Search

A method is presented to select the optimal time points at which to measure DCE-MRI signal intensities, leaving time in the MR exam for high-spatial resolution image acquisition.Simplicial complexes are generated from the Kety-Tofts model pharmacokinetic parameters KThe DCE-MRI data acquired in women with invasive breast cancer (N = 27) were used to retrospectively compare parameter maps fit to full and subsampled time courses. Simplicial complexes were generated for a fixed range of Kety-Tofts model parameters and for the parameter ranges weighted by estimates from the fully sampled data. The largest-area manifolds determined the optimal three time points for each case. Simulations were performed along with retrospectively subsampled data fits. The agreement was computed between the model parameters fit to three points and those fit to all points.The optimal three-point sample times were from the data-informed simplicial complex analysis and determined to be 65, 204, and 393 s after arrival of the contrast agent to breast tissue. In the patient data, tumor-median parameter values fit using all points and the three selected time points agreed with concordance correlation coefficients of 0.97 for KIt is possible to accurately estimate pharmacokinetic parameters from three properly selected time points inserted into a clinical DCE-MRI breast exam. This technique can provide guidance on when to capture images for quantitative data between high-spatial-resolution DCE-MRI images.

To implement physics-based regularization as a stopping condition in tuning an untrained deep neural network for reconstructing MR images from accelerated data.The ConvDecoder (CD) neural network was trained with a physics-based regularization term incorporating the spoiled gradient echo equation that describes variable-flip angle data. Fully-sampled variable-flip angle k-space data were retrospectively accelerated by factors of R = {8, 12, 18, 36} and reconstructed with CD, CD with the proposed regularization (CD + r), locally low-rank (LR) reconstruction, and compressed sensing with L1-wavelet regularization (L1). Final images from CD + r training were evaluated at the "argmin" of the regularization loss; whereas the CD, LR, and L1 reconstructions were chosen optimally based on ground truth data. The performance measures used were the normalized RMS error, the concordance correlation coefficient, and the structural similarity index.The CD + r reconstructions, chosen using the stopping condition, yielded structural similarity indexs that were similar to the CD (p = 0.47) and LR structural similarity indexs (p = 0.95) across R and that were significantly higher than the L1 structural similarity indexs (p = 0.04). The concordance correlation coefficient values for the CD + r TThe use of an untrained neural network together with a physics-based regularization loss shows promise as a measure for determining the optimal stopping point in training without relying on fully-sampled ground truth data.

Introduction. X-ray mammography is the standard-of-care screening protocol for breast cancer due to its low cost, widespread availability, and greater specificity. While magnetic resonance imaging (MRI) has lower specificity, it has superior tissue contrast, and dynamic contrast-enhanced (DCE) MRI has been shown to increase MRI specificity due to its ability to estimate tissue vascular properties. Because of the increased expense and scan time for MRI, there is an ongoing effort to develop abbreviated breast MRI scans for screening high-risk patients at a lower cost without compromising quantitative information. Here we investigate the effects of the limited dynamic time course afforded by an abbreviated breast MRI exam on the quantitative tissue information computed from retrospectively abbreviated DCE-MRI data. Methods. Data acquisition. We evaluate the error in perfusion model parameters computed using DCE-MRI time courses sourced from three datasets acquired with very different temporal resolutions (dt). These datasets are the ACRIN 6883 multi-site breast trial (dt = 15 s), The University of Texas at Austin (UTA) neoadjuvant therapy study (dt = 7.3 s), and The University of Chicago (UC) ultra-fast breast DCE-MRI study (dt = 3.4 s). Ten representative patients are chosen from each of these datasets for a total of 30 DCE-MRI patient datasets. All 30 full-time courses (FTCs) are retrospectively truncated into a series of abbreviated-time courses (ATCs). An ATC containing the first n post-contrast injection time points of a DCE-MRI time course is referred to as “ATC n.” For the ACRIN dataset, n is the inclusive set of integers from 7 to 18; and, similarly, for the UC data, n is the inclusive set of integers from 12 to 23. For the UTA dataset, n is the inclusive set ranging from 13 through 53, incrementing by eight. Data Analysis. The groups of FTCs and ATCs are analyzed by one of three models, determined by the specifics of the acquisition details of each dataset. The standard Kety-Tofts (SKT) model is fit to the UTA time courses, the reference-region (RR) model is fit to the ACRIN time courses, and the Patlak model is fit to the UC time courses. The volume transfer constant (Ktrans) characterizes tissue enhancement in all three models; whereas, the extravascular/extracellular volume fraction (ve) is specific to the SKT and RR models, and the plasma volume fraction (vp) is specific to the Patlak model. Due to the absence of an arterial input function (AIF) for the ACRIN dataset, the RR model was most appropriate with the pectoral muscle serving as the reference region. The UTA dataset has a population AIF and is thus able to be modeled by the SKT. Lastly, the UC dataset does not capture the tissue washout necessary for ve estimation, so the Patlak model was chosen for analyzing tissue enhancement. Results and Conclusion. The longest ATCs of 4.5, 6.4, and 1.3 min. yielded average errors of 9.1%, 7%, and 3.6% in Ktrans for the ACRIN, UTA, and UC datasets, respectively; and the shortest ATCs of 2, 1.6, and 0.6 min. yielded higher average errors of 24.2%, 22.8%, and 65% in Ktrans, respectively. This is expected from simulations as even the most aggressive ATCs did not substantially exclude the tissue enhancement, characterized by Ktrans. Errors in ve were higher overall since shorter ATCs exclude much of the washout phase. As Ktrans has been shown to discriminate between malignant and benign lesions in full length DCE-MRI scans, it is promising that tolerable errors are observed in the abbreviated MRI setting. There is potential for implementing quantitative abbreviated DCE-MRI scans in the clinic for enhanced diagnostic specificity while freeing up scan time for additional imaging sequences without interfering with standard-of-care image acquisition. Citation Format: Kalina P Slavkova, Julie C DiCarlo, Anum K Syed, Chengyue Wu, John Virostko, Anna G Sorace, Thomas E Yankeelov. Characterizing errors in perfusion model parameters derived from retrospectively abbreviated quantitative DCE-MRI data [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PS3-26.

Digital twins employ mathematical and computational models to virtually represent a physical object (e.g., planes and human organs), predict the behavior of the object, and enable decision-making to optimize the future behavior of the object. While digital twins have been widely used in engineering for decades, their applications to oncology are only just emerging. Due to advances in experimental techniques quantitatively characterizing cancer, as well as advances in the mathematical and computational sciences, the notion of building and applying digital twins to understand tumor dynamics and personalize the care of cancer patients has been increasingly appreciated. In this review, we present the opportunities and challenges of applying digital twins in clinical oncology, with a particular focus on integrating medical imaging with mechanism-based, tissue-scale mathematical modeling. Specifically, we first introduce the general digital twin framework and then illustrate existing applications of image-guided digital twins in healthcare. Next, we detail both the imaging and modeling techniques that provide practical opportunities to build patient-specific digital twins for oncology. We then describe the current challenges and limitations in developing image-guided, mechanism-based digital twins for oncology along with potential solutions. We conclude by outlining five fundamental questions that can serve as a roadmap when designing and building a practical digital twin for oncology and attempt to provide answers for a specific application to brain cancer. We hope that this contribution provides motivation for the imaging science, oncology, and computational communities to develop practical digital twin technologies to improve the care of patients battling cancer.

Background The purpose of this study was to determine whether advanced quantitative magnetic resonance imaging (MRI) can be deployed outside of large, research-oriented academic hospitals and into community care settings to predict eventual pathological complete response (pCR) to neoadjuvant therapy (NAT) in patients with locally advanced breast cancer. Methods Patients with stage II/III breast cancer (N = 28) were enrolled in a multicenter study performed in community radiology settings. Dynamic contrast-enhanced (DCE) and diffusion-weighted (DW)-MRI data were acquired at four time points during the course of NAT. Estimates of the vascular perfusion and permeability, as assessed by the volume transfer rate (Ktrans) using the Patlak model, were generated from the DCE-MRI data while estimates of cell density, as assessed by the apparent diffusion coefficient (ADC), were calculated from DW-MRI data. Tumor volume was calculated using semi-automatic segmentation and combined with Ktrans and ADC to yield bulk tumor blood flow and cellularity, respectively. The percent change in quantitative parameters at each MRI scan was calculated and compared to pathological response at the time of surgery. The predictive accuracy of each MRI parameter at different time points was quantified using receiver operating characteristic curves. Results Tumor size and quantitative MRI parameters were similar at baseline between groups that achieved pCR (n = 8) and those that did not (n = 20). Patients achieving a pCR had a larger decline in volume and cellularity than those who did not achieve pCR after one cycle of NAT (p Ktrans, ADC, cellularity, and bulk tumor flow from baseline (pre-treatment) were all significantly greater (p Conclusions Quantitative analysis of DCE-MRI and DW-MRI can be implemented in the community care setting to accurately predict the response of breast cancer to NAT. Dissemination of quantitative MRI into the community setting allows for the incorporation of these parameters into the standard of care and increases the number of clinical community sites able to participate in novel drug trials that require quantitative MRI.

This study characterizes the error that results when performing quantitative analysis of abbreviated dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) data of the breast with the Standard Kety–Tofts (SKT) model and its Patlak variant. More specifically, we used simulations and patient data to determine the accuracy with which abbreviated time course data could reproduce the pharmacokinetic parameters, Ktrans (volume transfer constant) and ve (extravascular/extracellular volume fraction), when compared to the full time course data. SKT analysis of simulated abbreviated time courses (ATCs) based on the imaging parameters from two available datasets (collected with a 3T MRI scanner) at a temporal resolution of 15 s (N = 15) and 7.23 s (N = 15) found a concordance correlation coefficient (CCC) greater than 0.80 for ATCs of length 3.0 and 2.5 min, respectively, for the Ktrans parameter. Analysis of the experimental data found that at least 90% of patients met this CCC cut-off of 0.80 for the ATCs of the aforementioned lengths. Patlak analysis of experimental data found that 80% of patients from the 15 s resolution dataset and 90% of patients from the 7.27 s resolution dataset met the 0.80 CCC cut-off for ATC lengths of 1.25 and 1.09 min, respectively. This study provides evidence for both the feasibility and potential utility of performing a quantitative analysis of abbreviated breast DCE-MRI in conjunction with acquisition of current standard-of-care high resolution scans without significant loss of information in the community setting.

Background: Early and accurate prediction of response to neoadjuvant therapy (NAT) would empower personalization of breast cancer treatment regimens based on expected response. Noninvasive, quantitative dynamic contrast-enhanced (DCE-MRI) and diffusion-weighted magnetic resonance imaging (DW-MRI), when performed during the course of NAT, can accurately predict the ultimate pathological response. While these techniques have been incorporated into clinical trials in the academic setting, they have not yet been translated to the community setting, where most cancer patients receive their care. Implementation of quantitative MRI in the community setting widely expands the potential impact it can provide by 1) allowing community settings to participate in clinical trials that require quantitative MRI, and 2) advancing quantitative MRI towards standard-of-care for prediction of response in breast cancer. Methods: Women with locally advanced breast cancer (N = 28) were imaged four times during the course of NAT: 1) prior to the start of NAT, 2) after 1 cycle of NAT, 3) after 2-4 cycles of NAT, and 4) 1 cycle after MRI #3. Imaging data was acquired on 3T Siemens Skyra scanners equipped with breast coils and sited in a community hospital and radiology clinic, respectively. DW-MRI and DCE-MRI were acquired over 10 slices of 5 mm thickness. DW-MRI was acquired with diagonal monopolar diffusion-encoding gradients with b-values of 0, 200, and 800 s/mm2 in a total scan time of 1 minute 39 seconds. Voxel wise tumor cellularity was quantified using the apparent diffusion coefficient (ADC). For DCE-MRI, a gadolinium-based contrast agent was administered intravenously at 2 mL/sec after the acquisition of baseline scans. DCE-MRI data was acquired dynamically with a temporal resolution of 7.27 seconds for a total acquisition time of 8 minutes. The volume transfer constant Ktrans was calculated using Patlak analysis of DCE-MRI data to characterize the tumor vasculature. The tumor was semi-automatically segmented using a manually drawn region of interest followed by fuzzy c-means clustering of DCE-MRI data to identify a functional tumor volume. Measurements of tumor volume were combined with both ADC and Ktrans to yield metrics of tumor cellularity and bulk tumor flow, respectively. Results: Women who achieved pathological complete response at the time of surgery (pCR; n=8) displayed significantly different treatment-induced changes in MRI-derived tumor parameters versus women who did not achieve pCR (non-pCR, n=20). After 1 cycle of NAT, women who achieved pCR had smaller functional tumor volume and lower cellularity (p < 0.05) than non-pCR study participants. At the third and fourth MRI, tumor volume, ADC, Ktrans, cellularity, and bulk tumor flow were all significantly different between the pCR and non-pCR cohorts (p < 0.05). Of note, longest tumor diameter was not predictive of pCR at any time point in this study. Conclusions: This study demonstrates that quantitative DCE- and DW-MRI can be implemented successfully in community care facilities within standard-of-care settings for imaging locally advanced breast cancer. Metrics extracted from the change in DW-MRI (ADC) and DCE-MRI (Ktrans) can accurately predict pathological complete response to neoadjuvant therapy and may be more sensitive to tumor response than the RECIST criteria. Furthermore, incorporating quantitative metrics with tumor volume further increases the ability to predict pathological response to NAT in locally advanced breast cancer. While there are still challenges to address to effectively implement these quantitative metrics into the clinical workflow, this study is first in its kind to transition a decade’s worth of quantitative MRI advancements from academic settings into standard-of-care. Citation Format: John Virostko, Anna G Sorace, Kalina P Slavkova, Anum S Kazerouni, Angela M Jarrett, Julie C DiCarlo, Stefanie Woodard, Sarah Avery, Boone W Goodgame, Debra Patt, Thomas E Yankeelov. Quantitative multiparametric MRI predicts response to neoadjuvant therapy in the community setting [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P3-03-03.

Background: Standard of care (SOC) breast MRI exams typically acquire 4-7 frames of dynamic contrast-enhanced MRI (DCE-MRI) for cancer screening and staging. Post-contrast images depict lesion spiculations and boundaries to identify and characterize tumors. Pharmacokinetic (PK) analysis of DCE-MRI involves modeling blood flow to the lesion and surrounding tissue and has shown promise in diagnosis and prediction of therapeutic response. Currently, SOC DCE-MRI requires ~60-90 seconds per volume for images with sufficient quality and spatial resolution. However, PK analysis of DCE-MRI requires faster time course sampling. For this reason, PK modeling is limited to research scans with lower spatial resolution and higher temporal resolution. PK modeling would improve feedback of treatment response, and implementation in the SOC exam would increase imaging trial participation. In this study, we tested the estimation of Ktrans, a mixed perfusion and permeability PK parameter, from three images at optimal time points after contrast agent (CA) injection, and compared it to the Ktrans estimation from analysis of the full-length time course.. Methods: Women (N=23) with newly diagnosed invasive breast cancers who were eligible for neoadjuvant therapy (NAT) were scanned with a research MRI protocol as part of a treatment-monitoring study. Images acquired prior to the start of NAT were used. MRI was performed on 3.0T Siemens Skyra scanners at two sites with bilateral breast coils. The research protocol included ten sagittal slices centered about the primary tumor. The DCE-MRI images came from a fast sequence with 1.3 × 1.3 × 5.0 mm resolution acquired at 7.3 seconds per frame (66 frames total,) with a gadolinium-based CA injected one minute into the scan. A population arterial input function was used to implement a mathematical graph-based search of possible tissue CA concentration curves from the expected range of PK parameters. The search results gave a set of three optimized sub-sampled timepoints, Topt, from the full set of sample times, Tfull, at which to best sample the CA concentration curves to optimally estimate PK values. The imaging data was analyzed to find one parameter map from image times Tfull, and another from the subset of images at times Topt. The difference in Ktrans was computed at each parameter map voxel, and the concordance correlation coefficient (CCC) was computed per patient to determine agreement. The median Ktrans values were also compared for each patient. Results: The graph-based search of CA concentration curves found optimal times Topt of 37, 66, and 153 seconds after injection. The averaged values over all patients for median and maximum Ktrans from the original Tfull image set were 0.07 and 0.5 (min)-1. The average difference in Ktrans values between the Topt and Tfull sets was 0.02 (min)-1. When the median Ktrans values for each patient were compared, the average difference in median Ktrans values was 15% +/- 9%. The concordance correlation coefficients comparing the Topt and Tfull -sampled parameter maps for each patient were 0.89 +/- 0.12, showing high agreement. Discussion: This retrospective analysis suggests that it is possible to estimate PK parameters from a few properly selected post-contrast images inserted into a SOC DCE-MRI exam. The combination of optimal timing with fast acquisition techniques for high-resolution imaging could be used to provide quantitative data while preserving post-contrast images with the necessary spatial resolution for clinical reading. Importantly, the test images were acquired in the community setting with widely available MRI hardware, further indicating the potential for integration with SOC exams. Funding: NCI U24 CA226110, NCI U01 CA174706, NCI U01 CA142565, CPRIT RR160005 Citation Format: Julie C DiCarlo, Angela M Jarrett, Anum S Kazerouni, John Virostko, Anna G Sorace, Kalina P Slavkova, Debra Patt, Boone W Goodgame, Sarah Avery, Thomas E Yankeelov. Three timepoint pharmacokinetic modeling to incorporate within standard of care MRI breast exams [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P3-02-09.

Introduction. Dynamic contrast-enhanced (DCE) MRI provides quantitative information on tissue properties that enhances the specificity of breast cancer diagnosis; however, mammography remains the standard screening protocol due to its lower cost. There is a push to develop more accessible abbreviated breast MRI scans for screening high-risk patients without compromising diagnostic power. Here we analyze the effects of the limited dynamic time course afforded by an abbreviated breast MRI exam on the diagnostic performance of quantitative DCE-MRI. Methods and Results. We evaluate the ability of five quantitative measures to retrospectively differentiate malignant (N=21) and benign (N=24) lesions using DCE-MRI data acquired with 15 second temporal resolution in a cohort of 45 patients from the ACRIN 6883 multi-site breast trial. The first two measures are the volume transfer constant (Ktrans) estimated by, respectively, fitting the standard Kety-Tofts (STK) perfusion model and the Patlak approximation of the STK model (Patlak model) to patient data sets that have been truncated into a series of abbreviated-time courses (ATCs). An ATC is defined as containing the first n time points of a time course and is referred to as “ATC n.” For the first measure, n is the inclusive set of integers from 7 to 14, and, for the second measure, n is the set of 4 and 7 since the Patlak approximation only holds during the initial enhancement. For each patient, the fitting procedure provides Ktrans values for each voxel within the region of interest (ROI). The values are averaged and statistically evaluated for performance in discriminating malignant from benign tumors. For Ktrans, the maximum AUC of 0.61 was achieved using ATC 14 (i.e., 3.50 minutes), while for the Patlak model, the maximum AUC of 0.55 was achieved using ATC 7 (i.e., 1.75 minutes). The third measure is a modified median signal enhancement ratio (SER) computed for a series of four ATCs per patient, where n is now the inclusive set of integers 8 through 11. The SER is defined as (S1-S0)/(S2-S0), where S0 is the pre-contrast signal, S1 is the peak enhancement, and S2 is the signal at the last time point. Abbreviating the time course effectively shifts S2 to an earlier time point. For each ATC for each patient, the SERs are computed for all voxels within the ROI after which the median is computed. We calculate the AUC under the ROC curve to evaluate the diagnostic performance of each ATC-derived median SER. ATC 10 (i.e., 2.5 minutes) yields a maximum AUC of 0.79 among all ATCs. The fourth and fifth measures are, respectively, the area under the enhancement phase and the slope of the washout phase of the patient DCE time courses. The fourth measure is computed by numerically integrating the time course of each voxel within a patient’s ROI between the first and seventh time points and averaging the values. Similarly, for the fifth measure, the slope is calculated between the seventh and last time points within the ROI and averaged. We find that the fourth measure yields an AUC under the ROC curve of 0.76, and the fifth measure yields an AUC of 0.77. Discussion and Conclusion. The highest AUC from the two pharmacokinetic measures is 0.61, which suggests ineffective diagnostic ability in this data set. The median SER computed using ATC10 yields an AUC of 0.79, showing promise as a quantitative diagnostic tool in the abbreviated scan setting. Inter-site variability and the low temporal resolution of this data set may explain the relatively lower performance of measures one, two, four, and five as compared to measure three. It is necessary to repeat this analysis using a state-of-the-art acquisition of DCE-MRI data to determine the specificity of the remaining quantitative measures to imaging specifications. Citation Format: Kalina P Slavkova, Julie C DiCarlo, Anum K Syed, Chengyue Wu, John Virostko, Anna G Sorace, Thomas E Yankeelov. Investigating the feasibility of performing quantitative DCE-MRI in an abbreviated breast examination [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P6-02-04.

Receptor tyrosine kinases (RTKs) typically contain multiple autophosphorylation sites in their cytoplasmic domains. Once activated, these autophosphorylation sites can recruit downstream signaling proteins containing Src homology 2 (SH2) and phosphotyrosine-binding (PTB) domains, which recognize phosphotyrosine-containing short linear motifs (SLiMs). These domains and SLiMs have polyspecific or promiscuous binding activities. Thus, multiple signaling proteins may compete for binding to a common SLiM and vice versa. To investigate the effects of competition on RTK signaling, we used a rule-based modeling approach to develop and analyze models for ligand-induced recruitment of SH2/PTB domain-containing proteins to autophosphorylation sites in the insulin-like growth factor 1 (IGF1) receptor (IGF1R). Models were parameterized using published datasets reporting protein copy numbers and site-specific binding affinities. Simulations were facilitated by a novel application of model restructuration, to reduce redundancy in rule-derived equations. We compare predictions obtained via numerical simulation of the model to those obtained through simple prediction methods, such as through an analytical approximation, or ranking by copy number and/or KD value, and find that the simple methods are unable to recapitulate the predictions of numerical simulations. We created 45 cell line-specific models that demonstrate how early events in IGF1R signaling depend on the protein abundance profile of a cell. Simulations, facilitated by model restructuration, identified pairs of IGF1R binding partners that are recruited in anti-correlated and correlated fashions, despite no inclusion of cooperativity in our models. This work shows that the outcome of competition depends on the physicochemical parameters that characterize pairwise interactions, as well as network properties, including network connectivity and the relative abundances of competitors., Author summary Cells rely on networks of interacting biomolecules to sense and respond to environmental perturbations and signals. However, it is unclear how information is processed to generate appropriate and specific responses to signals, especially given that these networks tend to share many components. For example, receptors that detect distinct ligands and regulate distinct cellular activities commonly interact with overlapping sets of downstream signaling proteins. Here, to investigate the downstream signaling of a well-studied receptor tyrosine kinase (RTK), the insulin-like growth factor 1 (IGF1) receptor (IGF1R), we formulated and analyzed 45 cell line-specific mathematical models, which account for recruitment of 18 different binding partners to six sites of receptor autophosphorylation in IGF1R. The models were parameterized using available protein copy number and site-specific affinity measurements, and restructured to allow for network generation. We find that recruitment is influenced by the protein abundance profile of a cell, with different patterns of recruitment in different cell lines. Furthermore, in a given cell line, we find that pairs of IGF1R binding partners may be recruited in a correlated or anti-correlated fashion. We demonstrate that the simulations of the model have greater predictive power than protein copy number and/or binding affinity data, and that even a simple analytical model cannot reproduce the predicted recruitment ranking obtained via simulations. These findings represent testable predictions and indicate that the outputs of IGF1R signaling depend on cell line-specific properties in addition to the properties that are intrinsic to the biomolecules involved.