Joseph Y. Lo

Fakrul Islam Tushar, Ehsan Abadi, Saman Sotoudeh-Paima et al.

Research studies of artificial intelligence models in medical imaging have been hampered by poor generalization. This problem has been especially concerning over the last year with numerous applications of deep learning for COVID-19 diagnosis. Virtual imaging trials (VITs) could provide a solution for objective evaluation of these models. In this work utilizing the VITs, we created the CVIT-COVID dataset including 180 virtually imaged computed tomography (CT) images from simulated COVID-19 and normal phantom models under different COVID-19 morphology and imaging properties. We evaluated the performance of an open-source, deep-learning model from the University of Waterloo trained with multi-institutional data and an in-house model trained with the open clinical dataset called MosMed. We further validated the model's performance against open clinical data of 305 CT images to understand virtual vs. real clinical data performance. The open-source model was published with nearly perfect performance on the original Waterloo dataset but showed a consistent performance drop in external testing on another clinical dataset (AUC=0.77) and our simulated CVIT-COVID dataset (AUC=0.55). The in-house model achieved an AUC of 0.87 while testing on the internal test set (MosMed test set). However, performance dropped to an AUC of 0.65 and 0.69 when evaluated on clinical and our simulated CVIT-COVID dataset. The VIT framework offered control over imaging conditions, allowing us to show there was no change in performance as CT exposure was changed from 28.5 to 57 mAs. The VIT framework also provided voxel-level ground truth, revealing that performance of in-house model was much higher at AUC=0.87 for diffuse COVID-19 infection size >2.65% lung volume versus AUC=0.52 for focal disease with <2.65% volume. The virtual imaging framework enabled these uniquely rigorous analyses of model performance.

Lavsen Dahal, Yubraj Bhandari, Geoffrey Rubin et al.

Abdominal CT disease classification is challenging because each scan is a large 3D volume with many possible findings, while diagnostic evidence is often confined to specific organs or anatomical compartments. Most study-level classifiers aggregate encoder features using anatomy-agnostic pooling or attention, creating a mismatch between localized disease evidence and global evidence aggregation. We propose ORACLE--CT, an encoder-agnostic anatomy-aware aggregation framework that uses multi-organ segmentation to define label-specific anatomical supports and restrict attention pooling to relevant regions. The framework supports single-organ, multi-organ union, comparative, localized, and global support strategies. We evaluate ORACLE--CT with three encoder families: DINOv3, I3D--ResNet-121, and the radiology-native Pillar--0 encoder. Models are trained end-to-end on MERLIN and evaluated internally and under frozen external transfer to Duke--Abdomen and AMOS. Compared with global average pooling, support-masked pooling improved MERLIN macro-AUROC/AUPRC from 0.838/0.638 to 0.858/0.676 for DINOv3 and from 0.829/0.617 to 0.848/0.659 for I3D--ResNet-121. On harmonized 10-label external evaluation, DINOv3 improved on Duke--Abdomen from 0.802/0.628 to 0.835/0.683 and on AMOS from 0.742/0.313 to 0.762/0.350, with similar gains for I3D--ResNet-121. For Pillar--0, most gains came from learned attention, with smaller additional benefit from anatomical masking. ORACLE--CT improves discrimination and external robustness while preserving an auditable link between predictions and anatomical evidence.

Fakrul Islam Tushar, Husam Nujaim, Wanyi Fu et al.

Organ segmentation of medical images is a key step in virtual imaging trials. However, organ segmentation datasets are limited in terms of quality (because labels cover only a few organs) and quantity (since case numbers are limited). In this study, we explored the tradeoffs between quality and quantity. Our goal is to create a unified approach for multi-organ segmentation of body CT, which will facilitate the creation of large numbers of accurate virtual phantoms. Initially, we compared two segmentation architectures, 3D-Unet and DenseVNet, which were trained using XCAT data that is fully labeled with 22 organs, and chose the 3D-Unet as the better performing model. We used the XCAT-trained model to generate pseudo-labels for the CT-ORG dataset that has only 7 organs segmented. We performed two experiments: First, we trained 3D-UNet model on the XCAT dataset, representing quality data, and tested it on both XCAT and CT-ORG datasets. Second, we trained 3D-UNet after including the CT-ORG dataset into the training set to have more quantity. Performance improved for segmentation in the organs where we have true labels in both datasets and degraded when relying on pseudo-labels. When organs were labeled in both datasets, Exp-2 improved Average DSC in XCAT and CT-ORG by 1. This demonstrates that quality data is the key to improving the model's performance.

Fakrul Islam Tushar, Lavsen Dahal, Saman Sotoudeh-Paima et al.

Purpose: The credibility of Artificial Intelligence (AI) models for medical imaging continues to be a challenge, affected by the diversity of models, the data used to train the models, and applicability of their combination to produce reproducible results for new data. In this work, we aimed to explore whether emerging Virtual Imaging Trials (VIT) methodologies can provide an objective resource to approach this challenge. Approach: The study was conducted for the case example of COVID-19 diagnosis using clinical and virtual computed tomography (CT) and chest radiography (CXR) processed with convolutional neural networks. Multiple AI models were developed and tested using 3D ResNet-like and 2D EfficientNetv2 architectures across diverse datasets. Results: Model performance was evaluated using the area under the curve (AUC) and the DeLong method for AUC confidence intervals. The models trained on the most diverse datasets showed the highest external testing performance, with AUC values ranging from 0.73-0.76 for CT and 0.70-0.73 for CXR. Internal testing yielded higher AUC values (0.77-0.85 for CT and 0.77-1.0 for CXR), highlighting a substantial drop in performance during external validation, which underscores the importance of diverse and comprehensive training and testing data. Most notably, the VIT approach provided objective assessment of the utility of diverse models and datasets, while offering insight into the influence of dataset characteristics, patient factors, and imaging physics on AI efficacy. Conclusions: The VIT approach enhances model transparency and reliability, offering nuanced insights into the factors driving AI performance and bridging the gap between experimental and clinical settings.

Fakrul Islam Tushar, Umme Hafsa Momy, Joseph Y. Lo et al.

We introduce iTRIALSPACE, a programmable evaluation framework for controlled assessment of lung CT models. Standard benchmarks are static retrospective collections that entangle lesion size, lobe prevalence, anatomy, and acquisition context, making it difficult to determine what structurally drives model accuracy. iTRIALSPACE addresses this limitation by composing real clinical CTs and lesion profiles into controlled virtual lesion trials through a four-stage pipeline: multidataset nodule profiling, explicit trial specification, anatomy-aware mask insertion, and ControlNet-conditioned CT synthesis. The framework is built on a unified 54-attribute nodule-profile dataset spanning 13,140 annotated nodules from seven public CT sources and instantiated as 13 trial modes. We evaluate iTRIALSPACE in a 55,469-sample Virtual Lesion Study spanning three medical VLMs, four spatialguidance conditions, and three clinical tasks. Across all 13 modes, the synthetic substrate remains within the real-to-real FID baseline, and synthetic performance rankings transfer strongly to real clinical data ($ρ$ = 0.93, p < 10$^{-15}$). Controlled trial modes expose findings unavailable to fixed-distribution benchmarks, including shortcut-driven size prediction collapse under lobe-equalized sampling and hostto-donor variance ratios of 8.9x and 3.3x in twin-cross analysis. These results position iTRIALSPACE as an auditable evaluation infrastructure for controlled, falsifiable testing beyond static retrospective benchmarks.

Lavsen Dahal, Yubraj Bhandari, Geoffrey Rubin et al.

Automated CT triage requires models that are simultaneously accurate across diverse pathologies and reliable under institutional shift. While Vision Transformers provide strong visual representations, many clinically significant findings are defined by quantitative imaging biomarkers rather than appearance alone. We introduce JANUS, a physiology-guided dual-stream architecture that conditions visual embeddings on macro-radiomic priors via Anatomically Guided Gating. On the MERLIN test set (N=5082), JANUS attains macro-AUROC 0.88 and AUPRC 0.74, outperforming all reproduced baselines. It generalizes to an external dataset N=2000; AUROC 0.87), with the largest gains on findings defined by size and attenuation as well as improved calibration on both datasets. We further quantify prediction suppression using the Physiological Veto Rate (PVR), showing that under domain shift JANUS reduces high-confidence false positives substantially more often than true positives. Together, these results are consistent with physically grounded conditioning that improves both discrimination and reliability in CT triage. Code is made publicly available at github repository https://github.com/lavsendahal/janus and model weights are at https://huggingface.co/lavsendahal/janus.

Kaouther Mouheb, Mobina Ghojogh Nejad, Lavsen Dahal et al.

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.

Yubraj Bhandari, Lavsen Dahal, Paul Segars et al.

Computational phantoms are widely used in medical imaging research, yet current systems to generate controlled, clinically meaningful anatomical variations remain limited. We present AbdomenGen, a sequential volume-conditioned diffusion framework for controllable abdominal anatomy generation. We introduce the \textbf{Volume Control Scalar (VCS)}, a standardized residual that decouples organ size from body habitus, enabling interpretable volume modulation. Organ masks are synthesized sequentially, conditioning on the body mask and previously generated structures to preserve global anatomical coherence while supporting independent, multi-organ control. Across 11 abdominal organs, the proposed framework achieves strong geometric fidelity (e.g., liver dice $0.83 \pm 0.05$), stable single-organ calibration over $[-3,+3]$ VCS, and disentangled multi-organ modulation. To showcase clinical utility with a hepatomegaly cohort selected from MERLIN, Wasserstein-based VCS selection reduces distributional distance of training data by 73.6\% . These results demonstrate calibrated, distribution-aware anatomical generation suitable for controllable abdominal phantom construction and simulation studies.

Fakrul Islam Tushar, Joseph Y. Lo

Using multiple open-access models trained on public datasets, we developed Tri-Reader, a comprehensive, freely available pipeline that integrates lung segmentation, nodule detection, and malignancy classification into a unified tri-stage workflow. The pipeline is designed to prioritize sensitivity while reducing the candidate burden for annotators. To ensure accuracy and generalizability across diverse practices, we evaluated Tri-Reader on multiple internal and external datasets as compared with expert annotations and dataset-provided reference standards.

Fakrul Islam Tushar, Ehsan Samei, Cynthia Rudin et al.

Objective: Although medical imaging datasets are increasingly available, abnormal and annotation-intensive findings critical to lung cancer screening, particularly small pulmonary nodules, remain underrepresented and inconsistently curated. Methods: We introduce NodMAISI, an anatomically constrained, nodule-oriented CT synthesis and augmentation framework trained on a unified multi-source cohort (7,042 patients, 8,841 CTs, 14,444 nodules). The framework integrates: (i) a standardized curation and annotation pipeline linking each CT with organ masks and nodule-level annotations, (ii) a ControlNet-conditioned rectified-flow generator built on MAISI-v2's foundational blocks to enforce anatomy- and lesion-consistent synthesis, and (iii) lesion-aware augmentation that perturbs nodule masks (controlled shrinkage) while preserving surrounding anatomy to generate paired CT variants. Results: Across six public test datasets, NodMAISI improved distributional fidelity relative to MAISI-v2 (real-to-synthetic FID range 1.18 to 2.99 vs 1.69 to 5.21). In lesion detectability analysis using a MONAI nodule detector, NodMAISI substantially increased average sensitivity and more closely matched clinical scans (IMD-CT: 0.69 vs 0.39; DLCS24: 0.63 vs 0.20), with the largest gains for sub-centimeter nodules where MAISI-v2 frequently failed to reproduce the conditioned lesion. In downstream nodule-level malignancy classification trained on LUNA25 and externally evaluated on LUNA16, LNDbv4, and DLCS24, NodMAISI augmentation improved AUC by 0.07 to 0.11 at <=20% clinical data and by 0.12 to 0.21 at 10%, consistently narrowing the performance gap under data scarcity.

Lavsen Dahal, Yubraj Bhandari, Geoffrey D. Rubin et al.

There is an urgent need for triage and classification of high-volume medical imaging modalities such as computed tomography (CT), which can improve patient care and mitigate radiologist burnout. Study-level CT triage requires calibrated predictions with localized evidence; however, off-the-shelf Vision Language Models (VLM) struggle with 3D anatomy, protocol shifts, and noisy report supervision. This study used the two largest publicly available chest CT datasets: CT-RATE and RADCHEST-CT (held-out external test set). Our carefully tuned supervised baseline (instantiated as a simple Global Average Pooling head) establishes a new supervised state of the art, surpassing all reported linear-probe VLMs. Building on this baseline, we present ORACLE-CT, an encoder-agnostic, organ-aware head that pairs Organ-Masked Attention (mask-restricted, per-organ pooling that yields spatial evidence) with Organ-Scalar Fusion (lightweight fusion of normalized volume and mean-HU cues). In the chest setting, ORACLE-CT masked attention model achieves AUROC 0.86 on CT-RATE; in the abdomen setting, on MERLIN (30 findings), our supervised baseline exceeds a reproduced zero-shot VLM baseline obtained by running publicly released weights through our pipeline, and adding masked attention plus scalar fusion further improves performance to AUROC 0.85. Together, these results deliver state-of-the-art supervised classification performance across both chest and abdomen CT under a unified evaluation protocol. The source code is available at https://github.com/lavsendahal/oracle-ct.

Lavsen Dahal, Joseph Y. Lo

In this retrospective multi-institutional study, a quantitative phenotyping framework, CT-IDP (CT Image-Derived Phenotypes) was developed on the MERLIN abdominal CT benchmark (training, validation, and test sets- 15,175, 5,018, and 5,082 studies, respectively) and externally evaluated on two independent dataset: Duke-Abdomen (2,000) and AMOS (1,107). Multi-organ segmentations were generated with TotalSegmentator and used to derive over 900 organ and compartment-level descriptors spanning morphometry, attenuation, and contextual/burden findings. Sparse disease-specific logistic regression with elastic-net regularization was trained on MERLIN and externally validated under a frozen specification. Performance was compared against a DINOv3-based vision-transformer baseline using AUC and average precision (AP), supported by phenotype-stratified audits and coefficient-level inspection. Macro-AUC for CT-IDP versus the baseline was 0.897 versus 0.880 on MERLIN, 0.877 versus 0.857 on the Duke-Abdomen dataset, and 0.780 versus 0.756 on AMOS.

Fengwei Tian, Payel Bhattacharjee, Heidi Hanson et al.

We present STAMP (Selective Task-Aware Mechanism for Text Privacy), a new framework for task-aware text privatization that achieves an improved privacy-utility trade-off. STAMP selectively allocates privacy budgets across tokens by jointly considering (i) each token's importance to the downstream task (as measured via a task- or query-specific representation), and (ii) its privacy sensitivity (e.g., names, dates, identifiers). This token-level partitioning enables fine-grained, group-wise control over the level of noise applied to different parts of the input, balancing privacy protection with task relevance. To privatize individual token embeddings, we introduce the polar mechanism, which perturbs only the direction of embeddings on the unit sphere while preserving their magnitude. Decoding is performed via cosine nearest-neighbor search, aligning the perturbation geometry with the decoding geometry. Unlike isotropic noise mechanisms, the polar mechanism maintains semantic neighborhoods in the embedding space and better preserves downstream utility. Experimental evaluations on SQuAD, Yelp, and AG News datasets demonstrate that STAMP, when combined with the normalized polar mechanism, consistently achieves superior privacy-utility trade-offs across varying per-token privacy budgets.

Fakrul Islam Tushar, Avivah Wang, Lavsen Dahal et al.

Background: Development of artificial intelligence (AI) models for lung cancer screening requires large, well-annotated low-dose computed tomography (CT) datasets and rigorous performance benchmarks. Purpose: To create a reproducible benchmarking resource leveraging the Duke Lung Cancer Screening (DLCS) and multiple public datasets to develop and evaluate models for nodule detection and classification. Materials & Methods: This retrospective study uses the DLCS dataset (1,613 patients; 2,487 nodules) and external datasets including LUNA16, LUNA25, and NLST-3D. For detection, MONAI RetinaNet models were trained on DLCS (DLCS-De) and LUNA16 (LUNA16-De) and evaluated using the Competition Performance Metric (CPM). For nodule-level classification, we compare five strategies: pretrained models (Models Genesis, Med3D), a self-supervised foundation model (FMCB), and ResNet50 with random initialization versus Strategic Warm-Start (ResNet50-SWS) pretrained with detection-derived candidate patches stratified by confidence. Results: For detection on the DLCS test set, DLCS-De achieved sensitivity 0.82 at 2 false positives/scan (CPM 0.63) versus LUNA16-De (0.62, CPM 0.45). For external validation on NLST-3D, DLCS-De (sensitivity 0.72, CPM 0.58) also outperformed LUNA16-De (sensitivity 0.64, CPM 0.49). For classification across multiple datasets, ResNet50-SWS attained AUCs of 0.71 (DLCS; 95% CI, 0.61-0.81), 0.90 (LUNA16; 0.87-0.93), 0.81 (NLST-3D; 0.79-0.82), and 0.80 (LUNA25; 0.78-0.82), matching or exceeding pretrained/self-supervised baselines. Performance differences reflected dataset label standards. Conclusion: This work establishes a standardized benchmarking resource for lung cancer AI research, supporting model development, validation, and translation. All code, models, and data are publicly released to promote reproducibility.

Fakrul Islam Tushar, Lavsen Dahal, Cindy McCabe et al.

AI models for lung cancer screening are limited by data scarcity, impacting generalizability and clinical applicability. Generative models address this issue but are constrained by training data variability. We introduce SYN-LUNGS, a framework for generating high-quality 3D CT images with detailed annotations. SYN-LUNGS integrates XCAT3 phantoms for digital twin generation, X-Lesions for nodule simulation (varying size, location, and appearance), and DukeSim for CT image formation with vendor and parameter variability. The dataset includes 3,072 nodule images from 1,044 simulated CT scans, with 512 lesions and 174 digital twins. Models trained on clinical + simulated data outperform clinical only models, achieving 10% improvement in detection, 2-9% in segmentation and classification, and enhanced synthesis. By incorporating anatomy-informed simulations, SYN-LUNGS provides a scalable approach for AI model development, particularly in rare disease representation and improving model reliability.

Xuan Liu, Yinhao Ren, Marc D. Ryser et al.

Accurate lesion tracking in temporal mammograms is essential for monitoring breast cancer progression and facilitating early diagnosis. However, automated lesion correspondence across exams remains a challenges in computer-aided diagnosis (CAD) systems, limiting their effectiveness. We propose MammoTracker, a mask-guided lesion tracking framework that automates lesion localization across consecutively exams. Our approach follows a coarse-to-fine strategy incorporating three key modules: global search, local search, and score refinement. To support large-scale training and evaluation, we introduce a new dataset with curated prior-exam annotations for 730 mass and calcification cases from the public EMBED mammogram dataset, yielding over 20000 lesion pairs, making it the largest known resource for temporal lesion tracking in mammograms. Experimental results demonstrate that MammoTracker achieves 0.455 average overlap and 0.509 accuracy, surpassing baseline models by 8%, highlighting its potential to enhance CAD-based lesion progression analysis. Our dataset will be available at https://gitlab.oit.duke.edu/railabs/LoGroup/mammotracker.

Payel Bhattacharjee, Fengwei Tian, Geoffrey D. Rubin et al.

Purpose: This study proposes a framework for fine-tuning large language models (LLMs) with differential privacy (DP) to perform multi-abnormality classification on radiology report text. By injecting calibrated noise during fine-tuning, the framework seeks to mitigate the privacy risks associated with sensitive patient data and protect against data leakage while maintaining classification performance. Materials and Methods: We used 50,232 radiology reports from the publicly available MIMIC-CXR chest radiography and CT-RATE computed tomography datasets, collected between 2011 and 2019. Fine-tuning of LLMs was conducted to classify 14 labels from MIMIC-CXR dataset, and 18 labels from CT-RATE dataset using Differentially Private Low-Rank Adaptation (DP-LoRA) in high and moderate privacy regimes (across a range of privacy budgets = {0.01, 0.1, 1.0, 10.0}). Model performance was evaluated using weighted F1 score across three model architectures: BERT-medium, BERT-small, and ALBERT-base. Statistical analyses compared model performance across different privacy levels to quantify the privacy-utility trade-off. Results: We observe a clear privacy-utility trade-off through our experiments on 2 different datasets and 3 different models. Under moderate privacy guarantees the DP fine-tuned models achieved comparable weighted F1 scores of 0.88 on MIMIC-CXR and 0.59 on CT-RATE, compared to non-private LoRA baselines of 0.90 and 0.78, respectively. Conclusion: Differentially private fine-tuning using LoRA enables effective and privacy-preserving multi-abnormality classification from radiology reports, addressing a key challenge in fine-tuning LLMs on sensitive medical data.

Michael E. Garcia-Alcoser, Mobina GhojoghNejad, Fakrul Islam Tushar et al.

Purpose: This study aims to evaluate the effectiveness of large language models (LLMs) in automating disease annotation of CT radiology reports. We compare a rule-based algorithm (RBA), RadBERT, and three lightweight open-weight LLMs for multi-disease labeling of chest, abdomen, and pelvis (CAP) CT reports. Materials and Methods: This retrospective study analyzed 40,833 CT reports from 29,540 patients, with 1,789 CAP reports manually annotated across three organ systems. External validation was conducted using the CT-RATE dataset. Three open-weight LLMs were tested with zero-shot prompting. Performance was evaluated using Cohen's Kappa and micro/macro-averaged F1 scores. Results: In 12,197 Duke CAP reports from 8,854 patients, Llama-3.1 8B and Gemma-3 27B showed the highest agreement ($κ$ median: 0.87). On the manually annotated set, Gemma-3 27B achieved the top macro-F1 (0.82), followed by Llama-3.1 8B (0.79), while the RBA scored lowest (0.64). On the CT-RATE dataset (lungs/pleura only), Llama-3.1 8B performed best (0.91), with Gemma-3 27B close behind (0.89). Performance differences were mainly due to differing labeling practices, especially for lung atelectasis. Conclusion: Lightweight LLMs outperform rule-based methods for CT report annotation and generalize across organ systems with zero-shot prompting. However, binary labels alone cannot capture the full nuance of report language. LLMs can provide a flexible, efficient solution aligned with clinical judgment and user needs.

Julia Yang, Alina Jade Barnett, Jon Donnelly et al.

Digital mammography is essential to breast cancer detection, and deep learning offers promising tools for faster and more accurate mammogram analysis. In radiology and other high-stakes environments, uninterpretable ("black box") deep learning models are unsuitable and there is a call in these fields to make interpretable models. Recent work in interpretable computer vision provides transparency to these formerly black boxes by utilizing prototypes for case-based explanations, achieving high accuracy in applications including mammography. However, these models struggle with precise feature localization, reasoning on large portions of an image when only a small part is relevant. This paper addresses this gap by proposing a novel multi-scale interpretable deep learning model for mammographic mass margin classification. Our contribution not only offers an interpretable model with reasoning aligned with radiologist practices, but also provides a general architecture for computer vision with user-configurable prototypes from coarse- to fine-grained prototypes.

Lavsen Dahal, Mobina Ghojoghnejad, Dhrubajyoti Ghosh et al.

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/.

Fakrul Islam Tushar, Vincent M. D'Anniballe, Geoffrey D. Rubin et al.

Weakly supervised learning with noisy data has drawn attention in the medical imaging community due to the sparsity of high-quality disease labels. However, little is known about the limitations of such weakly supervised learning and the effect of these constraints on disease classification performance. In this paper, we test the effects of such weak supervision by examining model tolerance for three conditions. First, we examined model tolerance for noisy data by incrementally increasing error in the labels within the training data. Second, we assessed the impact of dataset size by varying the amount of training data. Third, we compared performance differences between binary and multi-label classification. Results demonstrated that the model could endure up to 10% added label error before experiencing a decline in disease classification performance. Disease classification performance steadily rose as the amount of training data was increased for all disease classes, before experiencing a plateau in performance at 75% of training data. Last, the binary model outperformed the multilabel model in every disease category. However, such interpretations may be misleading, as the binary model was heavily influenced by co-occurring diseases and may not have learned the specific features of the disease in the image. In conclusion, this study may help the medical imaging community understand the benefits and risks of weak supervision with noisy labels. Such studies demonstrate the need to build diverse, large-scale datasets and to develop explainable and responsible AI.

Alina Jade Barnett, Fides Regina Schwartz, Chaofan Tao et al.

When we deploy machine learning models in high-stakes medical settings, we must ensure these models make accurate predictions that are consistent with known medical science. Inherently interpretable networks address this need by explaining the rationale behind each decision while maintaining equal or higher accuracy compared to black-box models. In this work, we present a novel interpretable neural network algorithm that uses case-based reasoning for mammography. Designed to aid a radiologist in their decisions, our network presents both a prediction of malignancy and an explanation of that prediction using known medical features. In order to yield helpful explanations, the network is designed to mimic the reasoning processes of a radiologist: our network first detects the clinically relevant semantic features of each image by comparing each new image with a learned set of prototypical image parts from the training images, then uses those clinical features to predict malignancy. Compared to other methods, our model detects clinical features (mass margins) with equal or higher accuracy, provides a more detailed explanation of its prediction, and is better able to differentiate the classification-relevant parts of the image.

Alina Jade Barnett, Fides Regina Schwartz, Chaofan Tao et al.

Interpretability in machine learning models is important in high-stakes decisions, such as whether to order a biopsy based on a mammographic exam. Mammography poses important challenges that are not present in other computer vision tasks: datasets are small, confounding information is present, and it can be difficult even for a radiologist to decide between watchful waiting and biopsy based on a mammogram alone. In this work, we present a framework for interpretable machine learning-based mammography. In addition to predicting whether a lesion is malignant or benign, our work aims to follow the reasoning processes of radiologists in detecting clinically relevant semantic features of each image, such as the characteristics of the mass margins. The framework includes a novel interpretable neural network algorithm that uses case-based reasoning for mammography. Our algorithm can incorporate a combination of data with whole image labelling and data with pixel-wise annotations, leading to better accuracy and interpretability even with a small number of images. Our interpretable models are able to highlight the classification-relevant parts of the image, whereas other methods highlight healthy tissue and confounding information. Our models are decision aids, rather than decision makers, aimed at better overall human-machine collaboration. We do not observe a loss in mass margin classification accuracy over a black box neural network trained on the same data.

Vincent M. D'Anniballe, Fakrul Islam Tushar, Khrystyna Faryna et al.

Purpose: To develop high throughput multi-label annotators for body (chest, abdomen, and pelvis) Computed Tomography (CT) reports that can be applied across a variety of abnormalities, organs, and disease states. Approach: We used a dictionary approach to develop rule-based algorithms (RBA) for extraction of disease labels from radiology text reports. We targeted three organ systems (lungs/pleura, liver/gallbladder, kidneys/ureters) with four diseases per system based on their prevalence in our dataset. To expand the algorithms beyond pre-defined keywords, attention-guided recurrent neural networks (RNN) were trained using the RBA-extracted labels to classify reports as being positive for one or more diseases or normal for each organ system. Confounding effects on model performance were evaluated using random initialization or pre-trained embedding as well as different sizes of training datasets. Performance was evaluated using the receiver operating characteristic (ROC) area under the curve (AUC) against 2,158 manually obtained labels. Results: Our models extracted disease labels from 261,229 radiology reports of 112,501 unique subjects. Pre-trained models outperformed random initialization across all diseases. As the training dataset size was reduced, performance was robust except for a few diseases with relatively small number of cases. Pre-trained classification AUCs achieved > 0.95 for all five disease outcomes across all three organ systems. Conclusions: Our label-extracting pipeline was able to encompass a variety of cases and diseases by generalizing beyond strict rules with exceptional accuracy. This method can be easily adapted to enable automated labeling of hospital-scale medical data sets for training image-based disease classifiers.

Mateusz Buda, Ashirbani Saha, Ruth Walsh et al.

Breast cancer screening is one of the most common radiological tasks with over 39 million exams performed each year. While breast cancer screening has been one of the most studied medical imaging applications of artificial intelligence, the development and evaluation of the algorithms are hindered due to the lack of well-annotated large-scale publicly available datasets. This is particularly an issue for digital breast tomosynthesis (DBT) which is a relatively new breast cancer screening modality. We have curated and made publicly available a large-scale dataset of digital breast tomosynthesis images. It contains 22,032 reconstructed DBT volumes belonging to 5,610 studies from 5,060 patients. This included four groups: (1) 5,129 normal studies, (2) 280 studies where additional imaging was needed but no biopsy was performed, (3) 112 benign biopsied studies, and (4) 89 studies with cancer. Our dataset included masses and architectural distortions which were annotated by two experienced radiologists. Additionally, we developed a single-phase deep learning detection model and tested it using our dataset to serve as a baseline for future research. Our model reached a sensitivity of 65% at 2 false positives per breast. Our large, diverse, and highly-curated dataset will facilitate development and evaluation of AI algorithms for breast cancer screening through providing data for training as well as common set of cases for model validation. The performance of the model developed in our study shows that the task remains challenging and will serve as a baseline for future model development.

Anindo Saha, Fakrul I. Tushar, Khrystyna Faryna et al.

Weakly supervised disease classification of CT imaging suffers from poor localization owing to case-level annotations, where even a positive scan can hold hundreds to thousands of negative slices along multiple planes. Furthermore, although deep learning segmentation and classification models extract distinctly unique combinations of anatomical features from the same target class(es), they are typically seen as two independent processes in a computer-aided diagnosis (CAD) pipeline, with little to no feature reuse. In this research, we propose a medical classifier that leverages the semantic structural concepts learned via multi-resolution segmentation feature maps, to guide weakly supervised 3D classification of chest CT volumes. Additionally, a comparative analysis is drawn across two different types of feature aggregation to explore the vast possibilities surrounding feature fusion. Using a dataset of 1593 scans labeled on a case-level basis via rule-based model, we train a dual-stage convolutional neural network (CNN) to perform organ segmentation and binary classification of four representative diseases (emphysema, pneumonia/atelectasis, mass and nodules) in lungs. The baseline model, with separate stages for segmentation and classification, results in AUC of 0.791. Using identical hyperparameters, the connected architecture using static and dynamic feature aggregation improves performance to AUC of 0.832 and 0.851, respectively. This study advances the field in two key ways. First, case-level report data is used to weakly supervise a 3D CT classifier of multiple, simultaneous diseases for an organ. Second, segmentation and classification models are connected with two different feature aggregation strategies to enhance the classification performance.

Wanyi Fu, Shobhit Sharma, Ehsan Abadi et al.

Objective: This study aims to develop and validate a novel framework, iPhantom, for automated creation of patient-specific phantoms or digital-twins (DT) using patient medical images. The framework is applied to assess radiation dose to radiosensitive organs in CT imaging of individual patients. Method: From patient CT images, iPhantom segments selected anchor organs (e.g. liver, bones, pancreas) using a learning-based model developed for multi-organ CT segmentation. Organs challenging to segment (e.g. intestines) are incorporated from a matched phantom template, using a diffeomorphic registration model developed for multi-organ phantom-voxels. The resulting full-patient phantoms are used to assess organ doses during routine CT exams. Result: iPhantom was validated on both the XCAT (n=50) and an independent clinical (n=10) dataset with similar accuracy. iPhantom precisely predicted all organ locations with good accuracy of Dice Similarity Coefficients (DSC) >0.6 for anchor organs and DSC of 0.3-0.9 for all other organs. iPhantom showed less than 10% dose errors for the majority of organs, which was notably superior to the state-of-the-art baseline method (20-35% dose errors). Conclusion: iPhantom enables automated and accurate creation of patient-specific phantoms and, for the first time, provides sufficient and automated patient-specific dose estimates for CT dosimetry. Significance: The new framework brings the creation and application of CHPs to the level of individual CHPs through automation, achieving a wider and precise organ localization, paving the way for clinical monitoring, and personalized optimization, and large-scale research.

Fakrul Islam Tushar, Vincent M. D'Anniballe, Rui Hou et al.

Purpose: To design multi-disease classifiers for body CT scans for three different organ systems using automatically extracted labels from radiology text reports.Materials & Methods: This retrospective study included a total of 12,092 patients (mean age 57 +- 18; 6,172 women) for model development and testing (from 2012-2017). Rule-based algorithms were used to extract 19,225 disease labels from 13,667 body CT scans from 12,092 patients. Using a three-dimensional DenseVNet, three organ systems were segmented: lungs and pleura; liver and gallbladder; and kidneys and ureters. For each organ, a three-dimensional convolutional neural network classified no apparent disease versus four common diseases for a total of 15 different labels across all three models. Testing was performed on a subset of 2,158 CT volumes relative to 2,875 manually derived reference labels from 2133 patients (mean age 58 +- 18;1079 women). Performance was reported as receiver operating characteristic area under the curve (AUC) with 95% confidence intervals by the DeLong method. Results: Manual validation of the extracted labels confirmed 91% to 99% accuracy across the 15 different labels. AUCs for lungs and pleura labels were: atelectasis 0.77 (95% CI: 0.74, 0.81), nodule 0.65 (0.61, 0.69), emphysema 0.89 (0.86, 0.92), effusion 0.97 (0.96, 0.98), and no apparent disease 0.89 (0.87, 0.91). AUCs for liver and gallbladder were: hepatobiliary calcification 0.62 (95% CI: 0.56, 0.67), lesion 0.73 (0.69, 0.77), dilation 0.87 (0.84, 0.90), fatty 0.89 (0.86, 0.92), and no apparent disease 0.82 (0.78, 0.85). AUCs for kidneys and ureters were: stone 0.83 (95% CI: 0.79, 0.87), atrophy 0.92 (0.89, 0.94), lesion 0.68 (0.64, 0.72), cyst 0.70 (0.66, 0.73), and no apparent disease 0.79 (0.75, 0.83). Conclusion: Weakly-supervised deep learning models were able to classify diverse diseases in multiple organ systems.

Rachel Lea Draelos, David Dov, Maciej A. Mazurowski et al.

Machine learning models for radiology benefit from large-scale data sets with high quality labels for abnormalities. We curated and analyzed a chest computed tomography (CT) data set of 36,316 volumes from 19,993 unique patients. This is the largest multiply-annotated volumetric medical imaging data set reported. To annotate this data set, we developed a rule-based method for automatically extracting abnormality labels from free-text radiology reports with an average F-score of 0.976 (min 0.941, max 1.0). We also developed a model for multi-organ, multi-disease classification of chest CT volumes that uses a deep convolutional neural network (CNN). This model reached a classification performance of AUROC greater than 0.90 for 18 abnormalities, with an average AUROC of 0.773 for all 83 abnormalities, demonstrating the feasibility of learning from unfiltered whole volume CT data. We show that training on more labels improves performance significantly: for a subset of 9 labels - nodule, opacity, atelectasis, pleural effusion, consolidation, mass, pericardial effusion, cardiomegaly, and pneumothorax - the model's average AUROC increased by 10% when the number of training labels was increased from 9 to all 83. All code for volume preprocessing, automated label extraction, and the volume abnormality prediction model will be made publicly available. The 36,316 CT volumes and labels will also be made publicly available pending institutional approval.