Deep learning-based automated liver contouring using a small sample of radiotherapy planning computed tomography images.

INTRODUCTION: No study has yet investigated the minimum amount of data required for deep learning-based liver contouring. Therefore, this study aimed to investigate the feasibility of automated liver contouring using limited data. METHODS: Radiotherapy planning Computed tomography (CT) images were subjected to various preprocessing methods, such as denoising and windowing. Segmentation was conducted using the modified Attention U-Net and Residual U-Net networks. Two different modified networks were trained separately for different training sizes. For each architecture, the model trained with the training set size that achieved the highest dice similarity coefficient (DSC) score was selected for further evaluation. Two unseen external datasets with different distributions from the training set were also used to examine the generalizability of the proposed method. RESULTS: The modified Residual U-Net and Attention U-Net networks achieved average DSCs of 97.62% and 96.48%, respectively, on the test set, using 62 training cases. The average Hausdorff distances (AHDs) for the modified Residual U-Net and Attention U-Net networks were 0.57 mm and 0.71 mm, respectively. Also, the modified Residual U-Net and Attention U-Net networks were tested on two unseen external datasets, achieving DSCs of 95.35% and 95.82% for data from another center and 95.16% and 94.93% for the AbdomenCT-1K dataset, respectively. CONCLUSION: This study demonstrates that deep learning models can accurately segment livers using a small training set. The method, utilizing simple preprocessing and modified network architectures, shows strong performance on unseen datasets, indicating its generalizability. IMPLICATIONS FOR PRACTICE: This promising result suggests its potential for automated liver contouring in radiotherapy planning.

Deep-learning Method for the Prediction of Three-Dimensional Dose Distribution for Left Breast Cancer Conformal Radiation Therapy.

AIMS: An increase in the demand of a new generation of radiotherapy planning systems based on learning approaches has been reported. At this stage, the new approach is able to improve the planning speed while saving a reasonable level of plan quality, compared with available planning systems. We believe that new achievements, such as deep-learning models, will be able to review the issue from a different point of view. MATERIALS AND METHODS: The data of 120 breast cancer patients were used to train and test the three-dimensional U-Res-Net model. The network input was computed tomography images and patients' contouring, while the patients' dose distribution was addressed as the output of the model proposed. The predicted dose distributions, created by the model for 10 test patients, were then compared with corresponding dose distributions calculated by a reliable treatment planning system. In particular, the dice similarity coefficients for different isodose volumes, dose difference and mean absolute errors (MAE) for all voxels inside the body, Dmean, D98%, D50%, D2%, V95% for planning target volume and organs at risk were calculated and were statistically analysed with the paired-samples t-test. RESULTS: The average dose difference for all patients and voxels in body was 0.60 ± 2.81%. The MAE varied from 3.85 ± 6.65% to 8.06 ± 10.00%. The average MAE for test cases was 5.71 ± 1.19%. The average dice similarity coefficients for isodose volumes was 0.91 ± 0.03. The three-dimensional gamma passing rates with 3 mm/3% criteria varied from 78.99% to 97.58% for planning target volume and organs at risk, respectively. CONCLUSIONS: The investigation showed that a deep-learning model can be applied to predict the three-dimensional dose distribution with optimal accuracy and precision for patients with left breast cancer. As further study, the model can be extended to predict dose distribution in other cancers.

Dosimetric verification of IMRT and 3D conformal treatment delivery using EPID.

PURPOSE: Electronic portal imaging devices (EPIDs) could potentially be useful for either in-vivo or pre-treatment dosimetric verification of external beam radiation therapy. The accuracy of EPID for dosimetric purposes is highly dependent on the specific method used for the determination of dose-response characteristics. The aim of this study was to develop a simple and time-saving EPID back-projection dosimetry algorithm for 2D dose verification in 3D conformal and intensity-modulated beams. METHODS: The procedure of dose reconstruction includes a first calibration step using ionization chamber measurements to convert the Electronic Portal Image (EPI) pixel values into an absorbed dose in water. Subsequently, several corrections were applied to the Portal Dose Images (PDIs) for the effect of field size, attenuator thickness, scattering radiation, beam hardening and EPID off-axis response. Furthermore, to consider tissue inhomogeneity for accurate dose reconstruction, the patient's water equivalent path length (WEPL) was calculated using a range of digitally reconstructed radiographs (DRRs) obtained at various thicknesses by Plastimatch software. The EPID-derived dose maps accuracy was assessed by comparing with the treatment planning system (TPS) calculated dose in the prostate region of Alderson phantom irradiated with 3D conformal and intensity-modulated beams. RESULTS: The gamma analysis for the dose plane showed agreements of 96.95% and 93.5% for 3D conformal and IMRT fields, respectively, with 3%/3 mm acceptance criteria. CONCLUSION: The presented algorithm can provide accurate absolute 2D dose maps for clinical use in the context of 3DCRT or IMRT Quality Assurance (QA) programs.

Evaluation of Breast Cancer Radiation Therapy Techniques in Outfield Organs of Rando Phantom with Thermoluminescence Dosimeter.

BACKGROUND: Given the importance of scattered and low doses in secondary cancer caused by radiation treatment, the point dose of critical organs, which were not subjected to radiation treatment in breast cancer radiotherapy, was measured. OBJECTIVE: The purpose of this study is to evaluate the peripheral dose in two techniques of breast cancer radiotherapy with two energies. MATERIAL AND METHODS: Eight different plans in two techniques (conventional and conformal) and two photon energies (6 and 15 MeV) were applied to Rando Alderson Phantom's DICOM images. Nine organs were contoured in the treatment planning system and specified on the phantom. To measure the photon dose, forty-eight thermoluminescence dosimeters (MTS700) were positioned in special places on the above nine organs and plans were applied to Rando phantom with Elekta presice linac. To obtain approximately the same dose distribution in the clinical organ volume, a wedge was used on planes with an energy of 6 MeV photon. RESULTS: Point doses in critical organs with 8 different plans demonstrated that scattering in low-energy photon is greater than high-energy photon. In contrast, neutron contamination in high-energy photon is not negligible. Using the wedge and shield impose greater scattering and neutron contamination on patients with low-and high-energy photon, respectively. CONCLUSION: Deciding on techniques and energies required for preparing an acceptable treatment plan in terms of scattering and neutron contamination is a key issue that may affect the probability of secondary cancer in a patient.

An Update of Couch Effect on the Attenuation of Megavoltage Radiotherapy Beam and the Variation of Absorbed Dose in the Build-up Region.

PURPOSE: Fiber carbon is the most common material used in treating couch as it causes less beam attenuation than other materials. Beam attenuation replaces build-up region, reduces skin-sparing effect and causes target volume under dosage. In this study, we aimed to evaluate beam attenuation and variation of build-up region in 550 TxT radiotherapy couch. MATERIALS AND METHODS: In this study, we utilized cylindrical PMMA Farmer chamber, DOSE-1 electrometer and set PMMA phantom in isocenter of gantry and the Farmer chamber on the phantom. Afterwards, the gantry rotated 10°, and attenuation was assessed. To measure build-up region, we used Markus chamber, Solid water phantom and DOSE-1 electrometer. Doing so, we set Solid water phantom on isocenter of gantry and placed Markus chamber in it, then we quantified the build-up region at 0° and 180° gantry angels and compared the obtained values. RESULTS: Notable attenuation and build-up region variation were observed in 550 TxT treatment table. The maximum rate of attenuation was 5.95% for 6 MV photon beam, at 5×5 cm2 field size and 130° gantry angle, while the maximum variation was 7 mm for 6 MV photon beam at 10×10 cm2 field size. CONCLUSION: Fiber carbon caused beam attenuation and variation in the build-up region. Therefore, the application of fiber carbon is recommended for planning radiotherapy to prevent skin side effects and to decrease the risk of cancer recurrence.

SIMIND Monte Carlo simulation of a single photon emission CT.

In this study, we simulated a Siemens E.CAM SPECT system using SIMIND Monte Carlo program to acquire its experimental characterization in terms of energy resolution, sensitivity, spatial resolution and imaging of phantoms using (99m)Tc. The experimental and simulation data for SPECT imaging was acquired from a point source and Jaszczak phantom. Verification of the simulation was done by comparing two sets of images and related data obtained from the actual and simulated systems. Image quality was assessed by comparing image contrast and resolution. Simulated and measured energy spectra (with or without a collimator) and spatial resolution from point sources in air were compared. The resulted energy spectra present similar peaks for the gamma energy of (99m)Tc at 140 KeV. FWHM for the simulation calculated to 14.01 KeV and 13.80 KeV for experimental data, corresponding to energy resolution of 10.01 and 9.86% compared to defined 9.9% for both systems, respectively. Sensitivities of the real and virtual gamma cameras were calculated to 85.11 and 85.39 cps/MBq, respectively. The energy spectra of both simulated and real gamma cameras were matched. Images obtained from Jaszczak phantom, experimentally and by simulation, showed similarity in contrast and resolution. SIMIND Monte Carlo could successfully simulate the Siemens E.CAM gamma camera. The results validate the use of the simulated system for further investigation, including modification, planning, and developing a SPECT system to improve the quality of images.