions: AI-powered subcellular-level analysis enables identification of 23.6% HER2 ultra-low WSIs among HER2 score 0 by pathologist, which is defined by AI-HER2-weak or higher membrane-specific HER2 expression ≥ 10%. Clinical validation of the AI-powered HER2 ultra-low definition is warranted.

Stem cell index-based RiskScore model for predicting prognosis in thyroid cancer and experimental verification

ObjectiveAn mRNA expression-based stemness index (mRNAsi) has been developed to characterize cancer stemness. However, the predictive value of mRNAsi-based signature in therapeutic resistance and immunotherapy in thyroid cancer (THCA) remains unclarified. This study evaluated and validated the role of mRNAsi in drug sensitivity, its relationship between mRNAsi and THCA clinical features and immunity based on bioinformatics. MethodsBased on transcriptome data of THCA patients from the Tumor Genome Atlas Project (TCGA) database, and expression data of multifunctional stem cell samples from the Progenitor Cell Biology Consortium (PCBC) databases, mRNAsi was calculated by the " one class logistic regression (OCLR)" method, Molecular subtypes of TCGA-THCA samples were identified with mRNAsi-related genes using ConsensusClusterPlus method. The gene mutation, clinical characteristics, immune characteristics, TIDE and drug sensitivity were compared among molecular subtypes. A prognostic model was designed with Lasso cox method. Modulation of malignant phenotype of THCA cell lines by model characterization genes is validated by CCK-8, flow cytometry. DNA methylation disorder in promoter region was analyzed between risk groups. The model was validated for survival in the internal Test dataset, while TCGA pan-cancer and immunotherapy datasets were further employed to validate the performance of this model. ResultsWe obtained a total of 78 stem cell samples, each containing the expression profile of 8087 mRNA genes. Based on mRNAsi, THCA was divided into 3 subtypes. Subtype C2 had the poorest prognosis and highest immune score, while subtype C3 had the best prognosis, lowest mRANsi and highest TIDE score. Patients in subtype C2 showed higher sensitivity to Cisplatin, Erlotinib, Paclitaxel, and Lapatinib. The prognostic signature was generated using 5 mRNAsi-related genes, which could predict prognosis for THCA. qRT-PCR results showed that the expression of 5 genes were various in Hth7 and KTC-1 cells, and inhibition CELSR3 expression increased percentage of apoptosis in Hth7 and KTC-1 cells. mRNAsi related DNA methylation sites were mainly enriched in tumor related pathways. Good performance of this model was validated in Test dataset, pan-cancer and immunotherapy datasets. ConclusionThis study identified three subtypes for classification and developed a prognostic model with mRNAsi-related genes, which provided great potential for prognosis and immunotherapy prediction.

Discovery of a T cell proliferation-associated regulator signature correlates with prognosis risk and immunotherapy response in bladder cancer.

The efficacy of immunotherapy is heavily influenced by T cell activity. This study aimed to examine how T cell proliferation regulators can predict the prognosis and response to immunotherapy in patients with bladder cancer (BCa). T cell proliferation-related subtypes were determined by employing the non-negative matrix factorization (NMF) algorithm that analyzed the expression patterns of T cell proliferation regulators. Subtypes were assessed for variations in prognosis, immune infiltration, and functional behaviors. Subsequently, a risk model related to T cell proliferation was created through Cox and Lasso regression analyses in the TCGA cohort and then confirmed in two GEO cohorts and an immunotherapy cohort. BCa patients were categorized into two subtypes (C1 and C2) according to the expression profiles of 31T cell proliferation-related genes (TRGs) with distinct prognoses and immune landscapes. The C2 subtype had a shorter overall survival (OS), with higher levels of M2 macrophage infiltration, and the activation of cancer-related pathways than the C1 subtype. Following this, thirteen prognosis-related genes that were involved in T cell proliferation were utilized to create the prognostic signature. The model's predictive accuracy was confirmed by analyzing both internal and external datasets. Individuals in the high-risk category experienced a poorer prognosis, increased immunosuppressive factors in the tumor microenvironment, and diminished responses to immunotherapy. Additionally, the immunotherapeutic prediction efficacy of the model was further confirmed by an immunotherapy cohort (anti-PD-L1 in the IMvigor210 cohort). Our study characterized two subtypes linked to T cell proliferation in BCa patients with distinct prognoses and tumor microenvironment (TME) patterns, providing new insights into the heterogeneity of T cell proliferation in BCa and its connection to the immune landscape. The signature has prospective clinical implications for predicting outcomes and may help physicians to select prospective responders who prioritize current immunotherapy.

Population Pharmacokinetics of Dasatinib in Healthy Subjects.

Dasatinib is one of the tyrosine kinase inhibitors. The main use of these agents is inhibition of cancerous cell proliferation. The therapeutic importance of tyrosine kinase inhibitors raises the necessity of many types of investigations, especially the pharmacokinetic analysis of these drugs in humans. This analysis, along with other investigations and clinical research, will contribute to the overall knowledge of the drug. This study focused on the population pharmacokinetics of dasatinib. The objective of the study was to investigate the sources of the variability of dasatinib in a population pharmacokinetics study in healthy participants. We utilized 4180 plasma observations from 110 subjects who were administered SPRYCEL® on two separate occasions under fasting conditions; data from 20% of the subjects (22 subjects) were extracted for the purpose of internal model evaluation and data from 88 subjects were used in modeling. The model was evaluated by visual predictive check of three different datasets. A two-compartmental model with first order absorption and transit compartment was considered the simplest base model to describe the data based on the corrected Bayesian information criterion evaluation. Covariates were tested through conditional sampling for the stepwise approach-screening procedure in Monolix 2020R1 version. Conditional sampling for the stepwise approach was used to include the correlated covariates within the base model in the forward inclusion step and then to eliminate them backwardly to ensure that the key covariates were kept in the model at the final stage. The effect of body mass index on the absorption rate constant was considered as significant covariate in the final established model. Visual predictive check for simulations, 20% of the original dataset (internal dataset) and an external dataset demonstrated the appropriateness of the final model. Population pharmacokinetic modeling was performed to describe dasatinib pharmacokinetics in healthy subjects. Body mass index was considered as a factor that might be used in the future along with studies on patients to adjust the dosing regimens. Dasatinib is classified as a highly variable drug; this variability was demonstrated in the study by the effect of body mass index on the absorption rate constant.

Development of an empirical model for the prediction of the sound absorption coefficient for thin and low-density fibrous materials

Currently, FEA software such as ABAQUS uses empirical models to predict the sound absorption coefficient of poroelastic materials. However, based on a recent review of the literature it was found that the current sound absorption empirical models are inadequate for accurate prediction of thin (t < 20 mm), low-density materials (ρB < 50 kg/m3). Therefore, the predictions of the sound pressure levels in vehicle cabins, using such software, will be inaccurate since the trim materials are thin and have a low density. Thus, this research aimed to develop an empirical model that can accurately predict the sound absorption coefficient of these materials. Hence, polypropylene fibres consisting of four different diameters were manufactured and converted into nonwovens. Thereafter, airflow resistivity and impedance tube experimental testing were performed on the specimens. Subsequently, statistical analysis of the data was performed using SAS software. SAS was used to identify which independent variables should be included in the models to be developed. The empirical models were developed using the regression analysis toolbox in Microsoft Excel. Once the models were developed, various checks were performed to validate the assumptions of linear regression. The software NumXL was used to perform Cook’s distance tests. Thereafter, the models were validated against the validation dataset, where it was found that the developed exponential model performed best. Finally, the exponential model was compared to existing models using two data sets i.e. an internal dataset, and an external dataset derived from the literature. The developed model outperformed all the historic models on both datasets.

#1017 External validation of a machine learning model for the prediction of bioimpedance-determined pre-dialysis overhydration

Abstract Background and Aims An accurate, automated determination of volume status in dialysis patients using a reproducible standard would be clinically useful. We have previously developed a machine learning model for the prediction of pre-dialysis overhydration (OH) index (litres) from the Fresenius Body Composition Monitor (BCM). External validation is a requisite step before the clinical implementation of a machine learning model. Method Data analyses were performed using Python. An XGBoost regressor model was internally developed and validated in a Fresenius dataset of 9,452 patients and 279,324 BCM target events from the Czech Republic and Italy between January 2016 and December 2022. An XGBoost regressor model with 60 input variables including the average BCM OH value in the preceding year, 2-week moving averages of dialysis session data and biomarkers/ medications data from the preceding 13 weeks demonstrated the best performance of the models tested. The external validation (EV) dataset consisted of 24 patients from a 7-week observational study in a single centre, in Beaumont, Ireland comprising 94 BCM sessions and 650 haemodialysis sessions. Individualised predictions were compared with BCM measures and clinical volume assessments performed at 2-week intervals. Mean squared error (MSE), Root mean square error (RMSE), and Bland-Altman plots were used for continuous outcomes. Precision, recall and F1 score were used to assess the algorithm's performance in predicting the following fluid categories: overhydration ≥1.1L, normohydration −1.1L to 1.1L, underhydration ≤-1.1L. Normohydration weight was calculated by subtracting the predicted OH index from the pre-dialysis weight. Results There was performance degradation between the internal validation (IV) and EV set. The RMSE for pre-dialysis OH was 1.04 kg in the IV set and 1.5 kg in the EV set. This could be attributed to differences in the distribution of the outcome variable (BCM OH validation dataset=1.97 ± 1.56 kg*, EV dataset=0.89 ± 1.7 kg), and the case-mix. The RMSE for normohydration weight was 1.86 kg in the EV set. Wide limits of agreement were noted on a Bland Altman plot comparing the average of BCM and predicted values plotted against the difference in values (mean bias - 0.54, 95% limits of agreement -3.44 kg to 2.36 kg). The overall accuracy of the model in predicting fluid classes was 60%. A histogram showed a close alignment in the distributions of predicted normohydration weights and BCM values. Conclusion The performance of the algorithms degraded between the internal validation and external validation datasets. However, the pre-dialysis algorithm showed an ability to discriminate fluid categories compared to nursing staff and the overall accuracy was acceptable. Additional training of the algorithms using fluid status classes would be expected to improve the precision of predictions. *Values less than the 1st percentile of – 2.17 L were excluded.

A multimodal integration pipeline for accurate diagnosis, pathogen identification, and prognosis prediction of pulmonary infections

Pulmonary infections pose formidable challenges in clinical settings with high mortality rates across all age groups worldwide. Accurate diagnosis and early intervention are crucial to improve patient outcomes. Artificial intelligence (AI) has the capability to mine imaging features specific to different pathogens and fuse multimodal features to reach a synergistic diagnosis, enabling more precise investigation and individualized clinical management. In this study, we successfully developed a multimodal integration (MMI) pipeline to differentiate among bacterial, fungal, and viral pneumonia and pulmonary tuberculosis based on a real-world dataset of 24,107 patients. The area under the curve (AUC) of the MMI system comprising clinical text and computed tomography (CT) image scans yielded 0.910 (95% confidence interval [CI]: 0.904-0.916) and 0.887 (95% CI: 0.867-0.909) in the internal and external testing datasets respectively, which were comparable to those of experienced physicians. Furthermore, the MMI system was utilized to rapidly differentiate between viral subtypes with a mean AUC of 0.822 (95% CI: 0.805-0.837) and bacterial subtypes with a mean AUC of 0.803 (95% CI: 0.775-0.830). Here, the MMI system harbors the potential to guide tailored medication recommendations, thus mitigating the risk of antibiotic misuse. Additionally, the integration of multimodal factors in the AI-driven system also provided an evident advantage in predicting risks of developing critical illness, contributing to more informed clinical decision-making. To revolutionize medical care, embracing multimodal AI tools in pulmonary infections will pave the way to further facilitate early intervention and precise management in the foreseeable future.

A Deep Learning Model for Automatically Quantifying the Anterior Segment in Ultrasound Biomicroscopy Images of Implantable Collamer Lens Candidates

ObjectiveThis study aimed to develop and evaluate a deep learning-based model that could automatically measure anterior segment (AS) parameters on preoperative ultrasound biomicroscopy (UBM) images of implantable Collamer lens (ICL) surgery candidates. MethodsA total of 1164 panoramic UBM images were preoperatively obtained from 321 patients who received ICL surgery in the Eye Center of Renmin Hospital of Wuhan University (Wuhan, China) to develop an imaging database. First, the UNet++ network was utilized to segment AS tissues automatically, such as corneal lens and iris. In addition, image processing techniques and geometric localization algorithms were developed to automatically identify the anatomical landmarks (ALs) of pupil diameter (PD), anterior chamber depth (ACD), angle-to-angle distance (ATA), and sulcus-to-sulcus distance (STS). Based on the results of the latter two processes, PD, ACD, ATA, and STS can be measured. Meanwhile, an external dataset of 294 images from Huangshi Aier Eye Hospital was employed to further assess the model's performance in other center. Lastly, a subset of 100 random images from the external test set was chosen to compare the performance of the model with senior experts. ResultsWhether in the internal test dataset or external test dataset, using manual labeling as the reference standard, the models achieved a mean Dice coefficient exceeding 0.880. Additionally, the intra-class correlation coefficients (ICCs) of ALs’ coordinates were all greater than 0.947, and the percentage of Euclidean distance distribution of ALs within 250 μm was over 95.24%.While the ICCs for PD, ACD, ATA, and STS were greater than 0.957, furthermore, the average relative error (ARE) of PD, ACD, ATA, and STS were below 2.41%. In terms of human versus machine performance, the ICCs between the measurements performed by the model and those by senior experts were all greater than 0.931. ConclusionA deep learning-based model could measure AS parameters using UBM images of ICL candidates, and exhibited a performance similar to that of a senior ophthalmologist.

Two-Stage Training Framework Using Multicontrast MRI Radiomics for IDH Mutation Status Prediction in Glioma.

Purpose To develop a radiomics framework for preoperative MRI-based prediction of isocitrate dehydrogenase (IDH) mutation status, a crucial glioma prognostic indicator. Materials and Methods Radiomics features (shape, first-order statistics, and texture) were extracted from the whole tumor or the combination of nonenhancing, necrosis, and edema regions. Segmentation masks were obtained via the federated tumor segmentation tool or the original data source. Boruta, a wrapper-based feature selection algorithm, identified relevant features. Addressing the imbalance between mutated and wild-type cases, multiple prediction models were trained on balanced data subsets using random forest or XGBoost and assembled to build the final classifier. The framework was evaluated using retrospective MRI scans from three public datasets (The Cancer Imaging Archive [TCIA, 227 patients], the University of California San Francisco Preoperative Diffuse Glioma MRI dataset [UCSF, 495 patients], and the Erasmus Glioma Database [EGD, 456 patients]) and internal datasets collected from the University of Texas Southwestern Medical Center (UTSW, 356 patients), New York University (NYU, 136 patients), and University of Wisconsin-Madison (UWM, 174 patients). TCIA and UTSW served as separate training sets, while the remaining data constituted the test set (1617 or 1488 testing cases, respectively). Results The best performing models trained on the TCIA dataset achieved area under the receiver operating characteristic curve (AUC) values of 0.89 for UTSW, 0.86 for NYU, 0.93 for UWM, 0.94 for UCSF, and 0.88 for EGD test sets. The best performing models trained on the UTSW dataset achieved slightly higher AUCs: 0.92 for TCIA, 0.88 for NYU, 0.96 for UWM, 0.93 for UCSF, and 0.90 for EGD. Conclusion This MRI radiomics-based framework shows promise for accurate preoperative prediction of IDH mutation status in patients with glioma. Keywords: Glioma, Isocitrate Dehydrogenase Mutation, IDH Mutation, Radiomics, MRI Supplemental material is available for this article. Published under a CC BY 4.0 license. See also commentary by Moassefi and Erickson in this issue.

Confidence-aware multi-modality learning for eye disease screening

Multi-modal ophthalmic image classification plays a key role in diagnosing eye diseases, as it integrates information from different sources to complement their respective performances. However, recent improvements have mainly focused on accuracy, often neglecting the importance of confidence and robustness in predictions for diverse modalities. In this study, we propose a novel multi-modality evidential fusion pipeline for eye disease screening. It provides a measure of confidence for each modality and elegantly integrates the multi-modality information using a multi-distribution fusion perspective. Specifically, our method first utilizes normal inverse gamma prior distributions over pre-trained models to learn both aleatoric and epistemic uncertainty for uni-modality. Then, the normal inverse gamma distribution is analyzed as the Student’s t distribution. Furthermore, within a confidence-aware fusion framework, we propose a mixture of Student’s t distributions to effectively integrate different modalities, imparting the model with heavy-tailed properties and enhancing its robustness and reliability. More importantly, the confidence-aware multi-modality ranking regularization term induces the model to more reasonably rank the noisy single-modal and fused-modal confidence, leading to improved reliability and accuracy. Experimental results on both public and internal datasets demonstrate that our model excels in robustness, particularly in challenging scenarios involving Gaussian noise and modality missing conditions. Moreover, our model exhibits strong generalization capabilities to out-of-distribution data, underscoring its potential as a promising solution for multimodal eye disease screening.

Assisting the analysis of insertions and deletions using regional allele frequencies.

Accurate estimation of population allele frequency (AF) is crucial for gene discovery and genetic diagnostics. However, determining AF for frameshift-inducing small insertions and deletions (indels) faces challenges due to discrepancies in mapping and variant calling methods. Here, we propose an innovative approach to assess indel AF. We developed CRAFTS-indels (Calculating Regional Allele Frequency Targeting Small indels), an algorithm that combines AF of distinct indels within a given region and provides "regional AF" (rAF). We tested and validated CRAFTS-indels using three independent datasets: gnomAD v2 (n=125,748 samples), an internal dataset (IGM; n=39,367), and the UK BioBank (UKBB; n=469,835). By comparing rAF against standard AF, we identified rare indels with rAF exceeding standard AF (sAF≤10-4 and rAF>10-4) as "rAF-hi" indels. Notably, a high percentage of rare indels were "rAF-hi", with a higher proportion in gnomAD v2 (11-20%) and IGM (11-22%) compared to the UKBB (5-9% depending on the CRAFTS-indels' parameters). Analysis of the overlap of regions based on their rAF with low complexity regions and with ClinVar classification supported the pertinence of rAF. Using the internal dataset, we illustrated the utility of CRAFTS-indel in the analysis of de novo variants and the potential negative impact of rAF-hi indels in gene discovery. In summary, annotation of indels with cohort specific rAF can be used to handle some of the limitations of current annotation pipelines and facilitate detection of novel gene disease associations. CRAFTS-indels offers a user-friendly approach to providing rAF annotation. It can be integrated into public databases such as gnomAD, UKBB and used by ClinVar to revise indel classifications.

SVPath: A Deep Learning Tool for Analysis of Stria Vascularis from Histology Slides

IntroductionThe stria vascularis (SV) may have a significant role in various otologic pathologies. Currently, researchers manually segment and analyze the stria vascularis to measure structural atrophy. Our group developed a tool, SVPath, that uses deep learning to extract and analyze the stria vascularis and its associated capillary bed from whole temporal bone histopathology slides (TBS).MethodsThis study used an internal dataset of 203 digitized hematoxylin and eosin-stained sections from a normal macaque ear and a separate external validation set of 10 sections from another normal macaque ear. SVPath employed deep learning methods YOLOv8 and nnUnet to detect and segment the SV features from TBS, respectively. The results from this process were analyzed with the SV Analysis Tool (SVAT) to measure SV capillaries and features related to SV morphology, including width, area, and cell count. Once the model was developed, both YOLOv8 and nnUnet were validated on external and internal datasets.ResultsYOLOv8 implementation achieved over 90% accuracy for cochlea and SV detection. nnUnet SV segmentation achieved a DICE score of 0.84–0.95; the capillary bed DICE score was 0.75–0.88. SVAT was applied to compare both the ears used in the study. There was no statistical difference in SV width, SV area, and average area of capillary between the two ears. There was a statistical difference between the two ears for the cell count per SV.ConclusionThe proposed method accurately and efficiently analyzes the SV from temporal histopathology bone slides, creating a platform for researchers to understand the function of the SV further.

Prediction of DNA methylation-based tumor types from histopathology in central nervous system tumors with deep learning.

Precision in the diagnosis of diverse central nervous system (CNS) tumor types is crucial for optimal treatment. DNA methylation profiles, which capture the methylation status of thousands of individual CpG sites, are state-of-the-art data-driven means to enhance diagnostic accuracy but are also time consuming and not widely available. Here, to address these limitations, we developed Deep lEarning from histoPathoLOgy and methYlation (DEPLOY), a deep learning model that classifies CNS tumors to ten major categories from histopathology. DEPLOY integrates three distinct components: the first classifies CNS tumors directly from slide images ('direct model'), the second initially generates predictions for DNA methylation beta values, which are subsequently used for tumor classification ('indirect model'), and the third classifies tumor types directly from routinely available patient demographics. First, we find that DEPLOY accurately predicts beta values from histopathology images. Second, using a ten-class model trained on an internal dataset of 1,796 patients, we predict the tumor categories in three independent external test datasets including 2,156 patients, achieving an overall accuracy of 95% and balanced accuracy of 91% on samples that are predicted with high confidence. These results showcase the potential future use of DEPLOY to assist pathologists in diagnosing CNS tumors within a clinically relevant short time frame.

Automated Analysis of Split Kidney Function from CT Scans Using Deep Learning and Delta Radiomics.

Background: Differential kidney function assessment is an important part of preoperative evaluation of various urological interventions. It is obtained through dedicated nuclear medical imaging and is not yet implemented through conventional Imaging. Objective: We assess if differential kidney function can be obtained through evaluation of contrast-enhanced computed tomography(CT) using a combination of deep learning and (2D and 3D) radiomic features. Methods: All patients who underwent kidney nuclear scanning at Mayo Clinic sites between 2018-2022 were collected. CT scans of the kidneys were obtained within a 3-month interval before or after the nuclear scans were extracted. Patients who underwent a urological or radiological intervention within this time frame were excluded. A segmentation model was used to segment both kidneys. 2D and 3D radiomics features were extracted and compared between the two kidneys to compute delta radiomics and assess its ability to predict differential kidney function. Performance was reported using receiver operating characteristics, sensitivity, and specificity. Results: Studies from Arizona & Rochester formed our internal dataset (n = 1,159). Studies from Florida were separately processed as an external test set to validate generalizability. We obtained 323 studies from our internal sites and 39 studies from external sites. The best results were obtained by a random forest model trained on 3D delta radiomics features. This model achieved an area under curve (AUC) of 0.85 and 0.81 on internal and external test sets, while specificity and sensitivity were 0.84,0.68 on the internal set, 0.70, and 0.65 on the external set. Conclusion: This proposed automated pipeline can derive important differential kidney function information from contrast-enhanced CT and reduce the need for dedicated nuclear scans for early-stage differential kidney functional assessment. Clinical Impact: We establish a machine learning methodology for assessing differential kidney function from routine CT without the need for expensive and radioactive nuclear medicine scans.

Signal enhancement ratio of multi-phase contrast-enhanced MRI: an imaging biomarker for survival in pancreatic adenocarcinoma

ObjectivesTo evaluate signal enhancement ratio (SER) for tissue characterization and prognosis stratification in pancreatic adenocarcinoma (PDAC), with quantitative histopathological analysis (QHA) as the reference standard.MethodsThis retrospective study included 277 PDAC patients who underwent multi-phase contrast-enhanced (CE) MRI and whole-slide imaging (WSI) from three centers (2015–2021). SER is defined as (SIlt − SIpre)/(SIea − SIpre), where SIpre, SIea, and SIlt represent the signal intensity of the tumor in pre-contrast, early-, and late post-contrast images, respectively. Deep-learning algorithms were implemented to quantify the stroma, epithelium, and lumen of PDAC on WSIs. Correlation, regression, and Bland-Altman analyses were utilized to investigate the associations between SER and QHA. The prognostic significance of SER on overall survival (OS) was evaluated using Cox regression analysis and Kaplan–Meier curves.ResultsThe internal dataset comprised 159 patients, which was further divided into training, validation, and internal test datasets (n = 60, 41, and 58, respectively). Sixty-five and 53 patients were included in two external test datasets. Excluding lumen, SER demonstrated significant correlations with stroma (r = 0.29–0.74, all p < 0.001) and epithelium (r = −0.23 to −0.71, all p < 0.001) across a wide post-injection time window (range, 25–300 s). Bland-Altman analysis revealed a small bias between SER and QHA for quantifying stroma/epithelium in individual training, validation (all within ± 2%), and three test datasets (all within ± 4%). Moreover, SER-predicted low stromal proportion was independently associated with worse OS (HR = 1.84 (1.17–2.91), p = 0.009) in training and validation datasets, which remained significant across three combined test datasets (HR = 1.73 (1.25–2.41), p = 0.001).ConclusionSER of multi-phase CE-MRI allows for tissue characterization and prognosis stratification in PDAC.Clinical relevance statementThe signal enhancement ratio of multi-phase CE-MRI can serve as a novel imaging biomarker for characterizing tissue composition and holds the potential for improving patient stratification and therapy in PDAC.Key PointsImaging biomarkers are needed to better characterize tumor tissue in pancreatic adenocarcinoma.Signal enhancement ratio (SER)-predicted stromal/epithelial proportion showed good agreement with histopathology measurements across three distinct centers.Signal enhancement ratio (SER)-predicted stromal proportion was demonstrated to be an independent prognostic factor for OS in PDAC.Graphical

Loss modeling with the size-biased lognormal mixture and the entropy regularized EM algorithm

The Erlang mixture with a common scale parameter is one of many popular models for modeling insurance losses. However, the actuarial literature recognizes and discusses some limitations of aforementioned model in approximate heavy-tailed distributions. In this paper, a size-biased left-truncated Lognormal (SB-ltLN) mixture is proposed as a robust alternative to the Erlang mixture for modeling left-truncated insurance losses with a heavy tail. The weak denseness property of the weighted Lognormal mixture is studied along with the tail behavior. Explicit analytical solutions are derived for moments and Tail Value at Risk based on the proposed model. An extension of the regularized expectation–maximization (REM) algorithm with Shannon's entropy weights (ewREM) is introduced for parameter estimation and variability assessment. The Operational Riskdata eXchange's left-truncated internal fraud loss data set is used to illustrate applications of the proposed model. Finally, the results of a simulation study show promising performance of the proposed SB-ltLN mixture in different simulation settings.

Improving Targeted Mass Spectrometry Data Analysis with Nested Active Machine Learning

Targeted mass spectrometry (MS) holds promise for precise protein and protein‐representative peptide identification and quantification, enhancing disease diagnosis. However, its clinical application is hindered by complex data analysis and expert review requirements. It is hypothesized that machine learning (ML) models can automate data analysis to accelerate the clinical application of MS. The approach involves an ML‐driven pipeline that extracts statistical and morphological features from an MS target region and feeds these features into ML algorithms to generate and assess predictive models. The findings demonstrate ML prediction models exhibit superior performance when trained on extracted features versus raw spectra intensity data and that random forest models exhibit robust classification performance in both internal and external validation datasets. These models remain effective across varying training dataset sizes and positive sample rates and are enhanced by a nested active learning approach. This approach can thus revolutionize clinical MS applications.