Can we predict overall survival using machine learning algorithms at 3-months for brain metastases from non-small cell lung cancer after gamma knife radiosurgery?

Gamma knife radiosurgery (GRKS) is widely used for patients with brain metastases; however, predictions of overall survival (OS) within 3-months post-GKRS remain imprecise. Specifically, more than 10% of non-small cell lung cancer (NSCLC) patients died within 8 weeks of post-GKRS, indicating potential overtreatment. This study aims to predict OS within 3-months post-GKRS using machine learning algorithms, and to identify prognostic features in NSCLC patients. We selected 120 NSCLC patients who underwent GKRS at Chungbuk National University Hospital. They were randomly assigned to training group (n = 80) and testing group (n = 40) with 14 features considered. We used 3 machine learning (ML) algorithms (Decision tree, Random forest, and Boosted tree classifier) to predict OS within 3-months for NSCLC patients. And we extracted important features and permutation features. Data validation was verified by physician and medical physicist. The accuracy of the ML algorithms for predicting OS within 3-months was 77.5% for the decision tree, 72.5% for the random forest, and 70% for the boosted tree classifier. The important features commonly showed age, receiving chemotherapy, and pretreatment each algorithm. Additionally, the permutation features commonly showed tumor volume (>10 cc) and age as critical factors each algorithm. The decision tree algorithm exhibited the highest accuracy. Analysis of the decision tree visualized data revealed that patients aged (>71 years) with tumor volume (>10 cc) were increased risk of mortality within 3-months. The findings suggest that ML algorithms can effectively predict OS within 3-months and identify crucial features in NSCLC patients. For NSCLC patients with poor prognoses, old age, and large tumor volumes, GKRS may not be a desirable treatment.

Optimal mask fixation method for frameless radiosurgery with Leksell Gamma Knife IconTM.

The Leksell Gamma Knife (LGK) IconTM is used for mask-based and frame-based fixation. The mask fixation provides a noninvasive method. However, an optimal mask fixation method is yet to be established. We evaluated the characteristics of three mask fixation methods (Plain, Folded, and Wide) for the LGK IconTM . Force-sensitive resistor sensors were attached to the forehead, supraorbital, zygoma, mandible, and occipital bone of the phantom, and digital humidity and temperature sensors were attached to both temporal lobes. Cone-beam computed tomography (CBCT) and high-definition motion management (HDMM) for each mask fixation method were used to evaluate the phantom motion during the initial application. Subsequently, the mask was removed and reapplied on the second (1st reapplication) and third days (2nd reapplication). In the initial application, forces acting on most portions of the phantom were stabilized within 1.5 h. The largest force acted on the occipital bone for the Plain and Wide methods and on the mandible for the Folded method. The temperature rapidly approaches the initial temperature, whereas the humidity gradually approached the initial humidity in all fixation methods. The Folded method exhibited a significantly lower translation along the Y-axis of the Leksell coordinate system, and rotations along all axes were under 0.5°. The HDMM values remained at 0.1 mm for all fixation methods. In the reapplications, the force acting on the occipital bone was significantly greater than that during the initial application for all mask fixation methods; the temperature and humidity remained unchanged. All mask fixation methods in the 1st reapplication were not significantly different from those in the 2nd reapplication. The Folded method is recommended as an optimal mask fixation for patients who require tight fixation; the Wide method can be considered if patient comfort is a priority.

Avoiding a Collision in Gamma Knife Radiosurgery : A Modified Mask Fixation Method.

OBJECTIVE: The latest version of the Leksell Gamma Knife IconTM allows for mask- and frame-based fixation. Although mask fixation provides fractionated treatment and immobilization using a noninvasive method, it is not free from collision. The authors investigated the collision problem with a modified mask fixation method. METHODS: This study presents a case of two meningiomas in the frontal area, where a collision occurs in the occipital area. A modified mask fixation method was introduced to avoid the collision : first, the edges of the head cushion were cut off and polystyrene beads with a diameter of approximately 5 cm were removed. Next, the head cushion was sealed using a stapler. Finally, the head cushion was flattened in the adapter. We compared the shot coordinates, 3-dimensional (3D) error, clearance distance, and vertical depth of the head cushion between the initial and modified mask fixations. RESULTS: When comparing the initial and modified mask fixations, the difference in the shot coordinates was +10.5 mm along the y-axis, the difference in the 3D error was approximately 18 mm, and the difference in clearance was -10.2 mm. The head cushion was approximately 8 mm deeper in the modified mask fixation. CONCLUSION: Based on these findings, we recommend a modified mask fixation method for gamma knife radiosurgery using ICON with a collision.

Virtual randomized study comparing lobectomy and particle beam therapy for clinical stage IA non-small cell lung cancer in operable patients.

To the best of our knowledge there have been no randomized controlled trials comparing lobectomy-a standard treatment for patients with early-stage non-small cell lung cancer (NSCLC)-and particle beam therapy (PBT), the best performing existing radiotherapy. We conducted a virtual randomized trial in medically operable patients with stage IA NSCLC to compare lobectomy and PBT effectiveness. A Markov model was developed to predict life expectancy after lobectomy and PBT in a cohort of patients with stage IA NSCLC. Ten thousand virtual patients were randomly assigned to each group. Sensitivity analyses were performed as model variables and scenarios changed to determine which treatment strategy was best for improving life expectancy. All estimated model parameters were determined using variables extracted from a systematic literature review of previously published articles. The preferred strategy differed depending on patient age. In young patients, lobectomy showed better life expectancy than that of PBT. The difference in life expectancy between lobectomy and PBT was statistically insignificant in older patients. Our model predicted lobectomy as the preferred strategy when operative mortality was under 5%. However, the preferred strategy changed to PBT if operative mortality post lobectomy was over 5%. For medically operable patients with stage IA NSCLC, our Markov model revealed the preferred strategy of lobectomy or PBT regarding operative mortality changed with varying age and comorbidity. Until randomized controlled trial results become available, we hope the current results will provide a rationale background for clinicians to decide treatment modalities for patients with stage IA NSCLC.

Signal Enhancement of Low Magnetic Field Magnetic Resonance Image Using a Conventional- and Cyclic-Generative Adversarial Network Models With Unpaired Image Sets.

In this study, the signal enhancement ratio of low-field magnetic resonance (MR) images was investigated using a deep learning-based algorithm. Unpaired image sets (0.06 Tesla and 1.5 Tesla MR images for different patients) were used in this study following three steps workflow. In the first step, the deformable registration of a 1.5 Tesla MR image into a 0.06 Tesla MR image was performed to ensure that the shapes of the unpaired set matched. In the second step, a cyclic-generative adversarial network (GAN) was used to generate a synthetic MR image of the original 0.06 Tesla MR image based on the deformed or original 1.5 Tesla MR image. Finally, an enhanced 0.06 Tesla MR image could be generated using the conventional-GAN with the deformed or synthetic MR image. The results from the optimized flow and enhanced MR images showed significant signal enhancement of the anatomical view, especially in the nasal septum, inferior nasal choncha, nasopharyngeal fossa, and eye lens. The signal enhancement ratio, signal-to-noise ratio (SNR) and correlation factor between the original and enhanced MR images were analyzed for the evaluation of the image quality. A combined method using conventional- and cyclic-GANs is a promising approach for generating enhanced MR images from low-magnetic-field MR.

Low-dose CBCT reconstruction via joint non-local total variation denoising and cubic B-spline interpolation.

This study develops an improved Feldkamp-Davis-Kress (FDK) reconstruction algorithm using non-local total variation (NLTV) denoising and a cubic B-spline interpolation-based backprojector to enhance the image quality of low-dose cone-beam computed tomography (CBCT). The NLTV objective function is minimized on all log-transformed projections using steepest gradient descent optimization with an adaptive control of the step size to augment the difference between a real structure and noise. The proposed algorithm was evaluated using a phantom data set acquired from a low-dose protocol with lower milliampere-seconds (mAs).The combination of NLTV minimization and cubic B-spline interpolation rendered the enhanced reconstruction images with significantly reduced noise compared to conventional FDK and local total variation with anisotropic penalty. The artifacts were remarkably suppressed in the reconstructed images. Quantitative analysis of reconstruction images using low-dose projections acquired from low mAs showed a contrast-to-noise ratio with spatial resolution comparable to images reconstructed using projections acquired from high mAs. The proposed approach produced the lowest RMSE and the highest correlation. These results indicate that the proposed algorithm enables application of the conventional FDK algorithm for low mAs image reconstruction in low-dose CBCT imaging, thereby eliminating the need for more computationally demanding algorithms. The substantial reductions in radiation exposure associated with the low mAs projection acquisition may facilitate wider practical applications of daily online CBCT imaging.

Deep Learning-Based Decision-Tree Classifier for COVID-19 Diagnosis From Chest X-ray Imaging.

The global pandemic of coronavirus disease 2019 (COVID-19) has resulted in an increased demand for testing, diagnosis, and treatment. Reverse transcription polymerase chain reaction (RT-PCR) is the definitive test for the diagnosis of COVID-19; however, chest X-ray radiography (CXR) is a fast, effective, and affordable test that identifies the possible COVID-19-related pneumonia. This study investigates the feasibility of using a deep learning-based decision-tree classifier for detecting COVID-19 from CXR images. The proposed classifier comprises three binary decision trees, each trained by a deep learning model with convolution neural network based on the PyTorch frame. The first decision tree classifies the CXR images as normal or abnormal. The second tree identifies the abnormal images that contain signs of tuberculosis, whereas the third does the same for COVID-19. The accuracies of the first and second decision trees are 98 and 80%, respectively, whereas the average accuracy of the third decision tree is 95%. The proposed deep learning-based decision-tree classifier may be used in pre-screening patients to conduct triage and fast-track decision making before RT-PCR results are available.

Proton-radiography-based quality assurance of proton range compensator.

The aim of this work was to study the feasibility of proton radiography (pRad) as a patient-specific range compensator (RC) quality assurance (QA) tool and to validate its clinical utility by performing QA on RCs having three kinds of possible defects. In order to achieve pRad for a single EBT film, proton beam currents were modulated with new weighting factors, maximizing the linearity of optical-density-to-thickness ratio. Two RCs, examined to be accurately manufactured as planned, were selected to estimate the feasibility of our pRad. The optical densities of the EBT film on which the RC was irradiated with the modulated proton beam were digitized to pixel values (pv) and then converted to thickness using a thickness-pv calibration curve. The thickness information on the pRad was compared with plan data that had been extracted from treatment planning system. The mean thickness difference (TD) over the flat RC regions was calculated as 0.39 mm, and the standard deviation as 0.22 mm, and the proton scattering effect was analyzed by step phantom measurement. Even proton scattering effected a TD of over 1 mm in the large gradient region, the percentage of pixels over the acceptance criterion was only within 1.11% and 3.49%, respectively, when a 1 mm distance to agreement tolerance limit was applied. The QA results for both precisely and imprecisely manufactured RCs demonstrated the high potential utility and clinical applicability of the pRad-based RC QA tool.

Application of Cerenkov radiation generated in plastic optical fibers for therapeutic photon beam dosimetry.

A Cerenkov fiber-optic dosimeter (CFOD) is fabricated using plastic optical fibers to measure Cerenkov radiation induced by a therapeutic photon beam. We measured the Cerenkov radiation generated in optical fibers in various irradiation conditions to evaluate the usability of Cerenkov radiation for a photon beam therapy dosimetry. As a results, the spectral peak of Cerenkov radiation was measured at a wavelength of 515 nm, and the intensity of Cerenkov radiation increased linearly with increasing irradiated length of the optical fiber. Also, the intensity peak of Cerenkov radiation was measured in the irradiation angle range of 30 to 40 deg. In the results of Monte Carlo N-particle transport code simulations, the relationship between fluxes of electrons over Cerenkov threshold energy and energy deposition of a 6 MV photon beam had a nearly linear trend. Finally, percentage depth doses for the 6 MV photon beam could be obtained using the CFOD and the results were compared with those of an ionization chamber. Here, the mean dose difference was about 0.6%. It is anticipated that the novel and simple CFOD can be effectively used for measuring depth doses in radiotherapy dosimetry.

Estimation of the secondary cancer risk induced by diagnostic imaging radiation during proton therapy.

We have estimated the secondary cancer risk (SCR) introduced by image-guided procedures during proton therapy. The physical dose from imaging radiation and the corresponding organ equivalent dose were calculated for the case of a lumbar spine patient. The maximum physical dose delivered to the patient during the imaging procedure was estimated to be ~0.35% of the prescribed dose of 46 Gy. However, this small imaging dose substantially raised the radiation-induced SCR by ~8%. In addition, the clinical benefit (improved accuracy during the procedure) and costs (extra SCR) associated with image-guided procedures were quantitatively modelled by systematically investigating the changes in SCR as a function of the prescribed dose, treatment target volume and imaging field size. The results showed that the SCR varied sensitively with the volume receiving the imaging and the therapeutic radiation, whereas the SCR depended to a lesser extent on the magnitude of the applied therapeutic radiation. These results showed that the additional SCR introduced by imaging radiation could be efficiently reduced by minimizing the imaging field size during image-guided procedures.

Secondary radiation doses of intensity-modulated radiotherapy and proton beam therapy in patients with lung and liver cancer.

PURPOSE: To compare the secondary radiation doses following intensity-modulated radiotherapy (IMRT) and proton beam therapy (PBT) in patients with lung and liver cancer. METHODS AND MATERIALS: IMRT and PBT were planned for three lung cancer and three liver cancer patients. The treatment beams were delivered to phantoms and the corresponding secondary doses during irradiation were measured at various points 20-50 cm from the beam isocenter using ion chamber and CR-39 detectors for IMRT and PBT, respectively. RESULTS: The secondary dose per Gy (i.e., a treatment dose of 1Gy) from PBT for lung and liver cancer, measured 20-50 cm from the isocenter, ranged from 0.17 to 0.086 mGy. The secondary dose per Gy from IMRT, however, ranged between 5.8 and 1.0 mGy, indicating that PBT is associated with a smaller dose of secondary radiation than IMRT. The internal neutron dose per Gy from PBT for lung and liver cancer, 20-50 cm from the isocenter, ranged from 0.03 to 0.008 mGy. CONCLUSIONS: The secondary dose from PBT is less than or compatible to the secondary dose from conventional IMRT. The internal neutron dose generated by the interaction between protons and body material is generally much less than the external neutron dose from the treatment head.

Development of a dual modality imaging system: a combined gamma camera and optical imager.

Several groups have reported the development of dual modality Gamma camera/optical imagers, which are useful tools for investigating biological processes in experimental animals. While previously reported dual modality imaging instrumentation usually employed a separated gamma camera and optical imager, we designed a detector using a position sensitive photomultiplier tube (PSPMT) that is capable of imaging both gamma rays and optical photons for combined gamma camera and optical imager. The proposed system consists of a parallel-hole collimator, an array-type crystal and a PSPMT. The top surface of the collimator and array crystals is left open to allow optical photons to reach the PSPMT. Pulse height spectra and planar images were obtained using a Tc-99m source and a green LED to estimate gamma and optical imaging performances. When both gamma rays and optical photon signals were detected, the signal interferences caused by each other signal were evaluated. A mouse phantom and an ICR mouse containing a gamma ray and optical photon source were imaged to assess the imaging capabilities of the system. The sensitivity, energy resolution and spatial resolution of the gamma image acquired using Tc-99m were 1.1 cps/kBq, 26% and 2.1 mm, respectively. The spatial resolution of the optical image acquired with an LED was 3.5 mm. Signal-to-signal interference due to the optical photon signal in the gamma pulse height spectrum was negligible. However, the pulse height spectrum of the optical photon signal was found to be affected by the gamma signal, and was obtained between signals generated by gamma rays with a correction using a veto gate. Gamma ray and optical photon images of the mouse phantom and ICR mouse were successfully obtained using the single detector. The experimental results indicated that both optical photon and gamma ray imaging are feasible using a detector based on the proposed PSPMT.

Performance improvement of small gamma camera using NaI(Tl) plate and position sensitive photo-multiplier tubes.

The purpose of this study was to improve the performance of a small gamma camera, utilizing a NaI(Tl) plate and a 5" position sensitive PMT. We attempted to build a NaI(Tl) plate crystal system which retained all its advantages, while at the same time integrating some of the advantages inherent in an array-type scintillation crystal system. Flood images were obtained with a lead hole mask, and position mapping was performed by detecting hole positions in the flood image. Energy calibration was performed using the energy spectra obtained from each hole position. Flood correction was performed using a uniformity correction table containing the relative efficiency of each image element. The spatial resolution was improved about 16% after correction at the centre field of view. Resolution deterioration at the outer field of view (OFOV) was considerably ameliorated, from 6.7 mm to 3.2 mm after correction. The sensitivity at the OFOV was also increased after correction, from 0.7 cps microCi(-1) to 2.0 cps microCi(-1). The correction also improved uniformity, from 5.2% to 2.1%, and linearity, from 0.5 mm to 0 mm. The results of this study indicate that the revised correction method can be employed to considerably improve the performance of a small gamma camera using a NaI(Tl) plate-type crystal. This method also provides high spatial resolution and linearity, like array-type crystals do, while retaining the specific advantages of plate-type crystals.