Development of porous structure for broadening Bragg-peak in scanning carbon-ion radiotherapy: Monte Carlo simulation and experimental validation.

PURPOSE: The present study aimed to develop a porous structure with plug-ins (PSP) to broaden the Bragg peak width (BPW, defined as the distance in water between the proximal and distal 80% dose) of the carbon ion beam while maintaining a sharp distal falloff width (DFW, defined as the distance along the beam axis where the dose in water reduces from 80% to 20%). METHODS: The binary voxel models of porous structure (PS) and PSP were established in the Monte Carlo code FLUKA and the corresponding physical models were manufactured by 3D printing. Both experiment and simulation were performed for evaluating the modulation capacity of PS and PSP. BPWs and DFWs derived from each integral depth dose curves were compared. Fluence homogeneity of 430 MeV/u carbon-ion beam passing through the PSP was recorded by analyzing radiochromic films at six different locations downstream the PSP in the experiment. Additionally, by changing the beam spot size and incident position on the PSP, totally 48 different carbon-ion beams were simulated and corresponding deviations of beam metrics were evaluated to test the modulating stability of PSP. RESULTS: According to the measurement data, the use of PSP resulted in an average increase of 0.63 mm in BPW and a decrease of 0.74 mm in DFW compared to PS. The 2D radiation field inhomogeneities were lower than 3 % when the beam passing through a ≥ 10 cm PMMA medium. Furthermore, employing a spot size of ≥ 6 mm ensures that beam metric deviations, including BPW, DFW, and range, remain within a deviation of 0.1 mm across various incident positions. CONCLUSION: The developed PSP demonstrated its capability to effectively broaden the BPW of carbon ion beams while maintaining a sharp DFW comparing to PS. The superior performance of PSP, indicates its potential for clinical use in the future.

Inter-fractional portability of deep learning models for lung target tracking on cine imaging acquired in MRI-guided radiotherapy.

MRI-guided radiotherapy systems enable beam gating by tracking the target on planar, two-dimensional cine images acquired during treatment. This study aims to evaluate how deep-learning (DL) models for target tracking that are trained on data from one fraction can be translated to subsequent fractions. Cine images were acquired for six patients treated on an MRI-guided radiotherapy platform (MRIdian, Viewray Inc.) with an onboard 0.35 T MRI scanner. Three DL models (U-net, attention U-net and nested U-net) for target tracking were trained using two training strategies: (1) uniform training using data obtained only from the first fraction with testing performed on data from subsequent fractions and (2) adaptive training in which training was updated each fraction by adding 20 samples from the current fraction with testing performed on the remaining images from that fraction. Tracking performance was compared between algorithms, models and training strategies by evaluating the Dice similarity coefficient (DSC) and 95% Hausdorff Distance (HD95) between automatically generated and manually specified contours. The mean DSC for all six patients in comparing manual contours and contours generated by the onboard algorithm (OBT) were 0.68 ± 0.16. Compared to OBT, the DSC values improved 17.0 - 19.3% for the three DL models with uniform training, and 24.7 - 25.7% for the models based on adaptive training. The HD95 values improved 50.6 - 54.5% for the models based on adaptive training. DL-based techniques achieved better tracking performance than the onboard, registration-based tracking approach. DL-based tracking performance improved when implementing an adaptive strategy that augments training data fraction-by-fraction.

Implementation and evaluation of an iterative-based algorithm for automatic beam angle optimization in breast cancer treatment planning.

INTRODUCTION: A beam angle optimization (BAO) algorithm was developed to evaluate its clinical feasibility and investigate the impact of varying BAO constraints on breast cancer treatment plans. MATERIALS AND METHODS: A two-part study was designed. In part 1, we retrospectively selected 20 patients treated with radiotherapy after breast-conserving surgery. For each patient, BAO plans were designed using beam angles optimized by the BAO algorithm and the same optimization constraints as manual plans. Dosimetric indices were compared between BAO and manual plans. In part 2, fifteen patients with left breast cancer were included. For each patient, three distinct cardiac constraints (mean heart dose < 5 Gy, 3 Gy or 1 Gy) were established during the BAO process to obtain three optimized beam sets which were marked as BAO_H1, BAO_H3, BAO_H5, respectively. These sets of beams were then utilized under identical IMRT constraints for planning. Comparative analysis was conducted among the three groups of plans. RESULTS: For part 1, no significant differences were observed between BAO plans and manual plans in all dosimetric indices, except for ipsilateral lung V5, where BAO plans performed slightly better than manual plans (35.5% ± 5.6% vs 36.9% ± 4.3%, p = 0.034). For part 2, Stricter BAO heart constraints resulted in more perpendicular beams. However, there was no significant difference between BAO_H1, BAO_H3 and BAO_H5 with the same IMRT constraint in the heart dose. Meanwhile, the left lung dose was increased while the right breast and lung doses were decreased with stricter heart constraints in BAO. When mean heart dose < 5 Gy in IMRT constraint, the mean dose to the right lung was decreased from 0.46 Gy for BAO_H5 to 0.33 Gy for BAO_H1 (p = 0.027). CONCLUSIONS: The BAO algorithm can achieve quality plans comparable to manual plans. IMRT constraints dominate the final plan dose, while varying BAO constraints alter the trade-off among structures, providing an additional degree of freedom in planning design.

A simple method to import CAD mesh format models in FLUKA.

BACKGROUND: Monte Carlo (MC) code FLUKA possesses widespread usage and accuracy in the simulation of particle beam radiotherapy. However, the conversion from computer-aided design (CAD) mesh format models to FLUKA readable geometries could not be implemented directly and conveniently. A simple method was required to be developed. PURPOSE: The present study proposed a simple method to voxelize CAD mesh format files by using a Python-based script and establishing geometric models in FLUKA. METHODS: Five geometric models including cube, sphere, cone, ridge filter (RGF), and 1D-Ripple Filter (1D-RiFi) were created and exported as CAD mesh format files (.stl). An open-source Python-based script was used to convert them into voxels by endowing X, Y, and Z (following the Cartesian coordinates system) of solid materials in the three-dimensional (3D) grid. A FLUKA (4-2.2, CERN) predefined routine was used to establish the voxelized geometry model (VGM), while Flair (3.2-1, CERN) was used to build the direct geometry model (DGM) in FLUKA for comparison purposes. Uniform carbon ion radiation fields 8×8 cm3 and 4×4 cm3 were generated to transport through the five pairs of models, 2D and 3D dose distributions were compared. The integral depth dose (IDD) in water of three different energy levels of carbon ion beams transported through 1D-RiFis were also simulated and compared. Moreover, the volume between CAD mesh and VGMs, as well as the computing speed between FLUKA DGMs and VGMs were simultaneously recorded. RESULTS: The volume differences between VGMs and CAD mesh models were not more than 0.6%. The maximum mean point-to-point deviation of IDD distribution was 0.7% ± 0.51% (mean ± standard deviation). The 3D dose Gamma-index passing rates were never lower than 97% with criteria of 1%-1 mm. The difference in computing CPU time was 2.89% ± 0.22 on average. CONCLUSIONS: The present study proposed and verified a Python-based method for converting CAD mesh format files into VGMs and establishing them in FLUKA simply as well as accurately.

Geometric and dosimetric evaluation of deep learning based auto-segmentation for clinical target volume on breast cancer.

BACKGROUND: Recently, target auto-segmentation techniques based on deep learning (DL) have shown promising results. However, inaccurate target delineation will directly affect the treatment planning dose distribution and the effect of subsequent radiotherapy work. Evaluation based on geometric metrics alone may not be sufficient for target delineation accuracy assessment. The purpose of this paper is to validate the performance of automatic segmentation with dosimetric metrics and try to construct new evaluation geometric metrics to comprehensively understand the dose-response relationship from the perspective of clinical application. MATERIALS AND METHODS: A DL-based target segmentation model was developed by using 186 manual delineation modified radical mastectomy breast cancer cases. The resulting DL model were used to generate alternative target contours in a new set of 48 patients. The Auto-plan was reoptimized to ensure the same optimized parameters as the reference Manual-plan. To assess the dosimetric impact of target auto-segmentation, not only common geometric metrics but also new spatial parameters with distance and relative volume ( R V ${R}_V$ ) to target were used. Correlations were performed using Spearman's correlation between segmentation evaluation metrics and dosimetric changes. RESULTS: Only strong (|R2 | > 0.6, p < 0.01) or moderate (|R2 | > 0.4, p < 0.01) Pearson correlation was established between the traditional geometric metric and three dosimetric evaluation indices to target (conformity index, homogeneity index, and mean dose). For organs at risk (OARs), inferior or no significant relationship was found between geometric parameters and dosimetric differences. Furthermore, we found that OARs dose distribution was affected by boundary error of target segmentation instead of distance and R V ${R}_V$ to target. CONCLUSIONS: Current geometric metrics could reflect a certain degree of dose effect of target variation. To find target contour variations that do lead to OARs dosimetry changes, clinically oriented metrics that more accurately reflect how segmentation quality affects dosimetry should be constructed.

Toward real-time automatic treatment planning (RTTP) with a one-step 3D fluence map prediction method and (nonorthogonal) convolution technique.

PURPOSE: To establish and evaluate a (quasi) real-time automated treatment planning (RTTP) strategy utilizing a one-step full 3D fluence map prediction model based on a nonorthogonal convolution operation for rectal cancer radiotherapy. METHODS: The RTTP approach directly extracts 3D projections from volumetric CT and anatomical data according to the beam incident direction. A 3D deep learning model with a nonorthogonal convolution operation was established that takes projections in cone beam space as input, extracts the features along and around the ray-trace path, and outputs a predicted fluence map (PFM) for each beam. The PFM is then converted to the MLC sequence with deliverable MUs to generate the final treatment plan. A total of 314 rectal adenocarcinoma patients with 2198 projection data samples were used in model training and validation. An extra 20 patients were used to test the feasibility of the RTTP method by comparing the plan quality, efficiency, deliverability performance, and physician blinded review results with the manual plans. RESULTS: Overall, the RTTP plans met the clinical dose criteria for target coverage, conformity, homogeneity, and organ-at-risk dose sparing. Compared to manual plans, the RTTP plans showed increases in PTV D1% by only 2.33% (p < 0.001) and a decrease in PTV D99% by 0.45% (p < 0.05). The RTTP plans showed a dose increase in the bladder, with a V50 of 14.01 ± 11.75% vs. 10.74 ± 8.51%, respectively, and no significant increases in the femoral head with the mean dose. The planning efficiency was improved in RTTP planning, with 39 s vs. 944 s in fluence map generation; the deliverability performance was saved by 1.91% (p < 0.001) in total MU. According to the blinded plan review by our physician, 55% of RTTP plans can be directly used in clinical radiotherapy treatment. CONCLUSION: The quasi RTTP method improves the planning efficiency and deliverability performance while maintaining a plan quality close to that of the optimized manual plans in rectal radiotherapy.

[Dosimetric Impact of Titanium Alloy Implant in Spinal IMRT Plan].

The purpose of this study is to establish and apply a correction method for titanium alloy implant in spinal IMRT plan, a corrected CT-density table was revised from normal CT-density table to include the density of titanium alloy implant. Dose distribution after and before correction were calculated and compared to evaluate the dose deviation. Plans were also copied to a spinal cancer simulation phantom. A titanium alloy fixation system for spine was implanted in this phantom. Plans were recalculated and compared with the measurement result. The result of this study shows that the max dose of spinal cord showed significant difference after correction, and the deviation between calculation results and measurement results was reduced after correction. The method for expanding the range CT-density table, which means that the density of titanium alloy was included, can reduce the error in calculation.

Technical note: First implementation of a one-stop solution of radiotherapy with full-workflow automation based on CT-linac combination.

BACKGROUND: Radiotherapy initiation is a laborious and time-consuming process that involves multiple steps and units. Workflow automation is in demand to improve the work efficiency and patient experience. PURPOSE: The purposes of this study are to describe the technical characteristics and clinical performance of an AI-powered one-stop radiotherapy workflow for initial treatment based on CT-linac combination, and provide insight into the behavior of full-workflow automation in radiotherapy. METHODS: Based on a CT-integrated linear accelerator and AI model implementation, the so-called "All-in-One" workflow incorporates routine procedures from simulation, autosegmentation, autoplanning, image guidance, beam delivery, and in vivo quality assurance (QA) into one scheme, while the patient is on the treatment couch. Clinical outcomes of the new workflow were evaluated for 10 enrolled patients with rectal cancer. RESULTS: For the enrolled patients, manual modifications of the autosegmented target volumes were necessary. The Dice similarity coefficient and 95% Hausdorff distance before and after the modifications were 0.892 ± 0.061 and 18.2 ± 13.0 mm, respectively. The autosegmented normal tissues and automatic plans were clinically acceptable without any modifications or reoptimization. The pretreatment IGRT corrections were within 2 mm in all directions, and the EPID-based in vivo QA showed γ passing rate of above 97% (3%/3 mm/10% threshold) at all the checkpoints, better than the results of rectal patients who followed a routine workflow. The duration of the whole process was 23.2 ± 3.5 minutes for the enrolled patients, depending mostly on the time required for manual modification and plan evaluation. CONCLUSION: The All-in-One workflow enables full-process automation of radiotherapy via seamless procedure integration. Compared to the routine workflow, the one-stop solution shortens the time scale it takes to ready the first treatment from days to minutes, significantly improving the patient experience and the workflow efficiency, and it also shows potential to facilitate clinical application of online adaptive replanning.

An overview of artificial intelligence in medical physics and radiation oncology.

Artificial intelligence (AI) is developing rapidly and has found widespread applications in medicine, especially radiotherapy. This paper provides a brief overview of AI applications in radiotherapy, and highlights the research directions of AI that can potentially make significant impacts and relevant ongoing research works in these directions. Challenging issues related to the clinical applications of AI, such as robustness and interpretability of AI models, are also discussed. The future research directions of AI in the field of medical physics and radiotherapy are highlighted.

An attention base U-net for parotid tumor autosegmentation.

A parotid neoplasm is an uncommon condition that only accounts for less than 3% of all head and neck cancers, and they make up less than 0.3% of all new cancers diagnosed annually. Due to their nonspecific imaging features and heterogeneous nature, accurate preoperative diagnosis remains a challenge. Automatic parotid tumor segmentation may help physicians evaluate these tumors. Two hundred eighty-five patients diagnosed with benign or malignant parotid tumors were enrolled in this study. Parotid and tumor tissues were segmented by 3 radiologists on T1-weighted (T1w), T2-weighted (T2w) and T1-weighted contrast-enhanced (T1wC) MR images. These images were randomly divided into two datasets, including a training dataset (90%) and an validation dataset (10%). A 10-fold cross-validation was performed to assess the performance. An attention base U-net for parotid tumor autosegmentation was created on the MRI T1w, T2 and T1wC images. The results were evaluated in a separate dataset, and the mean Dice similarity coefficient (DICE) for both parotids was 0.88. The mean DICE for left and right tumors was 0.85 and 0.86, respectively. These results indicate that the performance of this model corresponds with the radiologist's manual segmentation. In conclusion, an attention base U-net for parotid tumor autosegmentation may assist physicians to evaluate parotid gland tumors.

Deep learning empowered volume delineation of whole-body organs-at-risk for accelerated radiotherapy.

In radiotherapy for cancer patients, an indispensable process is to delineate organs-at-risk (OARs) and tumors. However, it is the most time-consuming step as manual delineation is always required from radiation oncologists. Herein, we propose a lightweight deep learning framework for radiotherapy treatment planning (RTP), named RTP-Net, to promote an automatic, rapid, and precise initialization of whole-body OARs and tumors. Briefly, the framework implements a cascade coarse-to-fine segmentation, with adaptive module for both small and large organs, and attention mechanisms for organs and boundaries. Our experiments show three merits: 1) Extensively evaluates on 67 delineation tasks on a large-scale dataset of 28,581 cases; 2) Demonstrates comparable or superior accuracy with an average Dice of 0.95; 3) Achieves near real-time delineation in most tasks with <2 s. This framework could be utilized to accelerate the contouring process in the All-in-One radiotherapy scheme, and thus greatly shorten the turnaround time of patients.

Evaluation of a hybrid automatic planning solution for rectal cancer.

BACKGROUND: Script-based planning and knowledge-based planning are two kinds of automatic planning solutions. Hybrid automatic planning may integrate the advantages of both solutions and provide a more robust automatic planning solution in the clinic. In this study, we evaluated and compared a commercially available hybrid planning solution with manual planning and script-based planning. METHODS: In total, 51 rectal cancer patients in our institution were enrolled in this study. Each patient generated 7 plans: one clinically accepted manual plan ([Formula: see text]), three script-based plans and three hybrid plans generated with the volumetric-modulated arc therapy technique and 3 different clinical goal settings: easy, moderate and hard ([Formula: see text], [Formula: see text], [Formula: see text], [Formula: see text], [Formula: see text] and [Formula: see text]). Planning goals included planning target volume (PTV) Dmax, bladder Dmean and femur head Dmean. The PTV prescription was the same (50.00 Gy) for the 3 goal settings. The hard setting required a lower PTV Dmax and stricter organ at risk (OAR) dose, while the easy setting was the opposite. Plans were compared using dose metrics and plan quality metric (PQM) scores, including bladder D15 and D50, left and right femur head D25 and D40, PTV D2, D98, CI (conformity index) and HI (homogeneity index). RESULTS: Compared to manual planning, hybrid planning with all settings significantly reduced the OAR dose (p < 0.05, paired t-test or Wilcoxon signed rank test) for all dose-volume indices, except D25 of the left femur head. For script-based planning, [Formula: see text] significantly increased the OAR dose for the femur head and D2 and the PTV homogeneity index (p < 0.05, paired t-test or Wilcoxon signed rank test). Meanwhile, the maximum dose of the PTV was largely increased with hard script-based planning (D2 = 56.06 ± 7.57 Gy). For all three settings, the comparison of PQM between hybrid planning and script-based planning showed significant differences, except for D25 of the left femur head and PTV D2. The total PQM showed that hybrid planning could provide a better and more robust plan quality than script-based planning. CONCLUSIONS: The hybrid planning solution was manual-planning comparable for rectal cancer. Hybrid planning can provide a better and more robust plan quality than script-based planning.

A clinically relevant online patient QA solution with daily CT scans and EPID-basedin vivodosimetry: a feasibility study on rectal cancer.

Objective.Adaptive radiation therapy (ART) could protect organs at risk (OARs) while maintain high dose coverage to targets. However, there is still a lack of efficient online patient quality assurance (QA) methods, which is an obstacle to large-scale adoption of ART. We aim to develop a clinically relevant online patient QA solution for ART using daily CT scans and EPID-basedin vivodosimetry.Approach.Ten patients with rectal cancer at our center were included. Patients' daily CT scans and portal images were collected to generate reconstructed 3D dose distributions. Contours of targets and OARs were recontoured on these daily CT scans by a clinician or an auto-segmentation algorithm, then dose-volume indices were calculated, and the percent deviation of these indices to their original plans were determined. This deviation was regarded as the metric for clinically relevant patient QA. The tolerance level was obtained using a 95% confidence interval of the QA metric distribution. These deviations could be further divided into anatomically relevant or delivery relevant indicators for error source analysis. Finally, our QA solution was validated on an additional six clinical patients.Main results.In rectal cancer, the 95% confidence intervals of the QA metric for PTV ΔD95(%) were [-3.11%, 2.35%], and for PTV ΔD2(%) were [-0.78%, 3.23%]. In validation, 68% for PTV ΔD95(%), and 79% for PTV ΔD2(%) of the 28 fractions are within tolerances of the QA metrics. one patient's dosimetric impact of anatomical variations during treatment were observed through the source of error analysis.Significance.The online patient QA solution using daily CT scans and EPID-basedin vivodosimetry is clinically feasible. Source of error analysis has the potential for distinguishing sources of error and guiding ART for future treatments.

Automatic Tumor Grading on Colorectal Cancer Whole-Slide Images: Semi-Quantitative Gland Formation Percentage and New Indicator Exploration.

Tumor grading is an essential factor for cancer staging and survival prognostication. The widely used the WHO grading system defines the histological grade of CRC adenocarcinoma based on the density of glandular formation on whole-slide images (WSIs). We developed a fully automated approach for stratifying colorectal cancer (CRC) patients' risk of mortality directly from histology WSI relating to gland formation. A tissue classifier was trained to categorize regions on WSI as glands, stroma, immune cells, background, and other tissues. A gland formation classifier was trained on expert annotations to categorize regions as different degrees of tumor gland formation versus normal tissues. The glandular formation density can thus be estimated using the aforementioned tissue categorization and gland formation information. This estimation was called a semi-quantitative gland formation ratio (SGFR), which was used as a prognostic factor in survival analysis. We evaluated gland formation percentage and validated it by comparing it against the WHO cutoff point. Survival data and gland formation maps were then used to train a spatial pyramid pooling survival network (SPPSN) as a deep survival model. We compared the survival prediction performance of estimated gland formation percentage and the SPPSN deep survival grade and found that the deep survival grade had improved discrimination. A univariable Cox model for survival yielded moderate discrimination with SGFR (c-index 0.62) and deep survival grade (c-index 0.64) in an independent institutional test set. Deep survival grade also showed better discrimination performance in multivariable Cox regression. The deep survival grade significantly increased the c-index of the baseline Cox model in both validation set and external test set, but the inclusion of SGFR can only improve the Cox model less in external test and is unable to improve the Cox model in the validation set.

A hybrid optimization strategy for deliverable intensity-modulated radiotherapy plan generation using deep learning-based dose prediction.

PURPOSE: To propose a clinically feasible automatic planning solution for external beam intensity-modulated radiotherapy, including dose prediction via a deep learning and voxel-based optimization strategy. MATERIALS AND METHODS: The dose distribution of patients was predicted using a U-Net-based deep learning network based on the patient's anatomy information. One hundred seventeen patients with nasopharyngeal cancer (NPC) and 200 patients with rectal cancer were enrolled in this study. For NPC cases, 94 cases were included in the training dataset, 13 in the validation dataset, and 10 in the testing dataset. For rectal cancer cases, 172 cases were included in the training set, 18 in the validation set, and 10 in the testing set. A voxel-based optimization strategy, "Voxel," was proposed to achieve treatment planning optimization by dividing body voxels into two parts: inside planning target volumes (PTVs) and outside PTVs. Fixed dose-volume objectives were attached to the total objective function to realize individualized planning intended as the "hybrid" optimizing strategy. Automatically generated plans were compared with clinically approved plans to evaluate clinical gains, according to dosimetric indices and dose-volume histograms (DVHs). RESULTS: Similarities were found between the DVH of the predicted dose and clinical plan, although significant differences were found in some organs at risk. Better organ sparing and suboptimal PTV coverage were shown using the voxel strategy; however, the deviations in homogeneity indices (HIs) and conformity indices (CIs) of the PTV between automatically generated plans and manual plans were reduced by the hybrid strategy ([manual plans]/[voxel plans[/[hybrid plans]: HI of PTV70 [1.06/1.12/1.02] and CI of PTV70 [0.79/0.58/0.76]). The optimization time for each patient was within 1 min and included fluence map optimization, leaf sequencing, and control point optimization. All the generated plans (voxel and hybrid strategy) could be delivered on uRT-linac 506c (United Imaging Healthcare, Shanghai, China). CONCLUSION: Deliverable plans can be generated by incorporating a voxel-based optimization strategy into a commercial treatment planning system (TPS). The hybrid optimization method shows the benefit and clinical feasibility in generating clinically acceptable plans.

Evaluation of Daily CT for EPID-Based Transit In Vivo Dosimetry.

PURPOSE: The difference in anatomical structure and positioning between planning and treatment may lead to bias in electronic portal image device (EPID)-based in vivo dosimetry calculations. The purpose of this study was to use daily CT instead of planning CT as a reference for EPID-based in vivo dosimetry calculations and to analyze the necessity of using daily CT for EPID-based in vivo dosimetry calculations in terms of patient quality assurance. MATERIALS AND METHODS: Twenty patients were enrolled in this study. The study design included eight different sites (the cervical, nasopharyngeal, and oral cavities, rectum, prostate, bladder, lung, and esophagus). All treatments were delivered with a CT-linac 506c (UIH, Shanghai) using 6 MV photon beams. This machine is equipped with diagnosis-level fan-beam CT and an amorphous silicon EPID XRD1642 (Varex Imaging Corporation, UT, USA). A Monte Carlo algorithm was developed to calculate the transmit EPID image. A pretreatment measurement was performed to assess system accuracy by delivering based on a homogeneous phantom (RW3 slab, PTW, Freiburg). During treatment, each patient underwent CT scanning before delivery either once or twice for a total of 268 fractions obtained daily CT images. Patients may have had a position correction that followed our image-guided radiation therapy (IGRT) procedure. Meanwhile, transmit EPID images were acquired for each field during delivery. After treatment, all patient CTs were reviewed to ensure that there was no large anatomical change between planning and treatment. The reference of transmit EPID images was calculated based on both planning and daily CTs, and the IGRT correction was corrected for the EPID calculation. The gamma passing rate (3 mm 3%, 2 mm 3%, and 2 mm 2%) was calculated and compared between the planning CT and daily CT. Mechanical errors [ ± 1 mm, ± 2 mm, ± 5 mm multileaf collimator (MLC) systematic shift and 3%, 5% monitor unit (MU) scaling] were also introduced in this study for comparing detectability between both types of CT. RESULT: The average (standard deviation) gamma passing rate (3 mm 3%, 2 mm 3%, and 2 mm 2%) in the RW3 slab phantom was 99.6% ± 1.0%, 98.9% ± 2.1%, and 97.2% ± 3.9%. For patient measurement, the average (standard deviation) gamma passing rates were 87.8% ± 14.0%, 82.2% ± 16.9%, and 74.2% ± 18.9% for using planning CTs as reference and 93.6% ± 8.2%, 89.7% ± 11.0%, and 82.8% ± 14.7% for using daily CTs as reference. There were significant differences between the planning CT and daily CT results. All p-values (Mann-Whitney test) were less than 0.001. In terms of error simulation, nonparametric test shows that there were significant differences between practical daily results and error simulation results (p < 0.001). The receiver operating characteristic (ROC) analysis indicated that the detectability of mechanical delivery error using daily CT was better than that of planning CT. AUCDaily CT = 0.63-0.96 and AUCPlanning CT = 0.49-0.93 in MLC systematic shift and AUCDaily CT = 0.56-0.82 and AUCPlanning CT = 0.45-0.73 in MU scaling. CONCLUSION: This study shows the feasibility and effectiveness of using two-dimensional (2D) EPID portal image and daily CT-based in vivo dosimetry for intensity-modulated radiation therapy (IMRT) verification during treatment. The daily CT-based in vivo dosimetry has better sensitivity and specificity to identify the variation of IMRT in MLC-related and dose-related errors than planning CT-based.

The impact of training sample size on deep learning-based organ auto-segmentation for head-and-neck patients.

To investigate the impact of training sample size on the performance of deep learning-based organ auto-segmentation for head-and-neck cancer patients, a total of 1160 patients with head-and-neck cancer who received radiotherapy were enrolled in this study. Patient planning CT images and regions of interest (ROIs) delineation, including the brainstem, spinal cord, eyes, lenses, optic nerves, temporal lobes, parotids, larynx and body, were collected. An evaluation dataset with 200 patients were randomly selected and combined with Dice similarity index to evaluate the model performances. Eleven training datasets with different sample sizes were randomly selected from the remaining 960 patients to form auto-segmentation models. All models used the same data augmentation methods, network structures and training hyperparameters. A performance estimation model of the training sample size based on the inverse power law function was established. Different performance change patterns were found for different organs. Six organs had the best performance with 800 training samples and others achieved their best performance with 600 training samples or 400 samples. The benefit of increasing the size of the training dataset gradually decreased. Compared to the best performance, optic nerves and lenses reached 95% of their best effect at 200, and the other organs reached 95% of their best effect at 40. For the fitting effect of the inverse power law function, the fitted root mean square errors of all ROIs were less than 0.03 (left eye: 0.024, others: <0.01), and theRsquare of all ROIs except for the body was greater than 0.5. The sample size has a significant impact on the performance of deep learning-based auto-segmentation. The relationship between sample size and performance depends on the inherent characteristics of the organ. In some cases, relatively small samples can achieve satisfactory performance.

Deep Learning for Differentiating Benign From Malignant Parotid Lesions on MR Images.

PURPOSE/OBJECTIVESS: Salivary gland tumors are a rare, histologically heterogeneous group of tumors. The distinction between malignant and benign tumors of the parotid gland is clinically important. This study aims to develop and evaluate a deep-learning network for diagnosing parotid gland tumors via the deep learning of MR images. MATERIALS/METHODS: Two hundred thirty-three patients with parotid gland tumors were enrolled in this study. Histology results were available for all tumors. All patients underwent MRI scans, including T1-weighted, CE-T1-weighted and T2-weighted imaging series. The parotid glands and tumors were segmented on all three MR image series by a radiologist with 10 years of clinical experience. A total of 3791 parotid gland region images were cropped from the MR images. A label (pleomorphic adenoma and Warthin tumor, malignant tumor or free of tumor), which was based on histology results, was assigned to each image. To train the deep-learning model, these data were randomly divided into a training dataset (90%, comprising 3035 MR images from 212 patients: 714 pleomorphic adenoma images, 558 Warthin tumor images, 861 malignant tumor images, and 902 images free of tumor) and a validation dataset (10%, comprising 275 images from 21 patients: 57 pleomorphic adenoma images, 36 Warthin tumor images, 93 malignant tumor images, and 89 images free of tumor). A modified ResNet model was developed to classify these images. The input images were resized to 224x224 pixels, including four channels (T1-weighted tumor images only, T2-weighted tumor images only, CE-T1-weighted tumor images only and parotid gland images). Random image flipping and contrast adjustment were used for data enhancement. The model was trained for 1200 epochs with a learning rate of 1e-6, and the Adam optimizer was implemented. It took approximately 2 hours to complete the whole training procedure. The whole program was developed with PyTorch (version 1.2). RESULTS: The model accuracy with the training dataset was 92.94% (95% CI [0.91, 0.93]). The micro-AUC was 0.98. The experimental results showed that the accuracy of the final algorithm in the diagnosis and staging of parotid cancer was 82.18% (95% CI [0.77, 0.86]). The micro-AUC was 0.93. CONCLUSION: The proposed model may be used to assist clinicians in the diagnosis of parotid tumors. However, future larger-scale multicenter studies are required for full validation.