Real-time motion management in MRI-guided radiotherapy: Current status and AI-enabled prospects

MRI-guided radiotherapy (MRIgRT) is a highly complex treatment modality, allowing adaptation to anatomical changes occurring from one treatment day to the other (inter-fractional), but also to motion occurring during a treatment fraction (intra-fractional). In this vision paper, we describe the different steps of intra-fractional motion management during MRIgRT, from imaging to beam adaptation, and the solutions currently available both clinically and at a research level. Furthermore, considering the latest developments in the literature, a workflow is foreseen in which motion-induced over- and/or under-dosage is compensated in 3D, with minimal impact to the radiotherapy treatment time. Considering the time constraints of real-time adaptation, a particular focus is put on artificial intelligence (AI) solutions as a fast and accurate alternative to conventional algorithms.

Experimental comparison of linear regression and LSTM motion prediction models for MLC-tracking on an MRI-linac

BACKGROUND

Magnetic resonance imaging (MRI)-guided radiotherapy with multileaf collimator (MLC)-tracking is a promising technique for intra-fractional motion management, achieving high dose conformality without prolonging treatment times. To improve beam-target alignment, the geometric error due to system latency should be reduced by using temporal prediction.

PURPOSE

To experimentally compare linear regression (LR) and long-short-term memory (LSTM) motion prediction models for MLC-tracking on an MRI-linac using multiple patient-derived traces with different complexities.

METHODS

Experiments were performed on a prototype 1.0 T MRI-linac capable of MLC-tracking. A motion phantom was programmed to move a target in superior-inferior (SI) direction according to eight lung cancer patient respiratory motion traces. Target centroid positions were localized from sagittal 2D cine MRIs acquired at 4 Hz using a template matching algorithm. The centroid positions were input to one of four motion prediction models. We used (1) a LSTM network which had been optimized in a previous study on patient data from another cohort (offline LSTM). We also used (2) the same LSTM model as a starting point for continuous re-optimization of its weights during the experiment based on recent motion (offline+online LSTM). Furthermore, we implemented (3) a continuously updated LR model, which was solely based on recent motion (online LR). Finally, we used (4) the last available target centroid without any changes as a baseline (no-predictor). The predictions of the models were used to shift the MLC aperture in real-time. An electronic portal imaging device (EPID) was used to visualize the target and MLC aperture during the experiments. Based on the EPID frames, the root-mean-square error (RMSE) between the target and the MLC aperture positions was used to assess the performance of the different motion predictors. Each combination of motion trace and prediction model was repeated twice to test stability, for a total of 64 experiments.

RESULTS

The end-to-end latency of the system was measured to be (389 ± 15) ms and was successfully mitigated by both LR and LSTM models. The offline+online LSTM was found to outperform the other models for all investigated motion traces. It obtained a median RMSE over all traces of (2.8 ± 1.3) mm, compared to the (3.2 ± 1.9) mm of the offline LSTM, the (3.3 ± 1.4) mm of the online LR and the (4.4 ± 2.4) mm when using the no-predictor. According to statistical tests, differences were significant (p-value <0.05) among all models in a pair-wise comparison, but for the offline LSTM and online LR pair. The offline+online LSTM was found to be more reproducible than the offline LSTM and the online LR with a maximum deviation in RMSE between two measurements of 10%.

CONCLUSIONS

This study represents the first experimental comparison of different prediction models for MRI-guided MLC-tracking using several patient-derived respiratory motion traces. We have shown that among the investigated models, continuously re-optimized LSTM networks are the most promising to account for the end-to-end system latency in MRI-guided radiotherapy with MLC-tracking.

Distortion-corrected image reconstruction with deep learning on an MRI-Linac

PURPOSE

MRI is increasingly utilized for image-guided radiotherapy due to its outstanding soft-tissue contrast and lack of ionizing radiation. However, geometric distortions caused by gradient nonlinearities (GNLs) limit anatomical accuracy, potentially compromising the quality of tumor treatments. In addition, slow MR acquisition and reconstruction limit the potential for effective image guidance. Here, we demonstrate a deep learning-based method that rapidly reconstructs distortion-corrected images from raw k-space data for MR-guided radiotherapy applications.

METHODS

We leverage recent advances in interpretable unrolling networks to develop a Distortion-Corrected Reconstruction Network (DCReconNet) that applies convolutional neural networks (CNNs) to learn effective regularizations and nonuniform fast Fourier transforms for GNL-encoding. DCReconNet was trained on a public MR brain dataset from 11 healthy volunteers for fully sampled and accelerated techniques, including parallel imaging (PI) and compressed sensing (CS). The performance of DCReconNet was tested on phantom, brain, pelvis, and lung images acquired on a 1.0T MRI-Linac. The DCReconNet, CS-, PI-and UNet-based reconstructed image quality was measured by structural similarity (SSIM) and RMS error (RMSE) for numerical comparisons. The computation time and residual distortion for each method were also reported.

RESULTS

Imaging results demonstrated that DCReconNet better preserves image structures compared to CS- and PI-based reconstruction methods. DCReconNet resulted in the highest SSIM (0.95 median value) and lowest RMSE (<0.04) on simulated brain images with four times acceleration. DCReconNet is over 10-times faster than iterative, regularized reconstruction methods.

CONCLUSIONS

DCReconNet provides fast and geometrically accurate image reconstruction and has the potential for MRI-guided radiotherapy applications.

Real-time radial reconstruction with domain transform manifold learning for MRI-guided radiotherapy

BACKGROUND

MRI-guidance techniques that dynamically adapt radiation beams to follow tumor motion in real time will lead to more accurate cancer treatments and reduced collateral healthy tissue damage. The gold-standard for reconstruction of undersampled MR data is compressed sensing (CS) which is computationally slow and limits the rate that images can be available for real-time adaptation.

PURPOSE

Once trained, neural networks can be used to accurately reconstruct raw MRI data with minimal latency. Here, we test the suitability of deep-learning-based image reconstruction for real-time tracking applications on MRI-Linacs.

METHODS

We use automated transform by manifold approximation (AUTOMAP), a generalized framework that maps raw MR signal to the target image domain, to rapidly reconstruct images from undersampled radial k-space data. The AUTOMAP neural network was trained to reconstruct images from a golden-angle radial acquisition, a benchmark for motion-sensitive imaging, on lung cancer patient data and generic images from ImageNet. Model training was subsequently augmented with motion-encoded k-space data derived from videos in the YouTube-8M dataset to encourage motion robust reconstruction.

RESULTS

AUTOMAP models fine-tuned on retrospectively acquired lung cancer patient data reconstructed radial k-space with equivalent accuracy to CS but with much shorter processing times. Validation of motion-trained models with a virtual dynamic lung tumor phantom showed that the generalized motion properties learned from YouTube lead to improved target tracking accuracy.

CONCLUSION

AUTOMAP can achieve real-time, accurate reconstruction of radial data. These findings imply that neural-network-based reconstruction is potentially superior to alternative approaches for real-time image guidance applications.

Integrated MRI-guided radiotherapy - opportunities and challenges

MRI can help to categorize tissues as malignant or non-malignant both anatomically and functionally, with a high level of spatial and temporal resolution. This non-invasive imaging modality has been integrated with radiotherapy in devices that can differentially target the most aggressive and resistant regions of tumours. The past decade has seen the clinical deployment of treatment devices that combine imaging with targeted irradiation, making the aspiration of integrated MRI-guided radiotherapy (MRIgRT) a reality. The two main clinical drivers for the adoption of MRIgRT are the ability to image anatomical changes that occur before and during treatment in order to adapt the treatment approach, and to image and target the biological features of each tumour. Using motion management and biological targeting, the radiation dose delivered to the tumour can be adjusted during treatment to improve the probability of tumour control, while simultaneously reducing the radiation delivered to non-malignant tissues, thereby reducing the risk of treatment-related toxicities. The benefits of this approach are expected to increase survival and quality of life. In this Review, we describe the current state of MRIgRT, and the opportunities and challenges of this new radiotherapy approach.

Pre-treatment and real-time image guidance for a fixed-beam radiotherapy system

PURPOSE

A radiotherapy system with a fixed treatment beam and a rotating patient positioning system could be smaller, more robust and more cost effective compared to conventional rotating gantry systems. However, patient rotation could cause anatomical deformation and compromise treatment delivery. In this work, we demonstrate an image-guided treatment workflow with a fixed beam prototype system that accounts for deformation during rotation to maintain dosimetric accuracy.

METHODS

The prototype system consists of an Elekta Synergy linac with the therapy beam orientated downward and a custom-built patient rotation system (PRS). A phantom that deforms with rotation was constructed and rotated within the PRS to quantify the performance of two image guidance techniques: motion compensated cone-beam CT (CBCT) for pre-treatment volumetric imaging and kilovoltage infraction monitoring (KIM) for real-time image guidance. The phantom was irradiated with a 3D conformal beam to evaluate the dosimetric accuracy of the workflow.

RESULTS

The motion compensated CBCT was used to verify pre-treatment position and the average calculated position was within -0.3 ± 1.1 mm of the phantom's ground truth position at 0°. KIM tracked the position of the target in real-time as the phantom was rotated and the average calculated position was within -0.2 ± 0.8 mm of the phantom's ground truth position. A 3D conformal treatment delivered on the prototype system with image guidance had a 3%/2 mm gamma pass rate of 96.3% compared to 98.6% delivered using a conventional rotating gantry linac.

CONCLUSIONS

In this work, we have shown that image guidance can be used with fixed-beam treatment systems to measure and account for changes in target position in order to maintain dosimetric coverage during horizontal rotation. This treatment modality could provide a viable treatment option when there insufficient space for a conventional linear accelerator or where the cost is prohibitive.

Small field detector correction factors: effects of the flattening filter for Elekta and Varian linear accelerators

Flattening filter‐free (FFF) beams are becoming the preferred beam type for stereotactic radiosurgery (SRS) and stereotactic ablative radiation therapy (SABR), as they enable an increase in dose rate and a decrease in treatment time. This work assesses the effects of the flattening filter on small field output factors for 6 MV beams generated by both Elekta and Varian linear accelerators, and determines differences between detector response in flattened (FF) and FFF beams. Relative output factors were measured with a range of detectors (diodes, ionization chambers, radiochromic film, and microDiamond) and referenced to the relative output factors measured with an air core fiber optic dosimeter (FOD), a scintillation dosimeter developed at Chris O'Brien Lifehouse, Sydney. Small field correction factors were generated for both FF and FFF beams. Diode measured detector response was compared with a recently published mathematical relation to predict diode response corrections in small fields. The effect of flattening filter removal on detector response was quantified using a ratio of relative detector responses in FFF and FF fields for the same field size. The removal of the flattening filter was found to have a small but measurable effect on ionization chamber response with maximum deviations of less than ±0.9% across all field sizes measured. Solid‐state detectors showed an increased dependence on the flattening filter of up to ±1.6%. Measured diode response was within ±1.1% of the published mathematical relation for all fields up to 30 mm, independent of linac type and presence or absence of a flattening filter. For 6 MV beams, detector correction factors between FFF and FF beams are interchangeable for a linac between FF and FFF modes, providing that an additional uncertainty of up to ±1.6% is accepted.

PACS number(s): 87.55.km, 87.56.bd, 87.56.Da

Twisted pair of optic fibers for background removal in radiation fields

In many situations in which an optic fiber carries a signal through a radiation field, an unwanted background signal is produced consisting of fluorescent and/or Cerenkov light. This presents a major problem in the measurement of the light signal, for example, in scintillation dosimetry of medical therapeutic beams. In this paper, we demonstrate a new method of measuring and removing the background signal through the use of a twisted pair of optic fibers. The twisted pair consists of a fiber carrying the scintillation signal that is twisted with a second optic fiber to form a double helix. The two twisted fibers will experience the same radiation environment provided the periodicity of the twist is correlated to the dose rate gradient. An expression for the required twist periodicity is presented. A scintillation dosimeter with a twisted pair optic fiber was tested in a megavoltage beam and found to accurately measure its beam characteristics. The twisted pair approach is not restricted to medical applications and can be used in many situations in which optical signals are carried through radiation fields.

A method to remove residual signals in fibre optic luminescence dosimeters

Whenever a fibre optic is used to convey a light signal through a radiation field, it is likely that an unwanted background signal will arise from Cerenkov or fluorescent light which will contaminate the signal. In luminescence dosimetry of high energy beams, when a fibre optic is used to convey the signal from the radiation field to the detector, Cerenkov light is the dominant contributor to the background signal and must be corrected for. In this work, a novel method is demonstrated to separate the signal from the unwanted background. A remotely operated shutter is used to block the signal, allowing the residual background in the fibre optic to be quantified. This background is subtracted from the total measurement acquired in a subsequent irradiation, enabling the luminescence signal to be extracted. Two types of shutter mechanism are considered: an electro-mechanical device to intercept the light path and an LCD device to block the light by cross-polarization. Both shutters were characterized and incorporated into a fibre optic dosimetry system used to measure the radiation dose produced by external beam radiation linear accelerators. The dosimeter using each of the shutters in turn was exposed to a 6 MV photon beam to determine their performance, including the measurement of field size dependent output factors. The mechanical shutter determined the output factors to within 0.29% of those measured with an ionization chamber, whereas the LCD shutter gave results that deviated by up to 2.4%. The switching precision of both shutters was good with standard deviations of less than 0.25% and both were able to completely block the light signal when closed. The use of shutters could therefore be applied to any fibre optic based system to quantify and remove a reproducible background arising from any source including ambient, fluorescent and Cerenkov light.

Real-time scintillation array dosimetry for radiotherapy: the advantages of photomultiplier detectors

PURPOSE

In this paper, a photomultiplier tube (PMT) array dosimetry system has been developed and tested for the real-time readout of multiple scintillation signals from fiber optic dosimeters. It provides array dosimetry with the advantages in sensitivity provided by a PMT, but without the need for a separate PMT for each detector element.

METHODS

The PMT array system consisted of a multianode PMT, a multichannel data acquisition system, housing and optic fiber connections suitable for clinical use. The reproducibility, channel uniformity, channel crosstalk, acquisition speed, and sensitivity of the PMT array were quantified using a constant light source. Its performance was compared to other readout systems used in scintillation dosimetry. An in vivo HDR brachytherapy treatment was used as an example of a clinical application of the dosimetry system to the measurement of dose at multiple sites in the rectum. The PMT array system was also tested in the pulsed beam of a linear accelerator to test its response speed and its application with two separate methods of Cerenkov background removal.

RESULTS

The PMT array dosimetry system was highly reproducible with a measurement uncertainty of 0.13% for a 10 s acquisition period. Optical crosstalk between neighboring channels was accounted for by omitting every second channel. A mathematical procedure was used to account for the crosstalk in next-neighbor channels. The speed and sensitivity of the PMT array system were found be superior to CCD cameras, allowing for measurement of more rapid changes in dose rate. This was further demonstrated by measuring the dose delivered by individual photon pulses of a linear accelerator beam.

CONCLUSIONS

The PMT array system has advantages over CCD camera-based systems for the readout of scintillation light. It provided a more sensitive, more accurate, and faster response to meet the demands of future developments in treatment delivery.