Thomas Niedermayr

All Publications

• Auto-delineation of treatment target volume for radiation therapy using large language model-aided multimodal learning. International journal of radiation oncology, biology, physics Rajendran, P., Chen, Y., Qiu, L., Niedermayr, T., Liu, W., Buyyounouski, M., Bagshaw, H., Han, B., Yang, Y., Kovalchuk, N., Gu, X., Hancock, S., Xing, L., Dai, X. 2024

  Abstract

  Artificial intelligence (AI)-aided methods have made significant progress in the auto-delineation of normal tissues. However, these approaches struggle with the auto-contouring of radiotherapy target volume. Our goal is to model the delineation of target volume as a clinical decision-making problem, resolved by leveraging large language model-aided multimodal learning approaches.A vision-language model, termed Medformer, has been developed, employing the hierarchical vision transformer as its backbone, and incorporating large language models to extract text-rich features. The contextually embedded linguistic features are seamlessly integrated into visual features for language-aware visual encoding through the visual language attention module. Metrics, including Dice similarity coefficient (DSC), intersection over union (IOU), and 95th percentile Hausdorff distance (HD95), were used to quantitatively evaluate the performance of our model. The evaluation was conducted on an in-house prostate cancer dataset and a public oropharyngeal carcinoma (OPC) dataset, totaling 668 subjects.Our Medformer achieved a DSC of 0.81 ± 0.10 versus 0.72 ± 0.10, IOU of 0.73 ± 0.12 versus 0.65 ± 0.09, and HD95 of 9.86 ± 9.77 mm versus 19.13 ± 12.96 mm for delineation of gross tumor volume (GTV) on the prostate cancer dataset. Similarly, on the OPC dataset, it achieved a DSC of 0.77 ± 0.11 versus 0.72 ± 0.09, IOU of 0.70 ± 0.09 versus 0.65 ± 0.07, and HD95 of 7.52 ± 4.8 mm versus 13.63 ± 7.13 mm, representing significant improvements (p < 0.05). For delineating the clinical target volume (CTV), Medformer achieved a DSC of 0.91 ± 0.04, IOU of 0.85 ± 0.05, and HD95 of 2.98 ± 1.60 mm, comparable to other state-of-the-art algorithms.Auto-delineation of the treatment target based on multimodal learning outperforms conventional approaches that rely purely on visual features. Our method could be adopted into routine practice to rapidly contour CTV/GTV.

  View details for DOI 10.1016/j.ijrobp.2024.07.2149

  View details for PubMedID 39117164

• Large Language Model-Augmented Auto-Delineation of Treatment Target Volume in Radiation Therapy. ArXiv Rajendran, P., Yang, Y., Niedermayr, T. R., Gensheimer, M., Beadle, B., Le, Q. T., Xing, L., Dai, X. 2024

  Abstract

  Radiation therapy (RT) is one of the most effective treatments for cancer, and its success relies on the accurate delineation of targets. However, target delineation is a comprehensive medical decision that currently relies purely on manual processes by human experts. Manual delineation is time-consuming, laborious, and subject to interobserver variations. Although the advancements in artificial intelligence (AI) techniques have significantly enhanced the auto-contouring of normal tissues, accurate delineation of RT target volumes remains a challenge. In this study, we propose a visual language model-based RT target volume auto-delineation network termed Radformer. The Radformer utilizes a hierarchical vision transformer as the backbone and incorporates large language models to extract text-rich features from clinical data. We introduce a visual language attention module (VLAM) for integrating visual and linguistic features for language-aware visual encoding (LAVE). The Radformer has been evaluated on a dataset comprising 2985 patients with head-and-neck cancer who underwent RT. Metrics, including the Dice similarity coefficient (DSC), intersection over union (IOU), and 95th percentile Hausdorff distance (HD95), were used to evaluate the performance of the model quantitatively. Our results demonstrate that the Radformer has superior segmentation performance compared to other state-of-the-art models, validating its potential for adoption in RT practice.

  View details for PubMedID 39040646

  View details for PubMedCentralID PMC11261986

• Design approach and benefits of the 3D-printed vaginal individualized applicator (VIA). Brachytherapy Ewongwo, A., Niedermayr, T., Kidd, E. A. 2024

  Abstract

  Interstitial gynecologic brachytherapy necessitates precise needle placement, requiring time and expertise. We aimed to simplify interstitial procedures and facilitate optimal needle distribution with individualized vaginal templates to guide interstitial needles.We developed the 3D-printed vaginal individualized applicator (VIA), a cylindrical template containing individualized internal channels that guide interstitial needles to cover the tumor extent. Eight patients underwent VIA only interstitial implants (VIA only), and five intact cervical cases were treated using tandem and customized VIA (VIA + T). Procedure length, number of needles utilized and dosimetric measures were evaluated.VIA was successfully designed and used clinically for 24 procedures (8 VIA only, 16 VIA + T). Average procedure needle insertion time reduced from 80.9 min for traditional interstitial to 42.9 min for VIA only, approximately 47% shorter with a similar mean high risk CTV volume (28.3 cc VIA only vs. 32.4 cc) and excellent dosimetry with average CTV V100% (94.3% and 94.4%). VIA + T was particularly useful in patients with small vaginal canals and large tumor size. For the five VIA + T patients average tumor size was 68.0cc (range 26.6-143.5 cc). VIA + T procedures were approximately 20% shorter than hybrid procedures with other applicators with mean length of 20.1 min and an average of 6.8 needles (range 3-12).Our novel 3D-printed VIA facilitates gynecologic interstitial brachytherapy by simplifying needle placement, reducing procedure time, and maintaining excellent dosimetry. VIA can be customized for various clinical scenarios, particularly beneficial for large tumors or small vaginal canals.

  View details for DOI 10.1016/j.brachy.2024.01.009

  View details for PubMedID 38402047

• Development and clinical implementation of simple needle attachment post placement interstitial template (SNAPP-IT) enabling a shorter, more direct needle path while preserving tumor visualization. Brachytherapy Baniel, C. C., Hui, C., Franco, P. A., Niedermayr, T., Kidd, E. A. 2023

  Abstract

  PURPOSE: Historical gynecologic interstitial brachytherapy templates block direct tumor visualization during needle placement, presenting an opportunity for clinical innovation to develop a novel interstitial template allowing direct tumor visualization during needle insertion.METHODS AND MATERIALS: We designed and implemented a novel interstitial template, simple needle attachment post placement interstitial template (SNAPP-IT), that allowed direct visualization of the target vaginal tumor during interstitial needle placement while maintaining the ability to individually secure needles to the template, allow a vaginal cylinder, suture holes for securing to the perineum, MRI compatibility and sterilizable for repeat use. Procedure outcomes including procedure time, needle path lengths, and plan dosimetry were prospectively captured in a patient database.RESULTS: Forty gynecologic interstitial brachytherapy cases were recorded (20 SNAPP-IT, 20 traditional templates). Needle insertion depth was reduced using the SNAPP-IT in comparison with traditional interstitial templates (11.8 cm vs. 3.6 cm, p < 0.0001). The average CTV volume was 25.6 cc for SNAPP-IT and 20.7 cc for traditional; both methods averaged a similar number of needles (15.8, 15.6). Dosimetric constraints were similarly met in both treatment groups. Procedures performed using the SNAPP-IT were shorter compared with those performed with traditional interstitial devices (83.4 minutes vs. 100.7 minutes) and there were no post-operative infections in the SNAPP-IT group.CONCLUSIONS: Implementation of a novel gynecologic interstitial brachytherapy template (SNAPP-IT) reduced procedure times, allowed direct tumor visualization, and decreased needle insertion depth. SNAPP-IT provides a useful alternative approach for vaginal interstitial brachytherapy, may increase brachytherapist efficiency with complex procedures and potentially expands access to interstitial brachytherapy.

  View details for DOI 10.1016/j.brachy.2023.12.002

  View details for PubMedID 38160101

• 3D-printed vaginal individualized applicator (VIA) simplifies procedure and optimizes dosimetry for gynecologic interstitial brachytherapy Niedermayr, T., Kidd, E. BMJ PUBLISHING GROUP. 2023: A84-A85

  View details for DOI 10.1136/ijgc-2023-IGCS.132

  View details for Web of Science ID 001262421600229

• Developing next generation 3D-printing for cervical cancer hybrid brachytherapy: a guided interstitial technique enabling improved flexibility, dosimetry, and efficiency. International journal of radiation oncology, biology, physics Marar, M., Niedermayr, T., Kidd, E. 2023

  Abstract

  We developed a 3D-printed tandem anchored radially guiding interstitial template (TARGIT) to increase the simplicity of intracavitary/interstitial (IC/IS) technique for tandem-and-ovoid (T&O) procedures in cervical cancer brachytherapy. This study compared dosimetry and procedure logistics between T&O implants using the original TARGIT versus the next-generation TARGIT-Flexible-eXtended (TARGIT-FX) 3D-printed template designed for practice-changing ease-of-use with further simplified needle insertion and increased flexibility in needle placement.This single-institution retrospective cohort study included patients undergoing T&O brachytherapy as part of definitive cervical cancer treatment. Procedures utilized the original TARGIT from November 2019 through February 2022 and the TARGIT-FX from March 2022 through November 2022. The FX design features full extension to the vaginal introitus with 9 needle channels and allows for needle additions or depth adjustments intra-procedure and post-CT/MRI imaging.A total of 148 implants were performed, 68 (46%) with TARGIT and 80 (54%) with TARGIT-FX, across 41 patients. Across implants, the TARGIT-FX achieved higher mean V100% (+2.8%, p = 0.0019), and across patients, the TARGIT-FX achieved higher D90 (+2.0 Gy, p = 0.037) and higher D98 (+2.7 Gy, p = 0.016) compared to the original TARGIT. Doses to organs at risk were overall similar between templates. Procedure times for TARGIT-FX implants were 30% shorter on average than for those using the original TARGIT (p < 0.0001), and 28% shorter on average for the subset of implants with HR-CTV ≥30 cc (p = 0.013). All residents (100%, N = 6) surveyed regarding the TARGIT-FX indicated ease-of-use for needle insertion and interest in applying the technique in future practice.The TARGIT-FX achieved shorter procedure times with increased tumor coverage and similar normal tissue sparing compared to the previously applied TARGIT and illustrates the potential of 3D-printing to enhance efficiency and shorten the learning curve for IC/IS procedure technique in cervical cancer brachytherapy.

  View details for DOI 10.1016/j.ijrobp.2023.04.005

  View details for PubMedID 37059235

• Radio-luminescent imaging for rapid, high resolution eye plaque loading verification. Medical physics Yan, H., De Jean, P., Grafil, E., Ashraf, R., Niedermayr, T., Astrahan, M., Mruthyunjaya, P., Beadle, B., Xing, L., Liu, W. 2022

  Abstract

  BACKGROUND: Eye plaque brachytherapy (EPB) is currently an optimal therapy for intraocular cancers. Due to the lack of an effective and practical technique to measure the seed radioactivity distribution, current quality assurance (QA) practice according to the AAPM TG129 only stipulates that the plaque assembly be visually inspected. Consequently, uniform seed activity is routinely adopted to avoid possible loading mistakes of differential seed loading. However, modulated dose delivery, which represents a general trend in radiotherapy to provide more personalized treatment for a given tumor and patient, requires differential activities in the loaded seeds.PURPOSE: In this study, a fast and low-cost radio-luminescent imaging and dose calculating system to verify the seed activity distribution for differential loading was developed.METHODS: A proof-of-concept system consisting of a thin scintillator sheet coupled to a camera/lens system was constructed. A seed-loaded plaque can be placed directly on the scintillator surface with the radioactive seeds facing the scintillator. The camera system collects the radioluminescent signal generated by the scintillator at its opposite side. The predicted dose distribution in the scintillator's sensitive layer was calculated using a Monte Carlo simulation with the planned plaque loading pattern of I-125 seeds. Quantitative comparisons of the distribution of relative measured signal intensity and that of the relative predicted dose in the sensitive layer were performed by gamma analysis, similar to IMRT QA.RESULTS: Data analyses showed high gamma (3%/0.3mm, global, 20% threshold) passing rates for correct seed loadings and low passing rates with distinguished high gamma value area for incorrect loadings, indicating that possible errors may be detected. The measurement and analysis only required a few extra minutes, significantly shorter than the time to assay the extra verification seeds the physicist already must perform as recommended by TG129.CONCLUSIONS: Radio-luminescent QA can be used to facilitate and assure the implementation of intensity modulated, customized plaque loading. This article is protected by copyright. All rights reserved.

  View details for DOI 10.1002/mp.16003

  View details for PubMedID 36183146

• Applying 3D-Printed Templates in High-Dose-Rate Brachytherapy for Cervix Cancer: Simplified Needle Insertion for Optimized Dosimetry. International journal of radiation oncology, biology, physics Marar, M., Simiele, E., Niedermayr, T., Kidd, E. A. 2022

  Abstract

  PURPOSE: In cervical cancer brachytherapy, adding interstitial (IS) needles to intracavitary (IC) applicators can enhance dosimetry by improving target coverage while limiting normal tissue dose. However, its use is limited to a subset of practitioners with appropriate technical skill. We designed Tandem Anchored Radially Guiding Interstitial Templates (TARGITs) with a 3D printing workflow to optimize needle placement and facilitate greater ease-of-use of IC/IS technique. This study compared dosimetry and procedure characteristics between tandem-and-ovoid implants (T&O) using TARGIT technique versus non-TARGIT technique.METHODS AND MATERIALS: This single-institution retrospective cohort study included patients undergoing T&O brachytherapy as part of definitive radiation treatment for cervical cancer between February 2017 and January 2021. TARGIT technique was implemented from November 2019 onwards; all prior procedures involved non-TARGIT technique using a No Needle (NN) or Freehand Needle (FN) approach. Target coverage, dose to organs-at-risk, and procedure times were evaluated and compared between TARGIT technique and non-TARGIT technique.RESULTS: The cohort included 70 patients with cervical cancer who underwent 302 T&O procedures: 68 (23%) with TARGIT technique and 234 (77%) with non-TARGIT technique, which included 133 NN and 101 FN implants. TARGIT implants involved longer average procedure times (+6.3 minutes, p<0.0001). TARGIT implants achieved a higher mean high-risk CTV V100% than non-TARGIT implants (+4.4%, p=0.001) including for large tumors 30 cc or greater (+8.1%, p=0.002). Average D90 was 4.6 Gy higher and average D98 was 3.2 Gy higher for TARGIT technique compared to non-TARGIT technique (p=0.006 and p=0.02). Total treatment doses to rectum, bowel, and bladder were not significantly different for TARGIT versus non-TARGIT technique.CONCLUSION: The 3D-printed TARGIT approach to T&O brachytherapy achieved greater tumor coverage while sparing normal tissues, particularly for large tumor volumes, with only a slight increase in average procedure time. TARGIT represents a creative technological solution for increasing accessibility of advanced IC/IS brachytherapy technique for cervical cancer definitive radiation treatment.

  View details for DOI 10.1016/j.ijrobp.2022.05.027

  View details for PubMedID 35654306

• Dose Prediction for Cervical Cancer Brachytherapy Using 3-D Deep Convolutional Neural Network IEEE TRANSACTIONS ON RADIATION AND PLASMA MEDICAL SCIENCES Ma, M., Kidd, E., Fahimian, B. P., Han, B., Niedermayr, T. R., Hristov, D., Xing, L., Yang, Y. 2022; 6 (2): 214-221

  View details for DOI 10.1109/TRPMS.2021.3098507

  View details for Web of Science ID 000750257400011

• Limited Time Penalty for Improved Dosimetry: Simplified Needle Insertion in Combined Tandem and Ovoid plus Interstitial Cases with Custom Templates Niedermayr, T., Kidd, E. LIPPINCOTT WILLIAMS & WILKINS. 2021: S14-S15

  View details for Web of Science ID 000701779700025

• MR to Ultrasound Image Registration with Segmentation-Based Learning for HDR Prostate Brachytherapy Chen, Y., Xing, L., Yu, L., Liu, W., Fahimian, B., Niedermayr, T., Bagshaw, H., Buyyounouski, M., Han, B. WILEY. 2021

  View details for Web of Science ID 000673145402120

• MR to ultrasound image registration with segmentation-based learning for HDR prostate brachytherapy. Medical physics Chen, Y. n., Xing, L. n., Yu, L. n., Liu, W. n., Fahimian, B. P., Niedermayr, T. n., Bagshaw, H. P., Buyyounouski, M. n., Han, B. n. 2021

  Abstract

  Propagation of contours from high-quality magnetic resonance (MR) images to treatment planning ultrasound (US) images with severe needle artifacts is a challenging task, which can greatly aid the organ contouring in high dose rate (HDR) prostate brachytherapy. In this study, a deep learning approach was developed to automatize this registration procedure for HDR brachytherapy practice.Because of the lack of training labels and difficulty of accurate registration from inferior image quality, a new segmentation-based registration framework was proposed for this multi-modality image registration problem. The framework consisted of two segmentation networks and a deformable registration network, based on the weakly-supervised registration strategy. Specifically, two 3D V-Nets were trained for the prostate segmentation on the MR and US images separately, to generate the weak supervision labels for the registration network training. Besides the image pair, the corresponding prostate probability maps from the segmentation were further fed to the registration network to predict the deformation matrix, and an augmentation method was designed to randomly scale the input and label probability maps during the registration network training. The overlap between the deformed and fixed prostate contours was analyzed to evaluate the registration accuracy. Three datasets were collected from our institution for the MR and US image segmentation networks, and the registration network learning, which contained 121, 104 and 63 patient cases, respectively.The mean Dice similarity coefficient (DSC) results of the two prostate segmentation networks are 0.86±0.05 and 0.90±0.03, for MR images and the US images after the needle insertion, respectively. The mean DSC, center-of-mass (COM) distance, Hausdorff distance (HD) and averaged symmetric surface distance (ASSD) results for the registration of manual prostate contours were 0.87±0.05, 1.70±0.89 mm, 7.21±2.07 mm, 1.61±0.64 mm, respectively. By providing the prostate probability map from the segmentation to the registration network, as well as applying the random map augmentation method, the evaluation results of the four metrics were all improved, such as an increase of DSC from 0.83±0.08 to 0.86±0.06 and from 0.86±0.06 to 0.87±0.05, respectively.A novel segmentation-based registration framework was proposed to automatically register prostate MR images to the treatment planning US images with metal artifacts, which not only largely saved the labor work on the data preparation, but also improved the registration accuracy. The evaluation results showed the potential of this approach in HDR prostate brachytherapy practice.

  View details for DOI 10.1002/mp.14901

  View details for PubMedID 33905566

• Intensity modulated Ir-192 brachytherapy using high-Z 3D printed applicators. Physics in medicine and biology Skinner, L. B., Niedermayr, T., Prionas, N., Perl, J., Fahimian, B. P., Kidd, E. 2020

  Abstract

  Gynecologic cancers are often asymmetric, yet current Ir-192 brachytherapy techniques provide only limited radial modulation of the dose. The shielded solutions investigated here solve this by providing the ability to modulate between highly asymmetric and radially symmetric dose distributions at a given location. To find applicator designs that can modulate between full dose and less than 50% dose, at the dimensions of the urethra, a 2D calculation algorithm was developed to narrow down the search space. Two shielding design types were then further investigated using Monte Carlo and Boltzmann-solver dose calculation algorithms. 3D printing techniques using ISO10993 certified biocompatible plastics and 3D printable tungsten-loaded plastics were tested. It was also found that shadowing effects set by the shape of the shielding cannot be easily modulated out, hence careful design is required. The shielded applicator designs investigated here, allow for reduction of the dose by over 50% at 5 mm from the applicator surface in desired regions, while also allowing radially symmetric dose with isodose line (IDL) deviations less than 0.5 mm from circular. The shielding designs were also chosen with treatment delivery time in mind. Treatment times for these shielded designs were found to be less than 1.4 times longer than a six-channel unshielded cylinder for the equivalent fully symmetric dose distribution. The 2D calculation methods developed here provide a simple way to rapidly evaluate shielding designs, while the 3D printing techniques also allow for devices with novel shapes to be rapidly prototyped. Both TOPAS Monte Carlo and Acuros BV calculations show that significant dose shaping, and organ at risk (OAR) sparing can be achieved without significantly compromising the plan in regions that require the full dose.

  View details for DOI 10.1088/1361-6560/ab9b54

  View details for PubMedID 32521512

• Factor 10 Expedience of Monthly Linac Quality Assurance via an Ion Chamber Array and Automation Scripts. Technology in cancer research & treatment Skinner, L. B., Yang, Y., Hsu, A., Xing, L., Yu, A. S., Niedermayr, T. 2019; 18: 1533033819876897

  Abstract

  PURPOSE: While critical for safe and accurate radiotherapy, monthly quality assurance of medical linear accelerators is time-consuming and takes physics resources away from other valuable tasks. The previous methods at our institution required 5 hours to perform the mechanical and dosimetric monthly linear accelerator quality assurance tests. An improved workflow was developed to perform these tests with higher accuracy, with fewer error pathways, in significantly less time.METHODS: A commercial ion chamber array (IC profiler, Sun Nuclear, Melbourne, Florida) is combined with automation scripts to consolidate monthly linear accelerator QA. The array was used to measure output, flatness, symmetry, jaw positions, gated dose constancy, energy constancy, collimator walkout, crosshair centering, and dosimetric leaf gap constancy. Treatment plans were combined with automation scripts that interface with Sun Nuclear's graphical user interface. This workflow was implemented on a standard Varian clinac, with no special adaptations, and can be easily applied to other C-arm linear accelerators.RESULTS: These methods enable, in 30 minutes, measurement and analysis of 20 of the 26 dosimetric and mechanical monthly tests recommended by TG-142. This method also reduces uncertainties in the measured beam profile constancy, beam energy constancy, field size, and jaw position tests, compared to our previous methods. One drawback is the increased uncertainty associated with output constancy. Output differences between IC profiler and farmer chamber in plastic water measurements over a 6-month period, across 4 machines, were found to have a 0.3% standard deviation for photons and a 0.5% standard deviation for electrons, which is sufficient for verifying output accuracy according to TG-142 guidelines. To minimize error pathways, automation scripts which apply the required settings, as well as check the exported data file integrity were employed.CONCLUSIONS: The equipment, procedure, and scripts used here reduce the time burden of routine quality assurance tests and in most instances improve precision over our previous methods.

  View details for DOI 10.1177/1533033819876897

  View details for PubMedID 31707931

• Optimizing efficiency and safety in external beam radiotherapy using automated plan check (APC) tool and six sigma methodology. Journal of applied clinical medical physics Liu, S. n., Bush, K. K., Bertini, J. n., Fu, Y. n., Lewis, J. M., Pham, D. J., Yang, Y. n., Niedermayr, T. R., Skinner, L. n., Xing, L. n., Beadle, B. M., Hsu, A. n., Kovalchuk, N. n. 2019; 20 (8): 56–64

  Abstract

  To develop and implement an automated plan check (APC) tool using a Six Sigma methodology with the aim of improving safety and efficiency in external beam radiotherapy.The Six Sigma define-measure-analyze-improve-control (DMAIC) framework was used by measuring defects stemming from treatment planning that were reported to the departmental incidence learning system (ILS). The common error pathways observed in the reported data were combined with our departmental physics plan check list, and AAPM TG-275 identified items. Prioritized by risk priority number (RPN) and severity values, the check items were added to the APC tool developed using Varian Eclipse Scripting Application Programming Interface (ESAPI). At 9 months post-APC implementation, the tool encompassed 89 check items, and its effectiveness was evaluated by comparing RPN values and rates of reported errors. To test the efficiency gains, physics plan check time and reported error rate were prospectively compared for 20 treatment plans.The APC tool was successfully implemented for external beam plan checking. FMEA RPN ranking re-evaluation at 9 months post-APC demonstrated a statistically significant average decrease in RPN values from 129.2 to 83.7 (P < .05). After the introduction of APC, the average frequency of reported treatment-planning errors was reduced from 16.1% to 4.1%. For high-severity errors, the reduction was 82.7% for prescription/plan mismatches and 84.4% for incorrect shift note. The process shifted from 4σ to 5σ quality for isocenter-shift errors. The efficiency study showed a statistically significant decrease in plan check time (10.1 ± 7.3 min, P = .005) and decrease in errors propagating to physics plan check (80%).Incorporation of APC tool has significantly reduced the error rate. The DMAIC framework can provide an iterative and robust workflow to improve the efficiency and quality of treatment planning procedure enabling a safer radiotherapy process.

  View details for DOI 10.1002/acm2.12678

  View details for PubMedID 31423729

• A multichannel superconducting tunnel junction detector for high-resolution X-ray spectroscopy of magnesium diboride films Applied Superconductivity Conference Friedrich, S., Vailionis, A., Drury, O., Niedermayr, T., Funk, T., Kang, W. N., Choi, E. M., Kim, H. J., Lee, S. I., Cramer, S. P., Kim, C., Labov, S. E. IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC. 2003: 1114–19

  View details for DOI 10.1109/TASC.2003.814169

  View details for Web of Science ID 000184241700259