Magdalini Paschali

All Publications

• Spectral Graph Sample Weighting for Interpretable Sub-cohort Analysis in Predictive Models for Neuroimaging. PRedictive Intelligence in MEdicine. PRIME (Workshop) Paschali, M., Jiang, Y. H., Siegel, S., Gonzalez, C., Pohl, K. M., Chaudhari, A., Zhao, Q. 2025; 15155: 24-34

  Abstract

  Recent advancements in medicine have confirmed that brain disorders often comprise multiple subtypes of mechanisms, developmental trajectories, or severity levels. Such heterogeneity is often associated with demographic aspects (e.g., sex) or disease-related contributors (e.g., genetics). Thus, the predictive power of machine learning models used for symptom prediction varies across subjects based on such factors. To model this heterogeneity, one can assign each training sample a factor-dependent weight, which modulates the subject's contribution to the overall objective loss function. To this end, we propose to model the subject weights as a linear combination of the eigenbases of a spectral population graph that captures the similarity of factors across subjects. In doing so, the learned weights smoothly vary across the graph, highlighting sub-cohorts with high and low predictability. Our proposed sample weighting scheme is evaluated on two tasks. First, we predict initiation of heavy alcohol drinking in young adulthood from imaging and neuropsychological measures from the National Consortium on Alcohol and NeuroDevelopment in Adolescence (NCANDA). Next, we detect Dementia vs. Mild Cognitive Impairment (MCI) using imaging and demographic measurements in subjects from the Alzheimer's Disease Neuroimaging Initiative (ADNI). Compared to existing sample weighting schemes, our sample weights improve interpretability and highlight sub-cohorts with distinct characteristics and varying model accuracy.

  View details for DOI 10.1007/978-3-031-74561-4_3

  View details for PubMedID 39525051

  View details for PubMedCentralID PMC11549025

• Assessing Completeness of Clinical Histories Accompanying Imaging Orders Using Adapted Open-Source and Closed-Source Large Language Models. Radiology Larson, D. B., Koirala, A., Cheuy, L. Y., Paschali, M., Van Veen, D., Na, H. S., Petterson, M. B., Fang, Z., Chaudhari, A. S. 2025; 314 (2): e241051

  Abstract

  Background Incomplete clinical histories are a well-known problem in radiology. Previous dedicated quality improvement efforts focusing on reproducible assessments of the completeness of free-text clinical histories have relied on tedious manual analysis. Purpose To adapt and evaluate open-source and closed-source large language models (LLMs) for their ability to automatically extract clinical history elements within imaging orders and to use the best-performing adapted open-source model to assess the completeness of a large sample of clinical histories as a benchmark for clinical practice. Materials and Methods This retrospective single-site study used previously extracted information accompanying CT, MRI, US, and radiography orders from August 2020 to May 2022 at an adult and pediatric emergency department of a 613-bed tertiary academic medical center. Two open-source (Llama 2-7B [Meta], Mistral-7B [Mistral AI]) and one closed-source (GPT-4 Turbo [OpenAI]) LLMs were adapted using prompt engineering, in-context learning, and fine-tuning (open-source only) to extract the elements "past medical history," "what," "when," "where," and "clinical concern" from clinical histories. Model performance, interreader agreement using Cohen κ (none to slight, 0.01-0.20; fair, 0.21-0.40; moderate, 0.41-0.60; substantial, 0.61-0.80; almost perfect, 0.81-1.00), and semantic similarity between the models and the adjudicated manual annotations of two board-certified radiologists with 16 and 3 years of postfellowship experience, respectively, were assessed using accuracy, Cohen κ, and BERTScore, an LLM metric that quantifies how well two pieces of text convey the same meaning; 95% CIs were also calculated. The best-performing open-source model was then used to assess completeness on a large dataset of unannotated clinical histories. Results A total of 50 186 clinical histories were included (794 training, 150 validation, 300 initial testing, 48 942 real-world application). Of the two open-source models, Mistral-7B outperformed Llama 2-7B in assessing completeness and was further fine-tuned. Both Mistral-7B and GPT-4 Turbo showed substantial overall agreement with radiologists (mean κ, 0.73 [95% CI: 0.67, 0.78] to 0.77 [95% CI: 0.71, 0.82]) and adjudicated annotations (mean BERTScore, 0.96 [95% CI: 0.96, 0.97] for both models; P = .38). Mistral-7B also rivaled GPT-4 Turbo in performance (weighted overall mean accuracy, 91% [95% CI: 89, 93] vs 92% [95% CI: 90, 94]; P = .31) despite being a smaller model. Using Mistral-7B, 26.2% (12 803 of 48 942) of unannotated clinical histories were found to contain all five elements. Conclusion An easily deployable fine-tuned open-source LLM (Mistral-7B), rivaling GPT-4 Turbo in performance, could effectively extract clinical history elements with substantial agreement with radiologists and produce a benchmark for completeness of a large sample of clinical histories. The model and code will be fully open-sourced. © RSNA, 2025 Supplemental material is available for this article.

  View details for DOI 10.1148/radiol.241051

  View details for PubMedID 39998369

• Foundation Models in Radiology: What, How, Why, and Why Not. Radiology Paschali, M., Chen, Z., Blankemeier, L., Varma, M., Youssef, A., Bluethgen, C., Langlotz, C., Gatidis, S., Chaudhari, A. 2025; 314 (2): e240597

  Abstract

  Recent advances in artificial intelligence have witnessed the emergence of large-scale deep learning models capable of interpreting and generating both textual and imaging data. Such models, typically referred to as foundation models (FMs), are trained on extensive corpora of unlabeled data and demonstrate high performance across various tasks. FMs have recently received extensive attention from academic, industry, and regulatory bodies. Given the potentially transformative impact that FMs can have on the field of radiology, radiologists must be aware of potential pathways to train these radiology-specific FMs, including understanding both the benefits and challenges. Thus, this review aims to explain the fundamental concepts and terms of FMs in radiology, with a specific focus on the requirements of training data, model training paradigms, model capabilities, and evaluation strategies. Overall, the goal of this review is to unify technical advances and clinical needs for safe and responsible training of FMs in radiology to ultimately benefit patients, providers, and radiologists.

  View details for DOI 10.1148/radiol.240597

  View details for PubMedID 39903075

• Multi-domain predictors of grip strength differentiate individuals with and without alcohol use disorder. Addiction biology Paschali, M., Zhao, Q., Sassoon, S. A., Pfefferbaum, A., Sullivan, E. V., Pohl, K. M. 2024; 29 (11): e70007

  Abstract

  Grip strength is considered one of the simplest and reliable indices of general health. Although motor ability and strength are commonly affected in people with alcohol use disorder (AUD), factors predictive of grip strength decline in AUD have not been investigated. Here, we employed a data-driven analysis predicting grip strength from measurements in 53 controls and 110 AUD participants, 53 of whom were comorbid with HIV infection. Controls and AUD were matched on sex, age, and body mass index. Measurements included commonly available metrics of brain structure, neuropsychological functioning, behavioural status, haematological and health status, and demographics. Based on 5-fold stratified cross-validation, a machine learning approach predicted grip strength separately for each cohort. The strongest (top 10%) predictors of grip were then tested against grip strength with correlational analysis. Leading grip strength predictors for both cohorts were cerebellar volume and mean corpuscular haemoglobin concentration. Predictors specific to controls were Backwards Digit Span, precentral gyrus volume, diastolic blood pressure, and mean platelet volume, which together significantly predicted grip strength (R2 = 0.255, p = 0.001). Unique predictors for AUD were comorbidity for HIV infection, social functioning, insular volume, and platelet count, which together significantly predicted grip strength (R2 = 0.162, p = 0.002). These cohort-specific predictors were doubly dissociated. Salient predictors of grip strength differed by diagnosis with only modest overlap. The constellation of cohort-specific predictive measurements of compromised grip strength provides insight into brain, behavioural, and physiological factors that may signal subtly affected yet treatable processes of physical decline and frailty.

  View details for DOI 10.1111/adb.70007

  View details for PubMedID 39532141

• Merlin: A Vision Language Foundation Model for 3D Computed Tomography. Research square Blankemeier, L., Cohen, J. P., Kumar, A., Veen, D. V., Gardezi, S., Paschali, M., Chen, Z., Delbrouck, J. B., Reis, E., Truyts, C., Bluethgen, C., Jensen, M., Ostmeier, S., Varma, M., Valanarasu, J., Fang, Z., Huo, Z., Nabulsi, Z., Ardila, D., Weng, W. H., Junior, E. A., Ahuja, N., Fries, J., Shah, N., Johnston, A., Boutin, R., Wentland, A., Langlotz, C., Hom, J., Gatidis, S., Chaudhari, A. 2024

  Abstract

  Over 85 million computed tomography (CT) scans are performed annually in the US, of which approximately one quarter focus on the abdomen. Given the current shortage of both general and specialized radiologists, there is a large impetus to use artificial intelligence to alleviate the burden of interpreting these complex imaging studies while simultaneously using the images to extract novel physiological insights. Prior state-of-the-art approaches for automated medical image interpretation leverage vision language models (VLMs) that utilize both the image and the corresponding textual radiology reports. However, current medical VLMs are generally limited to 2D images and short reports. To overcome these shortcomings for abdominal CT interpretation, we introduce Merlin - a 3D VLM that leverages both structured electronic health records (EHR) and unstructured radiology reports for pretraining without requiring additional manual annotations. We train Merlin using a high-quality clinical dataset of paired CT scans (6+ million images from 15,331 CTs), EHR diagnosis codes (1.8+ million codes), and radiology reports (6+ million tokens) for training. We comprehensively evaluate Merlin on 6 task types and 752 individual tasks. The non-adapted (off-the-shelf) tasks include zero-shot findings classification (31 findings), phenotype classification (692 phenotypes), and zero-shot cross-modal retrieval (image to findings and image to impressions), while model adapted tasks include 5-year chronic disease prediction (6 diseases), radiology report generation, and 3D semantic segmentation (20 organs). We perform internal validation on a test set of 5,137 CTs, and external validation on 7,000 clinical CTs and on two public CT datasets (VerSe, TotalSegmentator). Beyond these clinically-relevant evaluations, we assess the efficacy of various network architectures and training strategies to depict that Merlin has favorable performance to existing task-specific baselines. We derive data scaling laws to empirically assess training data needs for requisite downstream task performance. Furthermore, unlike conventional VLMs that require hundreds of GPUs for training, we perform all training on a single GPU. This computationally efficient design can help democratize foundation model training, especially for health systems with compute constraints. We plan to release our trained models, code, and dataset, pending manual removal of all protected health information.

  View details for DOI 10.21203/rs.3.rs-4546309/v1

  View details for PubMedID 38978576

  View details for PubMedCentralID PMC11230513

• Identifying high school risk factors that forecast heavy drinking onset in understudied young adults. Developmental cognitive neuroscience Zhao, Q., Paschali, M., Dehoney, J., Baker, F. C., de Zambotti, M., De Bellis, M. D., Goldston, D. B., Nooner, K. B., Clark, D. B., Luna, B., Nagel, B. J., Brown, S. A., Tapert, S. F., Eberson, S., Thompson, W. K., Pfefferbaum, A., Sullivan, E. V., Pohl, K. M. 2024; 68: 101413

  Abstract

  Heavy alcohol drinking is a major, preventable problem that adversely impacts the physical and mental health of US young adults. Studies seeking drinking risk factors typically focus on young adults who enrolled in 4-year residential college programs (4YCP) even though most high school graduates join the workforce, military, or community colleges. We examined 106 of these understudied young adults (USYA) and 453 4YCPs from the National Consortium on Alcohol and NeuroDevelopment in Adolescence (NCANDA) by longitudinally following their drinking patterns for 8 years from adolescence to young adulthood. All participants were no-to-low drinkers during high school. Whereas 4YCP individuals were more likely to initiate heavy drinking during college years, USYA participants did so later. Using mental health metrics recorded during high school, machine learning forecasted individual-level risk for initiating heavy drinking after leaving high school. The risk factors differed between demographically matched USYA and 4YCP individuals and between sexes. Predictors for USYA drinkers were sexual abuse, physical abuse for girls, and extraversion for boys, whereas 4YCP drinkers were predicted by the ability to recognize facial emotion and, for boys, greater openness. Thus, alcohol prevention programs need to give special consideration to those joining the workforce, military, or community colleges, who make up the majority of this age group.

  View details for DOI 10.1016/j.dcn.2024.101413

  View details for PubMedID 38943839

• Multimodal graph attention network for COVID-19 outcome prediction. Scientific reports Keicher, M., Burwinkel, H., Bani-Harouni, D., Paschali, M., Czempiel, T., Burian, E., Makowski, M. R., Braren, R., Navab, N., Wendler, T. 2023; 13 (1): 19539

  Abstract

  When dealing with a newly emerging disease such as COVID-19, the impact of patient- and disease-specific factors (e.g., body weight or known co-morbidities) on the immediate course of the disease is largely unknown. An accurate prediction of the most likely individual disease progression can improve the planning of limited resources and finding the optimal treatment for patients. In the case of COVID-19, the need for intensive care unit (ICU) admission of pneumonia patients can often only be determined on short notice by acute indicators such as vital signs (e.g., breathing rate, blood oxygen levels), whereas statistical analysis and decision support systems that integrate all of the available data could enable an earlier prognosis. To this end, we propose a holistic, multimodal graph-based approach combining imaging and non-imaging information. Specifically, we introduce a multimodal similarity metric to build a population graph that shows a clustering of patients. For each patient in the graph, we extract radiomic features from a segmentation network that also serves as a latent image feature encoder. Together with clinical patient data like vital signs, demographics, and lab results, these modalities are combined into a multimodal representation of each patient. This feature extraction is trained end-to-end with an image-based Graph Attention Network to process the population graph and predict the COVID-19 patient outcomes: admission to ICU, need for ventilation, and mortality. To combine multiple modalities, radiomic features are extracted from chest CTs using a segmentation neural network. Results on a dataset collected in Klinikum rechts der Isar in Munich, Germany and the publicly available iCTCF dataset show that our approach outperforms single modality and non-graph baselines. Moreover, our clustering and graph attention increases understanding of the patient relationships within the population graph and provides insight into the network's decision-making process.

  View details for DOI 10.1038/s41598-023-46625-8

  View details for PubMedID 37945590

  View details for PubMedCentralID 7869614

• Investigating pulse-echo sound speed estimation in breast ultrasound with deep learning. Ultrasonics Simson, W. A., Paschali, M., Sideri-Lampretsa, V., Navab, N., Dahl, J. J. 2023; 137: 107179

  Abstract

  Ultrasound is an adjunct tool to mammography that can quickly and safely aid physicians in diagnosing breast abnormalities. Clinical ultrasound often assumes a constant sound speed to form diagnostic B-mode images. However, the components of breast tissue, such as glandular tissue, fat, and lesions, differ in sound speed. Given a constant sound speed assumption, these differences can degrade the quality of reconstructed images via phase aberration. Sound speed images can be a powerful tool for improving image quality and identifying diseases if properly estimated. To this end, we propose a supervised deep-learning approach for sound speed estimation from analytic ultrasound signals. We develop a large-scale simulated ultrasound dataset that generates representative breast tissue samples by modeling breast gland, skin, and lesions with varying echogenicity and sound speed. We adopt a fully convolutional neural network architecture trained on a simulated dataset to produce an estimated sound speed map. The simulated tissue is interrogated with a plane wave transmit sequence, and the complex-value reconstructed images are used as input for the convolutional network. The network is trained on the sound speed distribution map of the simulated data, and the trained model can estimate sound speed given reconstructed pulse-echo signals. We further incorporate thermal noise augmentation during training to enhance model robustness to artifacts found in real ultrasound data. To highlight the ability of our model to provide accurate sound speed estimations, we evaluate it on simulated, phantom, and in-vivo breast ultrasound data.

  View details for DOI 10.1016/j.ultras.2023.107179

  View details for PubMedID 37939413

• Interactive Segmentation for COVID-19 Infection Quantification on Longitudinal CT Scans IEEE ACCESS Foo, M., Kim, S., Paschali, M., Goli, L., Burian, E., Makowski, M., Braren, R., Navab, N., Wendler, T. 2023; 11: 77596-77607

  View details for DOI 10.1109/ACCESS.2023.3297506

  View details for Web of Science ID 001041931100001

• Self-supervised Learning for Physiologically-Based Pharmacokinetic Modeling in Dynamic PET De Benetti, F., Simson, W., Paschali, M., Sari, H., Rominger, A., Shi, K., Navab, N., Wendler, T., Madabhushi, A., Greenspan, H., Mousavi, P., Salcudean, S., Duncan, J., Syeda-Mahmood, T., Taylor, R. SPRINGER INTERNATIONAL PUBLISHING AG. 2023: 290-299

  View details for DOI 10.1007/978-3-031-43907-0_28

  View details for Web of Science ID 001109628700028

• Bridging the Gap between Deep Learning and Hypothesis-Driven Analysis via Permutation Testing. PRedictive Intelligence in MEdicine. PRIME (Workshop) Paschali, M., Zhao, Q., Adeli, E., Pohl, K. M. 2022; 13564: 13-23

  Abstract

  A fundamental approach in neuroscience research is to test hypotheses based on neuropsychological and behavioral measures, i.e., whether certain factors (e.g., related to life events) are associated with an outcome (e.g., depression). In recent years, deep learning has become a potential alternative approach for conducting such analyses by predicting an outcome from a collection of factors and identifying the most "informative" ones driving the prediction. However, this approach has had limited impact as its findings are not linked to statistical significance of factors supporting hypotheses. In this article, we proposed a flexible and scalable approach based on the concept of permutation testing that integrates hypothesis testing into the data-driven deep learning analysis. We apply our approach to the yearly self-reported assessments of 621 adolescent participants of the National Consortium of Alcohol and Neurodevelopment in Adolescence (NCANDA) to predict negative valence, a symptom of major depressive disorder according to the NIMH Research Domain Criteria (RDoC). Our method successfully identifies categories of risk factors that further explain the symptom.

  View details for DOI 10.1007/978-3-031-16919-9_2

  View details for PubMedID 36342897

  View details for PubMedCentralID PMC9632755

• Detecting negative valence symptoms in adolescents based on longitudinal self-reports and behavioral assessments. Journal of affective disorders Paschali, M., Kiss, O., Zhao, Q., Adeli, E., Podhajsky, S., Muller-Oehring, E. M., Gotlib, I. H., Pohl, K. M., Baker, F. C. 2022

  Abstract

  BACKGROUND: Given the high prevalence of depressive symptoms reported by adolescents and associated risk of experiencing psychiatric disorders as adults, differentiating the trajectories of the symptoms related to negative valence at an individual level could be crucial in gaining a better understanding of their effects later in life.METHODS: A longitudinal deep learning framework is presented, identifying self-reported and behavioral measurements that detect the depressive symptoms associated with the Negative Valence System domain of the NIMH Research Domain Criteria (RDoC).RESULTS: Applied to the annual records of 621 participants (age range: 12 to 17 years) of the National Consortium on Alcohol and Neurodevelopment in Adolescence (NCANDA), the deep learning framework identifies predictors of negative valence symptoms, which include lower extraversion, poorer sleep quality, impaired executive control function and factors related to substance use.LIMITATIONS: The results rely mainly on self-reported measures and do not provide information about the underlying neural correlates. Also, a larger sample is required to understand the role of sex and other demographics related to the risk of experiencing symptoms of negative valence.CONCLUSIONS: These results provide new information about predictors of negative valence symptoms in individuals during adolescence that could be critical in understanding the development of depression and identifying targets for intervention. Importantly, findings can inform preventive and treatment approaches for depression in adolescents, focusing on a unique predictor set of modifiable modulators to include factors such as sleep hygiene training, cognitive-emotional therapy enhancing coping and controllability experience and/or substance use interventions.

  View details for DOI 10.1016/j.jad.2022.06.002

  View details for PubMedID 35688394

• OperA: Attention-Regularized Transformers for Surgical Phase Recognition Czempiel, T., Paschali, M., Ostler, D., Kim, S., Busam, B., Navab, N., DeBruijne, M., Cattin, P. C., Cotin, S., Padoy, N., Speidel, S., Zheng, Y., Essert, C. SPRINGER INTERNATIONAL PUBLISHING AG. 2021: 604-614

  View details for DOI 10.1007/978-3-030-87202-1_58

  View details for Web of Science ID 000712021400058

• Rethinking Ultrasound Augmentation: A Physics-Inspired Approach Tirindelli, M., Eilers, C., Simson, W., Paschali, M., Azampour, M., Navab, N., DeBruijne, M., Cattin, P. C., Cotin, S., Padoy, N., Speidel, S., Zheng, Y., Essert, C. SPRINGER INTERNATIONAL PUBLISHING AG. 2021: 690-700

  View details for DOI 10.1007/978-3-030-87237-3_66

  View details for Web of Science ID 000712019200066

• Longitudinal Quantitative Assessment of COVID-19 Infection Progression from Chest CTs Kim, S., Goli, L., Paschali, M., Khakzar, A., Keicher, M., Czempiel, T., Burian, E., Braren, R., Navab, N., Wendler, T., DeBruijne, M., Cattin, P. C., Cotin, S., Padoy, N., Speidel, S., Zheng, Y., Essert, C. SPRINGER INTERNATIONAL PUBLISHING AG. 2021: 273-282

  View details for DOI 10.1007/978-3-030-87234-2_26

  View details for Web of Science ID 000712024400026

• Ultrasound-Guided Robotic Navigation with Deep Reinforcement Learning Hase, H., Azampour, M., Tirindelli, M., Paschali, M., Simson, W., Fatemizadeh, E., Navab, N., IEEE IEEE. 2020: 5534-5541

  View details for DOI 10.1109/IROS45743.2020.9340913

  View details for Web of Science ID 000714033803042

• SIGNAL CLUSTERING WITH CLASS-INDEPENDENT SEGMENTATION Gasperini, S., Paschali, M., Hopke, C., Wittmann, D., Navab, N., IEEE IEEE. 2020: 3982-3986

  View details for Web of Science ID 000615970404046

• Manifold Exploring Data Augmentation with Geometric Transformations for Increased Performance and Robustness Paschali, M., Simson, W., Roy, A., Goebl, R., Wachinger, C., Navab, N., Chung, A. C., Gee, J. C., Yushkevich, P. A., Bao, S. SPRINGER INTERNATIONAL PUBLISHING AG. 2019: 517-529

  View details for DOI 10.1007/978-3-030-20351-1_40

  View details for Web of Science ID 000493380900040

• 3DQ: Compact Quantized Neural Networks for Volumetric Whole Brain Segmentation Paschali, M., Gasperini, S., Roy, A., Fang, M., Navab, N., Shen, D., Liu, T., Peters, T. M., Staib, L. H., Essert, C., Zhou, S., Yap, P. T., Khan, A. SPRINGER INTERNATIONAL PUBLISHING AG. 2019: 438-446

  View details for DOI 10.1007/978-3-030-32248-9_49

  View details for Web of Science ID 000548733600049

• Generalizability vs. Robustness: Investigating Medical Imaging Networks Using Adversarial Examples Paschali, M., Conjeti, S., Navarro, F., Navab, N., Frangi, A. F., Schnabel, J. A., Davatzikos, C., AlberolaLopez, C., Fichtinger, G. SPRINGER INTERNATIONAL PUBLISHING AG. 2018: 493-501

  View details for DOI 10.1007/978-3-030-00928-1_56

  View details for Web of Science ID 000477770600056