About

Amanda Finegold Swain, MD, is an Assistant Professor of Clinical Family Medicine and Community Health at the University of Pennsylvania's Perelman School of Medicine. She is a physician at Penn Presbyterian Medical Center (Penn Family Care) and serves as the Faculty Director of UME Medical Humanities at the Perelman School of Medicine. Dr. Swain is also an instructor in Narrative Medicine within the Penn Masters of Bioethics program. Her professional focus includes medical humanities and narrative medicine, emphasizing the intersections of medicine, storytelling, and patient care. She has a background in anthropology from Brandeis University and completed her medical degree at Mount Sinai School of Medicine. Her work involves providing education and resources related to behavioral sleep medicine and engaging in community outreach and global initiatives.

Research topics

• Biology
• Computer Science
• Artificial Intelligence
• Chemistry
• Physics
• Biochemistry
• Nuclear magnetic resonance
• Neuroscience
• Genetics
• Molecular biology

Selected publications

• An imaging biomarker to detect non-glucogenic shift in brain energy metabolism in Alzheimer’s disease

  Journal of Translational Medicine · 2026-05-20

  articleOpen access

  Cerebral glucose hypometabolism in Alzheimer’s disease (AD) leads to enhanced metabolism of fatty acids (FAs) and branched-chain amino acids (BCAAs) as a compensatory mechanism. While there have been some 13C labeled studies investigating the metabolism of FAs and BCCAs, their clinical translation is challenging. In this study, we investigated the potential of measuring neurometabolic perturbations through macromolecular signal at 0.9 ppm (MM09) in proton magnetic resonance (1H MR) spectrum. This signal represents a composite macromolecular signal with contributions from lipids and BCAA associated methyl resonances and may be sensitive to metabolic alterations occurring during glucose hypometabolism in AD. MM09 levels were measured from localized 1H MR spectra in the hippocampus and thalamus/hypothalamus of male and female APPNL−F/NL−F (AD) mice. In addition, the levels of glutamate in these regions were also recorded as it is known to be reduced under glucose hypometabolism in AD. We further studied the metabolic association of MM09 with glutamate in Pearson correlation plots. To find the statistical significance of difference two-way ANOVA analysis with post-hoc Tukey HSD tests were used. Male AD mice exhibited significantly reduced MM09 (15.42 ± 1.32 vs. 16.93 ± 1.15 mM; p = 0.008) and glutamate levels (15.27 ± 1.65 vs. 17.24 ± 1.21 mM; p = 0.004) in the hippocampus. Female AD mice did not show any changes in glutamate or MM09 levels. MM09 also showed a strong positive correlation with glutamate (R = 0.74; p < 0.0001). The observed reductions in MM09 and glutamate in male AD mice are consistent with neurometabolic alterations associated with impaired glucose metabolism, whereas the absence of such changes in female AD mice may reflect sex-specific metabolic resilience. The strong association between MM09 and glutamate suggests that MM09 may capture neurochemical changes linked to metabolic adaptations in AD. Because the MM09 resonance occurs in a relatively uncrowded region of the 1H MR spectrum, it may represent a promising spectroscopic marker for investigating metabolic shifts in AD and warrants further evaluation in clinical studies.

  PublisherDOI

• Unsupervised anomaly detection for tumor delineation in a preclinical model of glioblastoma using CEST MRI

  bioRxiv (Cold Spring Harbor Laboratory) · 2026-02-19

  articleOpen access1st author

  Introduction: vivo assessment of metabolic and macromolecular information through a Z-spectrum. CEST may provide insight into metabolic changes present in early-stage disease that are not visible in routine clinical imaging, thereby improving tumor delineation. In this work, we use an unsupervised anomaly detection (UAD) strategy to learn the distribution of features present in Z-spectra of healthy tissue and capture their deviations in pathology, foregoing the need for extensive labels. The approach leverages the metabolic information provided by CEST to improve the detection and delineation of glioblastoma and inform further treatment planning. Methods: A 1D convolutional autoencoder (CAE) was implemented to reconstruct Z-spectra from individual tissue voxels. The network was trained on Z-spectra acquired at 9.4T from healthy Sprague-Dawley rats and tested on data acquired from F98 glioma-bearing rats post Gd-administration. For baseline comparisons, Isolation Forest and Local Outlier Factor, which have shown success in anomaly detection, were implemented. For the CAE, our anomaly score was determined to be the mean squared reconstruction error. To facilitate clinical translation and evaluate the robustness of our model for under sampled Z-spectra, acceleration factors of 2x and 7x were performed with two sampling schemes: uniformly skipping frequency offsets and selecting offsets based on feature importance identified by Shapley value analysis and Integrated Gradients (IG). Binarization was performed by determining an optimal anomaly threshold, followed by comparison to ground truth tumor masks. Metrics related to model performance were assessed for baseline anomaly detectors on the fully sampled dataset and for the CAE on fully and under sampled datasets. Results: The best baseline anomaly detector was Isolation Forest, with an ROC-AUC of 0.967 and an F1-score of 0.584. Our method, the CAE, accurately reconstructed Z-spectral features, achieving Dice scores of up to 0.72 and outperforming the baseline model with an ROC-AUC of 0.968 and F1-score of 0.642. This model performance remained robust across sampling schemes and acceleration factors, with ROC-AUCs of ~0.96 and similar Dice scores (up to 0.7). Feature importance analysis indicated that offsets in the range of ±3.0 to 5.0ppm contributed most to the anomaly score. Discussion: This study successfully demonstrated a UAD pipeline utilizing the Z-spectrum from CEST MRI for metabolically informed tumor delineation. The framework captures biochemical deviations that may precede or extend beyond morphologic abnormalities, enabling sensitive detection of tumor regions and intra-tumoral heterogeneity that previous methods may fail to capture. The offsets from the feature analysis indicated a strong contribution from the magnetization transfer (MT) pool to the spectral deviations captured by the model, with additional contributions from relayed nuclear Overhauser effect (rNOE) and amide proton transfer (APT). Model robustness with under sampling further highlights the pipeline's potential in accelerated acquisitions, thus improving clinical practicality. While there is a need for validation on larger cohorts and clinical datasets, the current results demonstrate that this label-free, Z-spectral anomaly mapping can serve as an interpretable and scalable tool for monitoring tumor heterogeneity and progression, with potential applicability to other diffuse or metabolically subtle pathologies.

  PublisherOA PDFDOI

• Contrast-induced changes in chemical exchange saturation transfer MRI differentiate tumor progression from pseudoprogression

  bioRxiv (Cold Spring Harbor Laboratory) · 2026-05-05

  articleOpen access

  Tumor pseudo-progression (PsP) refers to an initial increase in tumor size or the appearance of new lesions. These pseudo-progressive lesions are predominantly composed of infiltrative inflammatory cells, such as macrophages. This phenomenon commonly occurs in patients undergoing radiation therapy or immunotherapy and typically indicates a positive treatment response. However, it often leads to premature treatment cessation due to misinterpretation as disease progression. Non-invasive imaging biomarkers capable of distinguishing pseudo-progression from true progression would greatly aid in treatment decision-making. In our preliminary study, we explored the potential of gadoterate meglumine (Gd-DOTA, a macrocyclic Gd-contrast) in combination with amine chemical-exchange saturation transfer (amine-CEST) imaging to differentiate tumor from radiation necrosis by assessing Gd-DOTA uptake by infiltrating immune cells, such as macrophages. To evaluate whether amine-CEST, in combination with Gd-DOTA, can differentiate macrophages from cancer cells, we incubated them with Gd-DOTA for 30 minutes. Subsequently, the cells were processed, and amine-CEST imaging was performed on a 9.4 Tesla preclinical scanner. Upon treatment with Gd-DOTA, we did not observe a significant change in amine-CEST contrast in F98 cells compared with untreated cells, whereas treated macrophages exhibited a marked decrease (~40%) in amine-CEST signal compared with untreated macrophages. This reduction in signal was attributed to the uptake of Gd-DOTA by macrophages, which notably shortened water T1 relaxation, thereby quenching the amine-CEST signal. Conversely, cancer cells showed no appreciable change in the amine-CEST signal, indicating no Gd-DOTA uptake. Furthermore, to validate that T1 shortening influences amine-CEST signal, cancer cells were also treated with manganese chloride (MnCl2) for 30 minutes. The uptake of MnCl2 by cancer cells similarly induced T1 shortening, as observed in macrophages, resulting in a decrease in the amine-CEST signal from these cells. Next, we performed the amin-CEST imaging on F98 tumor-bearing rats and radiation necrotic rats. Post-injection with Gd-DOTA showed no appreciable change in the amine-CEST contrast in the tumor-bearing rat, whereas a significant decrease in contrast was observed in the radiation necrotic rat. This further demonstrates that no change in the amine-CEST contrast in tumor-bearing rats is due to cancer cells failing to take up Gd-DOTA. The decrease in amine-CEST contrast in radiation-treated rats reflects the uptake of Gd-DOTA by macrophages infiltrating the radiation-necrotic regions. This straightforward imaging approach holds promise for clinical translation. It offers a novel method for characterizing pseudo-progressive lesions and monitoring diverse treatment responses in cancer patients using standard clinical scanners.

  PublisherOA PDFDOI

• In vivo imaging of glutamate uncovers the neuroprotective effects of nicotinamide riboside on excitotoxicity in an Alzheimer’s mouse model

  Alzheimer s Research & Therapy · 2026-01-30

  articleOpen access1st author

  Nicotinamide adenine dinucleotide (NAD+) precursors, such as nicotinamide riboside (NR), have gained interest as potential therapeutics for alleviating Alzheimer’s disease (AD) pathology. Chemical exchange saturation transfer (CEST) magnetic resonance imaging (MRI) can provide insights into the effects of NR on AD by virtue of its sensitivity to monitoring the metabolic status of tissue in vivo. This study used glutamate-weighted CEST (GluCEST) MRI to monitor glutamate-associated metabolic changes following NR treatment in the 5xFAD mouse model of AD. Drinking water was supplemented with NR or provided as is to animals over the course of expected disease progression prior to imaging experiments. Following imaging, an immunohistochemical assay to monitor the expression of glial fibrillary acidic protein (GFAP) and ionized calcium-binding adaptor molecule 1 (Iba1) was performed to assess the extent of neuroinflammatory glial responses. A two-way ANCOVA with interaction was performed for statistical analysis of both CEST and IHC data. Results from GluCEST revealed significantly higher glutamate levels in the hippocampal dentate gyrus of AD mice compared to WT, with a significant reduction following treatment. GFAP staining mirrored this trend, implicating reactive astrogliosis as a mechanism for elevated glutamate. Similar patterns were observed in the cerebral peduncles, a white matter bundle, in which GFAP and Iba1 supported GluCEST findings and suggested neuroinflammation in axonal tracts. Our findings are in concordance with studies reporting elevated glutamate associated with reactive gliosis and morphological changes disrupting glutamate imbalance. Interestingly, NR restores glutamate homeostasis and alleviates neuroinflammatory processes, thus rescuing tissue from excitotoxic insults. Overall, this study demonstrates the potential of NR to mitigate glutamate-driven excitotoxicity in AD pathology, and highlights GluCEST as a sensitive in vivo, clinically translatable biomarker for neuroinflammation and excitotoxicity.

  PublisherDOI

• In Vivo Brain B₁⁺ Inhomogeneity Correction and NOE Image Enhancement at 7 T via Flexible Metasurfaces

  NMR in Biomedicine · 2025-04-02 · 3 citations

  articleOpen access

  ABSTRACT Nuclear Overhauser effect (NOE) MRI has been used for in vivo brain imaging to assess lipid and protein composition and benefits from 7 T field strengths due to the larger chemical shift dispersion. However, a continuing challenge is signal drop off observed in regions such as the medial temporal lobes due to “standing wave” effects from shorter radiofrequency (RF) wavelengths at ultra‐high fields. 2D periodic unit cell metasurfaces have been a promising approach for providing improvements in anatomical imaging but have not yet been evaluated in chemical exchange saturation transfer (CEST)‐based sequences. Here, we report the use of metasurfaces for enhancement of NOE imaging as well as for improvement of Lorentzian line fitting of full Z‐spectrum data. 3D NOE image data, B 1 + maps, and B 0 maps were acquired on five healthy volunteers using a 7 T MRI system with and without metasurfaces positioned near the temporal lobes. A frequency offset range of −5 to +5 ppm with additional separate acquisitions of ±20 and ±100 ppm offset images. A five‐pool Lorentzian line fitting model was employed to fit and quantitatively compared magnetization transfer (MT), amide proton transfer (APT), amine, and relayed NOE (rNOE) metabolite pools. NOE MTR ‐weighted contrast maps were also calculated via Z‐spectrum asymmetry analysis. The metasurfaces globally enhanced the transmit efficiency within the imaging slab by approximately 9.6% and reduced B 1 + inhomogeneity by approximately 16.6% and increased transmit efficiency by 55.8% in the temporal lobes. Amplitude fit maps showed decreases in contrast magnitude ranging from 1 to 16% and changes in image uniformity ranging from a 4.3 decrease to a 34.7% increase, while NOE MTR ‐weighted contrast maps demonstrated similar changes. The results presented here demonstrate that metasurfaces can enhance CEST‐based techniques complementing previously reported benefits in anatomical imaging.

  PublisherOA PDFDOI

• A Comparative Study of CrCEST and PCrCEST for Mapping of Oxidative Phosphorylation in Exercised Muscle

  Proceedings on CD-ROM - International Society for Magnetic Resonance in Medicine. Scientific Meeting and Exhibition/Proceedings of the International Society for Magnetic Resonance in Medicine, Scientific Meeting and Exhibition · 2025-09-16

  article

  Motivation: This study compares the sensitivity and specificity of two CEST-based methods for assessing oxidative phosphorylation (OXPHOS) in participants' skeletal muscle and also shows that creatine in muscle is a fast-exchanging species. Goal(s): We compare the sensitivity and specificity of creatine CEST and phosphocreatine CEST for exercised calf muscle. Approach: We acquired z-spectra at multiple saturation parameters to compare creatine CEST and phosphocreatine CEST at 3.0T and 7.0T. We also acquired creatine CEST and phosphocreatine CEST from exercised calf muscle. Results: Our findings indicate that creatine rapidly exchanges in skeletal muscle. Additionally, creatine CEST demonstrates higher sensitivity and specificity than phosphocreatine CEST. Impact: OXPHOS deficiency in skeletal muscle is associated with primary mitochondrial disorders, muscle injury, cardiovascular disease, and diabetes mellitus. This study compares the sensitivity and specificity of two CEST-based methods for assessing OXPHOS.

  PublisherDOI

• NOE Imaging of Multiple Sclerosis Subjects at 7T Detects Diffuse Contrast Changes

  Proceedings on CD-ROM - International Society for Magnetic Resonance in Medicine. Scientific Meeting and Exhibition/Proceedings of the International Society for Magnetic Resonance in Medicine, Scientific Meeting and Exhibition · 2025-09-16

  article

  Motivation: While standard to the diagnosis of multiple sclerosis (MS), conventional structural MRI cannot provide detailed information on changes in lipid metabolism. Goal(s): To utilize NOE imaging to investigate changes between MS and healthy control subjects. Approach: NOE imaging was performed on 15 MS and 10 healthy subjects in conjunction with a multi-pool Lorentzian line fitting model to produce several contrasts including MT, APT, amine, and rNOE. Results: Statistically significant contrast decreases were observed in both the amine (15.3% in NAWM) and rNOE (11.4% in NAWM and 10.6% in NAGM) pools. Impact: This 7T NOE imaging method for patients with MS can provide complementary lipid metabolic information to standard structural imaging and can yield improved diagnostic outcomes for this patient population.

  PublisherDOI

• Magnetic Resonance Imaging for Improved Brain Tumor Detection

  bioRxiv (Cold Spring Harbor Laboratory) · 2025-05-11 · 1 citations

  preprintOpen access

  Abstract Precise demarcation of brain tumor boundaries is critical for optimizing treatment strategies and improving patient outcomes. In vivo characterization of tumor using PET/CT and MRI is clinical standard. PET/CT highlights the metabolic aspects of the tumor, while MRI provides information on functional, metabolic and structural changes. Even with technological advancements in both PET/CT and MRI, a method that can precisely delineate infiltrative tumor boundaries from normal-appearing brain regions (NABR) in vivo is still lacking. To address this limitation, we explored a relatively new MR imaging method, the Nuclear Overhauser Effect Magnetization Transfer Ratio (NOE MTR ), in conjunction with a gadolinium-based contrast agent (Gd-DOTA), to precisely delineate the tumor boundaries in a rat model of infiltrative gliosarcoma. NOE MTR imaging was performed in the rat model (n=5) before and after Gd-DOTA administration. The post-Gd-DOTA NOE MTR map was subtracted from the pre-Gd-DOTA map and compared with contrast-enhanced T 1 -weighted images and immuno-histological findings. The resulting NOE MTR difference map clearly highlighted both the tumor core and infiltrative boundaries, which was not discernible on the post-contrast T 1 -weighted images. The extended tumor boundaries observed on the NOE MTR difference map corroborated with the IHC image, which confirmed the presence of infiltrative tumor cells and macrophages in these regions. Guided by the NOE MTR difference map, regions of interest (ROI) were drawn to quantify NOE MTR signal changes in the tumor core, tumor boundaries, and NABR post-Gd-DOTA. Tumor core showed a significant ∼43% reduction in NOE MTR signal (plJ=lJ0.003), while the tumor periphery exhibited a moderate reduction of ∼10%, (plJ=lJ0.045). No appreciable change in was observed in the NABR (plJ=lJ0.371). In contrast, the post contrast T 1 -weighted signal changes in tumor core, tumor periphery and NABR were, 33.32% (p = 0.092), 3.8% (p = 0.478), and 8.7% (p = 0.464) respectively. These findings suggest that NOE MTR imaging provides enhanced tumor contrast, particularly at the infiltrative tumor margins, where conventional contrast enhanced T 1 -weighted MRI may underestimate tumor extent. Histological validation confirmed the presence of infiltrative tumor cells and macrophages in the tumor periphery, as highlighted by the NOE MTR difference map. Overall, NOE MTR imaging, in combination with Gd-DOTA administration, demonstrates superior delineation of brain tumor boundaries compared to conventional MRI. As NOE MTR imaging is a fast acquisition scan (under 10 minutes) and performed on standard 3 Tesla, it can be easily integrated into clinical protocols. By improving visualization of tumor infiltration and distinguishing tumor regions from NABR, NOE MTR imaging holds promise for advancing neuro-oncological diagnostics and treatment planning.

  PublisherOA PDFDOI

• Effects of nicotinamide riboside supplementation on glucose metabolism and lipid homeostasis in healthy mice

  Proceedings on CD-ROM - International Society for Magnetic Resonance in Medicine. Scientific Meeting and Exhibition/Proceedings of the International Society for Magnetic Resonance in Medicine, Scientific Meeting and Exhibition · 2025-09-16

  article1st authorCorresponding

  Motivation: The rising popularity of NAD+ precursors to alleviate metabolic stress in aging necessitates a close monitoring of tissue metabolic status. Goal(s): FINUTE and 2H-MRS are used monitor lipid homeostasis and glucose metabolism in the brains of healthy mice following supplementation of nicotinamide riboside (NR), a widely used NAD+ precursor. Approach: Mice administered saline or NR intraperitoneally for one week are imaged to assess lipid integrity. 2H-MRS is performed following administration of labeled glucose to monitor downstream metabolite labeling. Results: Mice administered NR show increased lipid integrity and improved glucose metabolism, determined through increased labeling of Glx (TCA) and lower labeling of lactate (glycolysis). Impact: Nicotinamide riboside has potential in improving cerebral glucose metabolism and lipid integrity as assessed by lipid-sensitive MRI and 2H-MR spectroscopy, necessitating further exploration of NR and other NAD+ precursors both clinically and in aging-related diseases using these methods.

  PublisherDOI

• Swin-Transformer based B0 inhomogeneity correction for GluCEST and NOE MRI

  Proceedings on CD-ROM - International Society for Magnetic Resonance in Medicine. Scientific Meeting and Exhibition/Proceedings of the International Society for Magnetic Resonance in Medicine, Scientific Meeting and Exhibition · 2025-09-16

  article

  Motivation: For B0 correction of GluCEST MRI, some deep learning-based algorithms have been developed to significantly accelerate the Z-spectrum calibration process. Goal(s): When applied to CEST imaging involving other metabolites, and to Nuclear Overhauser Effect (NOE) MRI, the performance of the model declined substantially. Our goal is to develop a new model that can handle different metabolites. Approach: To address this issue, we proposed a Swin-Transformer-based model designed to handle both NOE and Glutamate-weighted CEST MRI separately. Results: Preliminary results demonstrate strong performance on both GluCEST and NOE datasets, indicating the potential for a generalizable model applicable to other CEST agents. Impact: This success of the proposed method suggests that the Swin Transformer could potentially serve as a general model for B0 correction across various metabolites in a single model if sufficient data is available.

  PublisherDOI

Frequent coauthors

• Ravinder Reddy

  24 shared

• Ravi Prakash Reddy Nanga

  University of Pennsylvania

  17 shared

• Narayan Datt Soni

  14 shared

• Halvor Juul

  10 shared

• Paul Jacobs

  9 shared

• Blake Benyard

  8 shared

• John A. Detre

  University of Pennsylvania

  8 shared

• Neil Wilson

  University of Pennsylvania

  8 shared

Labs

• Amanda Finegold Swain LabPI

Education

• PhD Candidate, Bioengineering

  University of Pennsylvania

• B.S., Chemistry

  Emory University

  2021