arXiv subjects

Kyle J. Lafata

Publications and source records attributed to Kyle J. Lafata.

10 recordsLinked to original sources

arXiv2026

Lymphocyte Mimicry Correction via Region-Level Tissue Reasoning and Unbalanced Optimal Transport

Cell mimicry arises when different cell types appear morphologically similar. Human pathologists resolve this ambiguity using surrounding tissue context, whereas current vision models either lack contextual reasoning (cell foundation models) or cannot operate at the cell level (pathology MLLMs). We present Loki-OT, which propagates region-level tissue reasoning to individual cell predictions via Unbalanced Optimal Transport, using MLLM-derived density priors as soft guidance for ambiguous cell reassignment. Loki-OT is motivated by the observation that pretrained cell foundation model features already encode discriminative information, including tissue context, but standard cell-level supervision fails to use tissue context effectively. The resulting transport plan is distilled into a lightweight student MLP classifier that learns context-aware decision boundaries within the pretrained feature space. On the independent TCGA-BRCA cohort, Loki-OT achieved lower patient-level MAE than the fully supervised in-domain PanopTILs classifier and improved F1 in epithelium-rich mimicry tissues, using 278 weak region-level MLLM estimates built on a general-domain cell foundation model. Code: https://github.com/xiangli980/Lymphocyte_Mimicry_Correction_via_Loki_OT

cs.CV↗

arXiv2026

Zero-Shot Multi-Disease Labeling of Chest, Abdomen, and Pelvis CT Reports Using Open-Weight Large Language Models: The Effect of Labeling Conventions

Purpose: To compare five lightweight open-weight large language models (LLMs) with a rule-based algorithm (RBA) and fine-tuned RadBERT for zero-shot labeling of chest-abdomen-pelvis (CAP) CT reports, and to examine how labeling conventions affect measured performance. Materials and Methods: In this retrospective study, 40,833 CAP CT reports from 29,540 patients examined between 2012 and 2017 were analyzed; age and sex were unavailable. Five LLMs were prompted zero-shot to assign 15 labels across three organ systems and compared with an RBA and fine-tuned RadBERT. Inter-model agreement was assessed with Cohen kappa ($κ$) on 12,197 held-out reports. Macro-averaged F1 was computed against 1,789 radiologist-supervised annotations, the same annotations simplified to disregard clinical actionability, and the CT-RATE dataset. Nonoverlapping bootstrapped 95% CIs indicated relevant differences. Results: MedGemma 27B and MedGemma-1.5 4B showed the highest median agreement ($κ$ = 0.90). Against manual annotations, Gemma-3 27B achieved the highest macro-averaged F1 (0.82 [95% CI: 0.80, 0.83]) versus 0.66 for RadBERT and 0.64 for the RBA; a majority-vote ensemble scored 0.84. Scores were lowest averaged across models for subjective classes, kidney lesion (0.44) and atelectasis (0.67). Relabeling raised F1 for all models on kidney lesion, but for atelectasis only for the LLMs; the RBA and RadBERT declined. F1 against CT-RATE exceeded that against manual annotations for all models, reflecting its more literal convention. Conclusion: Lightweight open-weight LLMs outperformed rule-based and fine-tuned BERT labeling of CAP CT reports with zero-shot prompting. Models and annotators disagreed largely because they applied different labeling criteria.

cs.CL↗

arXiv2026

Reproducible Benchmarking for Lung Nodule Detection and Malignancy Classification Across Multiple Low-Dose CT Datasets

Evaluation of artificial intelligence (AI) models for low-dose CT lung cancer screening is limited by heterogeneous datasets, annotation standards, and evaluation protocols, making performance difficult to compare and translate across clinical settings. We establish a public, reproducible multi-dataset benchmark for lung nodule detection and nodule-level cancer classification and quantify cross-dataset generalizability. Using the Duke Lung Cancer Screening (DLCS) dataset as a clinically curated development set, we evaluate performance across LUNA16/LIDC-IDRI, NLST-3D, and LUNA25. Detection models trained on DLCS and LUNA16 were evaluated externally on NLST-3D using free-response ROC analysis. For malignancy classification, we compared five strategies: randomly initialized ResNet50, Models Genesis, Med3D, a Foundation Model for Cancer Biomarkers, and a Strategic Warm-Start (ResNet50-SWS) approach pretrained using detection-derived candidate patches stratified by confidence. Performance was summarized using AUC with 95% confidence intervals and DeLong tests. Detection performance varied substantially by training dataset, with DLCS-trained models outperforming LUNA16-trained models on external NLST-3D evaluation (sensitivity at 2 false positives per scan: 0.72 vs. 0.64; p < 0.001). For malignancy classification, ResNet50-SWS achieved AUCs of 0.71 (DLCS), 0.90 (LUNA16), 0.81 (NLST-3D), and 0.80 (LUNA25), consistently matching or exceeding alternative pretraining strategies. These results demonstrate that dataset characteristics strongly influence lung cancer AI performance and highlight the need for transparent, multi-dataset benchmarking.

cs.CV↗

arXiv2025

Large Intestine 3D Shape Refinement Using Point Diffusion Models for Digital Phantom Generation

Accurate 3D modeling of human organs is critical for constructing digital phantoms in virtual imaging trials. However, organs such as the large intestine remain particularly challenging due to their complex geometry and shape variability. We propose CLAP, a novel Conditional LAtent Point-diffusion model that combines geometric deep learning with denoising diffusion models to enhance 3D representations of the large intestine. Given point clouds sampled from segmentation masks, we employ a hierarchical variational autoencoder to learn both global and local latent shape representations. Two conditional diffusion models operate within this latent space to refine the organ shape. A pretrained surface reconstruction model is then used to convert the refined point clouds into meshes. CLAP achieves substantial improvements in shape modeling accuracy, reducing Chamfer distance by 26% and Hausdorff distance by 36% relative to the initial suboptimal shapes. This approach offers a robust and extensible solution for high-fidelity organ modeling, with potential applicability to a wide range of anatomical structures.

cs.CV↗

arXiv2025

Demographic Distribution Matching between real world and virtual phantom population

Virtual imaging trials (VITs) offer scalable and cost-effective tools for evaluating imaging systems and protocols. However, their translational impact depends on rigorous comparability between virtual and real-world populations. This study introduces DISTINCT (Distributional Subsampling for Covariate-Targeted Alignment), a statistical framework for selecting demographically aligned subsamples from large clinical datasets to support robust comparisons with virtual cohorts. We applied DISTINCT to the National Lung Screening Trial (NLST) and a companion virtual trial dataset (VLST). The algorithm jointly aligned typical continuous (age, BMI) and categorical (sex, race, ethnicity) variables by constructing multidimensional bins based on discretized covariates. For a given target size, DISTINCT samples individuals to match the joint demographic distribution of the reference population. We evaluated the demographic similarity between VLST and progressively larger NLST subsamples using Wasserstein and Kolmogorov-Smirnov (K-S) distances to identify the maximal subsample size with acceptable alignment. The algorithm identified a maximal aligned NLST subsample of 9,974 participants, preserving demographic similarity to the VLST population. Receiver operating characteristic (ROC) analysis using risk scores for lung cancer detection showed that area under the curve (AUC) estimates stabilized beyond 6,000 participants, confirming the sufficiency of aligned subsamples for virtual imaging trial evaluation. Stratified AUC analysis revealed substantial performance variation across demographic subgroups, reinforcing the importance of covariate alignment in comparative studies.

stat.AP↗

arXiv2025

Virtual Lung Screening Trial (VLST): An In Silico Study Inspired by the National Lung Screening Trial for Lung Cancer Detection

Clinical imaging trials play a crucial role in advancing medical innovation but are often costly, inefficient, and ethically constrained. Virtual Imaging Trials (VITs) present a solution by simulating clinical trial components in a controlled, risk-free environment. The Virtual Lung Screening Trial (VLST), an in silico study inspired by the National Lung Screening Trial (NLST), illustrates the potential of VITs to expedite clinical trials, minimize risks to participants, and promote optimal use of imaging technologies in healthcare. This study aimed to show that a virtual imaging trial platform could investigate some key elements of a major clinical trial, specifically the NLST, which compared Computed tomography (CT) and chest radiography (CXR) for lung cancer screening. With simulated cancerous lung nodules, a virtual patient cohort of 294 subjects was created using XCAT human models. Each virtual patient underwent both CT and CXR imaging, with deep learning models, the AI CT-Reader and AI CXR-Reader, acting as virtual readers to perform recall patients with suspicion of lung cancer. The primary outcome was the difference in diagnostic performance between CT and CXR, measured by the Area Under the Curve (AUC). The AI CT-Reader showed superior diagnostic accuracy, achieving an AUC of 0.92 (95% CI: 0.90-0.95) compared to the AI CXR-Reader's AUC of 0.72 (95% CI: 0.67-0.77). Furthermore, at the same 94% CT sensitivity reported by the NLST, the VLST specificity of 73% was similar to the NLST specificity of 73.4%. This CT performance highlights the potential of VITs to replicate certain aspects of clinical trials effectively, paving the way toward a safe and efficient method for advancing imaging-based diagnostics.

eess.IV↗

arXiv2024

XCAT-3.0: A Comprehensive Library of Personalized Digital Twins Derived from CT Scans

Virtual Imaging Trials (VIT) offer a cost-effective and scalable approach for evaluating medical imaging technologies. Computational phantoms, which mimic real patient anatomy and physiology, play a central role in VITs. However, the current libraries of computational phantoms face limitations, particularly in terms of sample size and diversity. Insufficient representation of the population hampers accurate assessment of imaging technologies across different patient groups. Traditionally, the more realistic computational phantoms were created by manual segmentation, which is a laborious and time-consuming task, impeding the expansion of phantom libraries. This study presents a framework for creating realistic computational phantoms using a suite of automatic segmentation models and performing three forms of automated quality control on the segmented organ masks. The result is the release of over 2500 new computational phantoms, so-named XCAT3.0 after the ubiquitous XCAT computational construct. This new formation embodies 140 structures and represents a comprehensive approach to detailed anatomical modeling. The developed computational phantoms are formatted in both voxelized and surface mesh formats. The framework is combined with an in-house CT scanner simulator to produce realistic CT images. The framework has the potential to advance virtual imaging trials, facilitating comprehensive and reliable evaluations of medical imaging technologies. Phantoms may be requested at https://cvit.duke.edu/resources/. Code, model weights, and sample CT images are available at https://xcat-3.github.io/.

eess.IV↗

arXiv2023

A personalized Uncertainty Quantification framework for patient survival models: estimating individual uncertainty of patients with metastatic brain tumors in the absence of ground truth

TodevelopanovelUncertaintyQuantification (UQ) framework to estimate the uncertainty of patient survival models in the absence of ground truth, we developed and evaluated our approach based on a dataset of 1383 patients treated with stereotactic radiosurgery (SRS) for brain metastases between January 2015 and December 2020. Our motivating hypothesis is that a time-to-event prediction of a test patient on inference is more certain given a higher feature-space-similarity to patients in the training set. Therefore, the uncertainty for a particular patient-of-interest is represented by the concordance index between a patient similarity rank and a prediction similarity rank. Model uncertainty was defined as the increased percentage of the max uncertainty-constrained-AUC compared to the model AUC. We evaluated our method on multiple clinically-relevant endpoints, including time to intracranial progression (ICP), progression-free survival (PFS) after SRS, overall survival (OS), and time to ICP and/or death (ICPD), on a variety of both statistical and non-statistical models, including CoxPH, conditional survival forest (CSF), and neural multi-task linear regression (NMTLR). Our results show that all models had the lowest uncertainty on ICP (2.21%) and the highest uncertainty (17.28%) on ICPD. OS models demonstrated high variation in uncertainty performance, where NMTLR had the lowest uncertainty(1.96%)and CSF had the highest uncertainty (14.29%). In conclusion, our method can estimate the uncertainty of individual patient survival modeling results. As expected, our data empirically demonstrate that as model uncertainty measured via our technique increases, the similarity between a feature-space and its predicted outcome decreases.

cs.LG↗

arXiv2023

Duke Spleen Data Set: A Publicly Available Spleen MRI and CT dataset for Training Segmentation

Spleen volumetry is primarily associated with patients suffering from chronic liver disease and portal hypertension, as they often have spleens with abnormal shapes and sizes. However, manually segmenting the spleen to obtain its volume is a time-consuming process. Deep learning algorithms have proven to be effective in automating spleen segmentation, but a suitable dataset is necessary for training such algorithms. To our knowledge, the few publicly available datasets for spleen segmentation lack confounding features such as ascites and abdominal varices. To address this issue, the Duke Spleen Data Set (DSDS) has been developed, which includes 109 CT and MRI volumes from patients with chronic liver disease and portal hypertension. The dataset includes a diverse range of image types, vendors, planes, and contrasts, as well as varying spleen shapes and sizes due to underlying disease states. The DSDS aims to facilitate the creation of robust spleen segmentation models that can take into account these variations and confounding factors.

eess.IV↗

arXiv2021

A Radiomics-Boosted Deep-Learning Model for COVID-19 and Non-COVID-19 Pneumonia Classification Using Chest X-ray Image

To develop a deep-learning model that integrates radiomics analysis for enhanced performance of COVID-19 and Non-COVID-19 pneumonia detection using chest X-ray image, two deep-learning models were trained based on a pre-trained VGG-16 architecture: in the 1st model, X-ray image was the sole input; in the 2nd model, X-ray image and 2 radiomic feature maps (RFM) selected by the saliency map analysis of the 1st model were stacked as the input. Both models were developed using 812 chest X-ray images with 262/288/262 COVID-19/Non-COVID-19 pneumonia/healthy cases, and 649/163 cases were assigned as training-validation/independent test sets. In 1st model using X-ray as the sole input, the 1) sensitivity, 2) specificity, 3) accuracy, and 4) ROC Area-Under-the-Curve of COVID-19 vs Non-COVID-19 pneumonia detection were 1) 0.90$\pm$0.07 vs 0.78$\pm$0.09, 2) 0.94$\pm$0.04 vs 0.94$\pm$0.04, 3) 0.93$\pm$0.03 vs 0.89$\pm$0.03, and 4) 0.96$\pm$0.02 vs 0.92$\pm$0.04. In the 2nd model, two RFMs, Entropy and Short-Run-Emphasize, were selected with their highest cross-correlations with the saliency maps of the 1st model. The corresponding results demonstrated significant improvements (p<0.05) of COVID-19 vs Non-COVID-19 pneumonia detection: 1) 0.95$\pm$0.04 vs 0.85$\pm$0.04, 2) 0.97$\pm$0.02 vs 0.96$\pm$0.02, 3) 0.97$\pm$0.02 vs 0.93$\pm$0.02, and 4) 0.99$\pm$0.01 vs 0.97$\pm$0.02. The reduced variations suggested a superior robustness of 2nd model design.

eess.IV↗