Using deep learning to model the biological dose prediction on bulky lung cancer patients of partial stereotactic ablation radiotherapy

PURPOSE

To develop a biological dose prediction model considering tissue bio-reactions in addition to patient anatomy for achieving a more comprehensive evaluation of tumor control and promoting the automatic planning of bulky lung cancer.

METHODS

A database containing images and partial stereotactic ablation boost radiotherapy (P-SABR) plans of 94 bulky lung cancer patients was studied. Patient-specific parameters of gross tumor boost volume (GTVb), planning gross target volume (PGTV), and identified organs at risk (OARs) were extracted via Numpy and simple ITK. The original dose and structure maps for P-SABR patients were resampled to have a voxel resolution of 3.9 × 3.9 × 3 mm³ . Biological equivalent dose (BED) distributions were reprogrammed based on physical dose volumes. A developed deep learning architecture, Nestnet, was adopted as the training framework. We utilized two approaches for data organization to correlate the structures and BED: (a) BED programming before training model (B-Nestnet); (b) BED programming after the training process (D-B Nestnet). The early-stop mechanism was adopted on the validation set to avoid overfitting. The evaluation criteria of predictive accuracy contain the minimum BED of GTVb and PGTV, the maximum and the mean BED of all targets, BED-volume metrics. For comparison, we also used the original Unet for BED prediction. The absolute differences were statistically analyzed with the paired-samples t test.

RESULTS

The statistical outcomes demonstrate that D-B Nestnet model predicts biological dose distributions accurately. The average absolute biases of [max, mean] BED for GTVb, PGTV are [2.1%, 3.3%] and [2.1%, 4.7%], respectively. Averaging across most of OARs, the D-B Nestnet model is capable of predicting the errors of the max and mean BED within 6.3% and 6.1%, respectively. While the compared models performed worse with averaged max and mean BED prediction errors surpassing 10% on some specific OARs.

CONCLUSIONS

The study developed a D-B Nestnet model capable of predicting BED distribution accurately for bulky lung cancer patients in P-SABR. The predicted BED map enables a quick intuitive evaluation of tumor ablation, modification of the ablation range to improve BED of tumor targets, and quality assessment. It represents a major step forward toward automated P-SABR planning on bulky lung cancer in real clinical practice.

Rat sensorimotor cortex tolerance to parallel transections induced by synchrotron-generated X-ray microbeams

Microbeam radiation therapy is a novel preclinical technique, which uses synchrotron-generated X-rays for the treatment of brain tumours and drug-resistant epilepsies. In order to safely translate this approach to humans, a more in-depth knowledge of the long-term radiobiology of microbeams in healthy tissues is required. We report here the result of the characterization of the rat sensorimotor cortex tolerance to microradiosurgical parallel transections. Healthy adult male Wistar rats underwent irradiation with arrays of parallel microbeams. Beam thickness, spacing and incident dose were 100 or 600 µm, 400 or 1200 µm and 360 or 150 Gy, respectively. Motor performance was carried over a 3-month period. Three months after irradiation rats were sacrificed to evaluate the effects of irradiation on brain tissues by histology and immunohistochemistry. Microbeam irradiation of sensorimotor cortex did not affect weight gain and motor performance. No gross signs of paralysis or paresis were also observed. The cortical architecture was not altered, despite the presence of cell death along the irradiation path. Reactive gliosis was evident in the microbeam path of rats irradiated with 150 Gy, whereas no increase was observed in rats irradiated with 360 Gy.

Medical physics aspects of the synchrotron radiation therapies: Microbeam radiation therapy (MRT) and synchrotron stereotactic radiotherapy (SSRT)

Stereotactic Synchrotron Radiotherapy (SSRT) and Microbeam Radiation Therapy (MRT) are both novel approaches to treat brain tumor and potentially other tumors using synchrotron radiation. Although the techniques differ by their principles, SSRT and MRT share certain common aspects with the possibility of combining their advantages in the future. For MRT, the technique uses highly collimated, quasi-parallel arrays of X-ray microbeams between 50 and 600 keV. Important features of highly brilliant Synchrotron sources are a very small beam divergence and an extremely high dose rate. The minimal beam divergence allows the insertion of so called Multi Slit Collimators (MSC) to produce spatially fractionated beams of typically ∼25-75 micron-wide microplanar beams separated by wider (100-400 microns center-to-center(ctc)) spaces with a very sharp penumbra. Peak entrance doses of several hundreds of Gy are extremely well tolerated by normal tissues and at the same time provide a higher therapeutic index for various tumor models in rodents. The hypothesis of a selective radio-vulnerability of the tumor vasculature versus normal blood vessels by MRT was recently more solidified. SSRT (Synchrotron Stereotactic Radiotherapy) is based on a local drug uptake of high-Z elements in tumors followed by stereotactic irradiation with 80 keV photons to enhance the dose deposition only within the tumor. With SSRT already in its clinical trial stage at the ESRF, most medical physics problems are already solved and the implemented solutions are briefly described, while the medical physics aspects in MRT will be discussed in more detail in this paper.

Absolute perfusion measurements and associated iodinated contrast agent time course in brain metastasis: a study for contrast-enhanced radiotherapy

Contrast-enhanced radiotherapy is an innovative treatment that combines the selective accumulation of heavy elements in tumors with stereotactic irradiations using medium energy X-rays. The radiation dose enhancement depends on the absolute amount of iodine reached in the tumor and its time course. Quantitative, postinfusion iodine biodistribution and associated brain perfusion parameters were studied in human brain metastasis as key parameters for treatment feasibility and quality. Twelve patients received an intravenous bolus of iodinated contrast agent (CA) (40 mL, 4 mL/s), followed by a steady-state infusion (160 mL, 0.5 mL/s) to ensure stable intratumoral amounts of iodine during the treatment. Absolute iodine concentrations and quantitative perfusion maps were derived from 40 multislice dynamic computed tomography (CT) images of the brain. The postinfusion mean intratumoral iodine concentration (over 30 minutes) reached 1.94 ± 0.12 mg/mL. Reasonable correlations were obtained between these concentrations and the permeability surface area product and the cerebral blood volume. To our knowledge, this is the first quantitative study of CA biodistribution versus time in brain metastasis. The study shows that suitable and stable amounts of iodine can be reached for contrast-enhanced radiotherapy. Moreover, the associated perfusion measurements provide useful information for the patient recruitment and management processes.

Synchrotron-generated microbeam sensorimotor cortex transections induce seizure control without disruption of neurological functions

Synchrotron-generated X-ray microplanar beams (microbeams) are characterized by the ability to deliver extremely high doses of radiation to spatially restricted volumes of tissue. Minimal dose spreading outside the beam path provides an exceptional degree of protection from radio-induced damage to the neurons and glia adjacent to the microscopic slices of tissue irradiated. The preservation of cortical architecture following high-dose microbeam irradiation and the ability to induce non-invasively the equivalent of a surgical cut over the cortex is of great interest for the development of novel experimental models in neurobiology and new treatment avenues for a variety of brain disorders. Microbeams (size 100 µm/600 µm, center-to-center distance of 400 µm/1200 µm, peak entrance doses of 360-240 Gy/150-100 Gy) delivered to the sensorimotor cortex of six 2-month-old naïve rats generated histologically evident cortical transections, without modifying motor behavior and weight gain up to 7 months. Microbeam transections of the sensorimotor cortex dramatically reduced convulsive seizure duration in a further group of 12 rats receiving local infusion of kainic acid. No subsequent neurological deficit was associated with the treatment. These data provide a novel tool to study the functions of the cortex and pave the way for the development of new therapeutic strategies for epilepsy and other neurological diseases.

Intracerebral delivery of carboplatin in combination with either 6 MV photons or monoenergetic synchrotron X-rays are equally efficacious for treatment of the F98 rat glioma

BACKGROUND

The purpose of the present study was to compare side-by-side the therapeutic efficacy of a 6-day infusion of carboplatin, followed by X-irradiation with either 6 MV photons or synchrotron X-rays, tuned above the K-edge of Pt, for treatment of F98 glioma bearing rats.

METHODS

Carboplatin was administered intracerebrally (i.c.) to F98 glioma bearing rats over 6 days using AlzetTM osmotic pumps starting 7 days after tumor implantation. Radiotherapy was delivered in a single 15 Gy fraction on day 14 using a conventional 6 MV linear accelerator (LINAC) or 78.8 keV synchrotron X-rays.

RESULTS

Untreated control animals had a median survival time (MeST) of 33 days. Animals that received either carboplatin alone or irradiation alone with either 78.8 keV or 6 MV had a MeSTs 38 and 33 days, respectively. Animals that received carboplatin in combination with X-irradiation had a MeST of > 180 days with a 55% cure rate, irrespective of whether they were irradiated with either 78.8 KeV synchrotron X-rays or 6MV photons.

CONCLUSIONS

These studies have conclusively demonstrated the equivalency of i.c. delivery of carboplatin in combination with X-irradiation with either 6 MV photons or synchrotron X-rays.

Monochromatic minibeams radiotherapy: from healthy tissue-sparing effect studies toward first experimental glioma bearing rats therapy

PURPOSE

The purpose of this study was to evaluate high-dose single fraction delivered with monochromatic X-rays minibeams for the radiotherapy of primary brain tumors in rats.

METHODS AND MATERIALS

Two groups of healthy rats were irradiated with one anteroposterior minibeam incidence (four minibeams, 123 Gy prescribed dose at 1 cm depth in the brain) or two interleaved incidences (54 Gy prescribed dose in a 5 × 5 × 4.8 mm(3) volume centered in the right hemisphere), respectively. Magnetic resonance imaging (MRI) follow-up was performed over 1 year. T2-weighted (T2w) images, apparent diffusion coefficient (ADC), and blood vessel permeability maps were acquired. F98 tumor bearing rats were also irradiated with interleaved minibeams to achieve a homogeneous dose of 54 Gy delivered to an 8 × 8 × 7.8 mm(3) volume centered on the tumor. Anatomic and functional MRI follow-up was performed every 10 days after irradiation. T2w images, ADC, and perfusion maps were acquired.

RESULTS

All healthy rats were euthanized 1 year after irradiation without any clinical alteration visible by simple examination. T2w and ADC measurements remain stable for the single incidence irradiation group. Localized Gd-DOTA permeability, however, was observed 9 months after irradiation for the interleaved incidences group. The survival time of irradiated glioma bearing rats was significantly longer than that of untreated animals (49 ± 12.5 days versus 23.3 ± 2 days, p < 0.001). The tumoral cerebral blood flow and blood volume tend to decrease after irradiation.

CONCLUSIONS

This study demonstrates the sparing effect of minibeams on healthy tissue. The increased life span achieved for irradiated glioma bearing rats was similar to the one obtained with other radiotherapy techniques. This experimental tumor therapy study shows the feasibility of using X-ray minibeams with high doses in brain tumor radiotherapy.

Treatment plans optimization for contrast-enhanced synchrotron stereotactic radiotherapy

PURPOSE

Synchrotron stereotactic radiotherapy (SSRT) is a treatment that involves the targeting of high-Z elements into tumors followed by stereotactic irradiation with monochromatic x-rays from a synchrotron source, tuned at an optimal energy. The irradiation geometry, as well as the secondary particles generated at a higher yield by the medium energy x-rays on the high-Z atoms (characteristic x-rays, photoelectrons, and Auger electrons), produces a localized dose enhancement in the tumor. Iodine-enhanced SSRT with systemic injections of iodinated contrast agents has been successfully developed in the past six years in the team, and is currently being transferred to clinical trials. The purpose of this work is to study the impact on the SSRT treatment of the contrast agent type, the beam quality, the irradiation geometry, and the beam weighting for defining an optimized SSRT treatment plan.

METHODS

Theoretical dosimetry was performed using the MCNPX particle transport code. The simulated geometry was an idealized phantom representing a human head. A virtual target was positioned in the central part of the phantom or off-centered by 4 cm. The authors investigated the dosimetric characteristics of SSRT for various contrast agents: Iodine, gadolinium, and gold; and for different beam qualities: Monochromatic x-ray beams from a synchrotron source (30-120 keV), polychromatic x-ray beams from an x-ray tube (80, 120, and 180 kVp), and a 6 MV x-ray beam from a linear accelerator. Three irradiation geometries were studied: One arc or three noncoplanar arcs dynamic arc therapy, and an irradiation with a finite number of beams. The resulting dose enhancements, beam profiles, and histograms dose volumes were compared for iodine-enhanced SSRT. An attempt to optimize the irradiation scheme by weighing the finite x-ray beams was performed. Finally, the optimization was studied on patient specific 3D CT data after contrast agent infusion.

RESULTS

It was demonstrated in this study that an 80 keV beam energy was a good compromise for treating human brain tumors with iodine-enhanced SSRT, resulting in a still high dose enhancement factor (about 2) and a superior bone sparing in comparison with lower energy x-rays. This beam could easily be produced at the European Synchrotron Radiation Facility medical beamline. Moreover, there was a significant diminution of dose delivered to the bone when using monochromatic x-rays rather than polychromatic x-rays from a conventional tube. The data showed that iodine SSRT exhibits a superior sparing of brain healthy tissue in comparison to high energy treatment. The beam weighting optimization significantly improved the treatment plans for off-centered tumors, when compared to nonweighted irradiations.

CONCLUSIONS

This study demonstrated the feasibility of realistic clinical plans for low energy monochromatic x-rays contrast-enhanced radiotherapy, suitable for the first clinical trials on brain metastasis with a homogeneous iodine uptake.

Monte Carlo simulations and atomic calculations for Auger processes in biomedical nanotheranostics

We present numerical simulations of X-ray emission and absorption in a biological environment for which we have modified the general-purpose computer code Geant4. The underlying mechanism rests on the use of heavy nanoparticles delivered to specific sites, such as cancerous tumors, and treated with monoenergetic X-rays at resonant atomic and molecular transitions. X-ray irradiation of high-Z atoms results in Auger decays of photon emission and electron ejections creating multiple electron vacancies. These vacancies may be filled either be radiative decays from higher electronic shells or by excitations from the K-shell at resonant energies by an external X-ray source, as described in an accompanying paper by Pradhan et al. in this volume. Our Monte Carlo models assume normal body material embedded with a layer of gold nanoparticles. The simulation results presented in this paper demonstrate that resonant excitations via Kalpha, Kbeta, etc., transitions result in a considerable enhancement in localized X-ray energy deposition at the layer with gold nanoparticles, compared with nonresonant processes and energies. The present results could be applicable to in vivo therapy and diagnostics (theranostics) of cancerous tumors using high-Z nanoparticles and monochromatic X-ray sources according to the resonant theranostics (RT) methodology.