Real‐time dynamic <scp>MR</scp> image reconstruction using compressed sensing and principal component analysis (<scp>CS</scp>‐<scp>PCA</scp>): Demonstration in lung tumor tracking

Dietz, Bryson; Yip, Eugene; Yun, Jihyun; Fallone, B. G.; Wachowicz, Keith

doi:10.1002/mp.12354

Cited by 16 publications

(23 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…), and presence of image artifacts from various sources (including bias field artifacts, which are not corrected for in this study) may all affect image quality and contouring performance. However, in our previous work [30][31][32] we have tested this autocontouring algorithm with images degraded retrospectively by additional noise and k-space undersampling, and has shown that the algorithm works reasonably well with lower image quality scenarios, albeit with slightly poorer performance metrics.…”

Section: Discussionmentioning

confidence: 99%

“…To present the results, the following notation is used: A 12 represents autocontours trained by expert 1 in session 2; whereas M 31 denotes manual contours drawn by expert 3 in session 1. The automatic contouring algorithm, trained using the contours from 30 training images (images 1-30) from each of the 6 manual contouring sessions (3 contourer 9 2 sessions) is used to generate 6 sets of 100 automatic contours (ROI AUTO ) on the 100 tracking images (images 31-130), labeled as A 11 , A 12 , A 21 , A 22 , A 31 , A 32 . These are compared against the 6 sets of manual contours (ROI STD ) drawn on the same tracking images, namely, M 11 , M 12 , M 21 , M 22 , M 31 , M 32.…”

Section: D Evaluation Of Contoursmentioning

confidence: 99%

“…, A 12 , A 21 , A22 , A31 , A32 are compared against the 6 sets of manual contours drawn on the same tracking images, namely, M11 , M 12 , M 21 , M 22 , M 31 , M 32 . DC is shown in Table I, with the diagonal element representing the automatic vs. manual SUSS DC, bolded elements representing automatic vs. manual SUDS DC, and underlined elements representing automatic vs. manual DU DC.…”

mentioning

confidence: 99%

“…Underlined elements represent DU comparisons. The manual contours for the 100 tracking images (i.e., images 31-130 in each data set), labelled M 11 , M 12 , M 21 , M 22 , M 31 , M32 , are compared against each other. DC is shown inTable III.…”

mentioning

confidence: 99%

See 3 more Smart Citations

Evaluating performance of a user‐trained MR lung tumor autocontouring algorithm in the context of intra‐ and interobserver variations

et al. 2017

Self Cite

View full text Add to dashboard Cite

Purpose: Real-time tracking of lung tumors using magnetic resonance imaging (MRI) has been proposed as a potential strategy to mitigate the ill-effects of breathing motion in radiation therapy. Several autocontouring methods have been evaluated against a "gold standard" of a single human expert user. However, contours drawn by experts have inherent intra-and interobserver variations. In this study, we aim to evaluate our user-trained autocontouring algorithm with manually drawn contours from multiple expert users, and to contextualize the accuracy of these autocontours within intra-and interobserver variations. Methods: Six nonsmall cell lung cancer patients were recruited, with institutional ethics approval. Patients were imaged with a clinical 3 T Philips MR scanner using a dynamic 2D balanced SSFP sequence under free breathing. Three radiation oncology experts, each in two separate sessions, contoured 130 dynamic images for each patient. For autocontouring, the first 30 images were used for algorithm training, and the remaining 100 images were autocontoured and evaluated. Autocontours were compared against manual contours in terms of Dice's coefficient (DC) and Hausdorff distances (d H ). Intra-and interobserver variations of the manual contours were also evaluated. Results: When compared with the manual contours of the expert user who trained it, the algorithm generates autocontours whose evaluation metrics (same session: DC = 0.90(0.03), d H = 3.8(1.6) mm; different session DC = 0.88(0.04), d H = 4.3(1.5) mm) are similar to or better than intraobserver variations (DC = 0.88(0.04), and d H = 4.3(1.7) mm) between two sessions. The algorithm's autocontours are also compared to the manual contours from different expert users with evaluation metrics (DC = 0.87(0.04), d H = 4.8(1.7) mm) similar to interobserver variations (DC = 0.87(0.04), d H = 4.7(1.6) mm). Conclusions: Our autocontouring algorithm delineates tumor contours (<20 ms per contour), in dynamic MRI of lung, that are comparable to multiple human experts (several seconds per contour), but at a much faster speed. At the same time, the agreement between autocontours and manual contours is comparable to the intra-and interobserver variations. This algorithm may be a key component of the real time tumor tracking workflow for our hybrid Linac-MR device in the future.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: D Evaluation Of Contoursmentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

See 2 more Smart Citations

Evaluating performance of a user‐trained MR lung tumor autocontouring algorithm in the context of intra‐ and interobserver variations

et al. 2017

Self Cite

View full text Add to dashboard Cite

show abstract

“…Of all available imaging modalities for IGRT, MRI is the most versatile and suitable candidate as it provides soft-tissue contrast to enable direct tumour visualization as well as OAR localization (Lagendijk et al 2014). Moreover, it provides real-time imaging to characterize and eventually track anatomical motion (Stemkens et al 2016, Dietz et al 2017 for MRI guided radiotherapy and for dose reconstruction (Glitzner et al 2015) and ultimately real-time plan adaptation (Kontaxis et al 2017).…”

Section: Introductionmentioning

confidence: 99%

First patients treated with a 1.5 T MRI-Linac: clinical proof of concept of a high-precision, high-field MRI guided radiotherapy treatment

et al. 2017

View full text Add to dashboard Cite

LETTER • OPEN ACCESSFirst patients treated with a 1.5 T MRI-Linac: clinical proof of concept of a high-precision, highfield MRI guided radiotherapy treatment AbstractThe integration of 1.5 T MRI functionality with a radiotherapy linear accelerator (linac) has been pursued since 1999 by the UMC Utrecht in close collaboration with Elekta and Philips. The idea behind this integrated device is to offer unrivalled, online and real-time, soft-tissue visualization of the tumour and the surroundings for more precise radiation delivery. The proof of concept of this device was given in 2009 by demonstrating simultaneous irradiation and MR imaging on phantoms, since then the device has been further developed and commercialized by Elekta. The aim of this work is to demonstrate the clinical feasibility of online, high-precision, high-field MRI guidance of radiotherapy using the first clinical prototype MRI-Linac.Four patients with lumbar spine bone metastases were treated with a 3 or 5 beam step-and-shoot IMRT plan. The IMRT plan was created while Letter Institute of Physics and Engineering in MedicineOriginal content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. 3 Author to whom any correspondence should be addressed. the patient was on the treatment table and based on the online 1.5 T MR images; pre-treatment CT was deformably registered to the online MRI to obtain Hounsfield values. Bone metastases were chosen as the first site as these tumors can be clearly visualized on MRI and the surrounding spine bone can be detected on the integrated portal imager. This way the portal images served as an independent verification of the MRI based guidance to quantify the geometric precision of radiation delivery. Dosimetric accuracy was assessed post-treatment from phantom measurements with an ionization chamber and film. Absolute doses were found to be highly accurate, with deviations ranging from 0.0% to 1.7% in the isocenter. The geometrical, MRI based targeting as confirmed using portal images was better than 0.5 mm, ranging from 0.2 mm to 0.4 mm.In conclusion, high precision, high-field, 1.5 T MRI guided radiotherapy is clinically feasible.

show abstract

Time domain principal component analysis for rapid, real‐time 2D MRI reconstruction from undersampled data

et al. 2021

Self Cite

View full text Add to dashboard Cite

A rapid real-time 2D accelerated method was developed for magnetic resonance imaging (MRI) using principal component analysis (PCA) in the temporal domain. This method employs a moving window of previous dynamic frames to reconstruct the current, real-time frame within this window. This technique could be particularly useful in real-time tracking applications such as in MR-guided radiotherapy,where low latency real-time reconstructions are essential. Methods: The method was tested retrospectively on 15 fully-sampled data sets of lung patient data acquired on a 3T Philips Achieva system. High frequency data are incoherently undersampled, while the central low-frequency data are always acquired to characterize the temporal fluctuations through PCA. The undersampling pattern is derived in such a way that all of k-space is acquired within a pre-determined number of frames. The missing data in the current frame are then filled in by fitting the temporal characterizations to the acquired undersampled data, using a pre-determined number of PCs. A subset of six patients was used to test the contour ability of the images. Various accelerations between 3x and 8x were tested along with the optimal number of PCs for fitting. A comparison was also performed with previous work from our group proposed by Dietz et al. as well as with a standard low resolution acquisition. In order to determine how the method would perform at lower signal to noise ratio (SNR), noise levels of 2×, 4×, and 6× were added to the 3T data. Metrics such as normalised mean square error and Dice coefficient were used to measure the reconstruction image quality and contour ability. Results:The proposed method demonstrated good temporal robustness as consistent metrics were detected for the duration of the imaging session. It was found that the optimal number of PCs for temporal fitting was dependent on the acceleration rate. For the data tested, five PCs were found to be optimal at the acceleration rates of 3× and 4×. This number decreases to three at accelerations of 5× and 6× and further decreases to two at an acceleration rate of 8×, likely due to greater instability with fewer acquired data points. The use of too many PCs for fitting increased the chances of noisy reconstruction which affected contourability. Conclusions: The proposed 2D real-time MR acceleration method demonstrated greater robustness in the metrics over time when compared with previous real-time PCA methods using metrics such as normalised mean squared error, peak SNR and structural similarity up to an acceleration of 8x. Improved temporal robustness of image structure contourability and accurate definition was also demonstrated using several metrics including the Dice coefficient. Reconstruction of raw acquired data can be performed at approximately 50 ms 6724

show abstract

Real‐time dynamic MR image reconstruction using compressed sensing and principal component analysis (CS‐PCA): Demonstration in lung tumor tracking

Cited by 16 publications

References 24 publications

Evaluating performance of a user‐trained MR lung tumor autocontouring algorithm in the context of intra‐ and interobserver variations

Evaluating performance of a user‐trained MR lung tumor autocontouring algorithm in the context of intra‐ and interobserver variations

First patients treated with a 1.5 T MRI-Linac: clinical proof of concept of a high-precision, high-field MRI guided radiotherapy treatment

Time domain principal component analysis for rapid, real‐time 2D MRI reconstruction from undersampled data

Contact Info

Product

Resources

About