Simulation CT Images Research Articles

Strategy to implement a convolutional neural network based ideal model observer via transfer learning for multi-slice simulated breast CT images

Objective. In this work, we propose a convolutional neural network (CNN)-based multi-slice ideal model observer using transfer learning (TL-CNN) to reduce the required number of training samples. Approach. To train model observers, we generate simulated breast CT image volumes that are reconstructed using the FeldkampDavisKress algorithm with a ramp and Hanning-weighted ramp filter. The observer performance is evaluated on the background-known-statistically (BKS)/signal-known-exactly task with a spherical signal, and the BKS/signal-known-statistically task with random signal generated by the stochastic grown method. We compare the detectability of the CNN-based model observer with that of conventional linear model observers for multi-slice images (i.e. a multi-slice channelized Hotelling observer (CHO) and volumetric CHO). We also analyze the detectability of the TL-CNN for different numbers of training samples to examine its performance robustness to a limited number of training samples. To further analyze the effectiveness of transfer learning, we calculate the correlation coefficients of filter weights in the CNN-based multi-slice model observer. Main results. When using transfer learning for the CNN-based multi-slice ideal model observer, the TL-CNN provides the same performance with a 91.7% reduction in the number of training samples compared to that when transfer learning is not used. Moreover, compared to the conventional linear model observer, the proposed CNN-based multi-slice model observers achieve 45% higher detectability in the signal-known-statistically detection tasks and 13% higher detectability in the SKE detection tasks. In correlation coefficient analysis, it is observed that the filters in most of the layers are highly correlated, demonstrating the effectiveness of the transfer learning for multi-slice model observer training. Significance. Deep learning-based model observers require large numbers of training samples, and the required number of training samples increases as the dimensions of the image (i.e. the number of slices) increase. With applying transfer learning, the required number of training samples is significantly reduced without performance drop.

The dosimetric underpinning of bladder filling criteria for prostate image-guided volumetric modulated arc therapy.

Repeated CT simulation imaging is common at our institution due to failure to achieve acceptable bladder filling in patients undergoing prostate radiotherapy. There is operational value in re-assessing the validity of the bladder filling assessment criteria by comparing the quality of two plans optimized based on either an "Accepted" or "Rejected" bladder status. Twenty prostate patients with repeated CT simulation imaging were included. For each patient, a VMAT plan created using the "Rejected" bladder was compared with the "Accepted" bladder plan. Then, delivered dose to bladder was estimated using ≥4 CBCTs to measure number of fractions with major bladderdose violation (>5% difference) for both plans. Bladder heights of fractions without major bladder dose violations were compared to those with a violation to determine a threshold height for bladder filling acceptability. Using the "Accepted" bladder plans for treatment resulted in 30/175 fractions with major dose violations. These 30 bladders were significantly shorter than those without major violation (mean 28 vs 43mm, p < 0.05). The "Rejected" bladder plans delivered a lower dose to the bladder by ≥5% than the '"Accepted" bladder plans in 59% of fractions, and the number of fractions with major dose violations was 17. Using a shorter bladder for plan optimization resulted in better bladder sparing during treatment and improved compliance to protocol specific bladder dose constraints. A bladder height range of 20-40 mm measured between the bladder dome and the superior aspect of the symphysis pubis is recommended for prostate radiotherapy requiring a full bladder protocol. Using real patient data from simulation and treatment, this study established a range of bladder height that can be measured easily in a clinical setting for assessing adequacy of bladder filling for prostate radiotherapy.

Comparison of dosimetric analysis of organs at risk and target volumes for ovoids and cylinders in endometrial carcinoma.

To compare dose volume parameters of target and organs at risk in vaginal vault brachytherapy using ovoids or cylinder in post-operative endometrial carcinoma. The study was done among 25 histologically proven post-operative endometrial carcinoma patients requiring vaginal brachytherapy. All patients underwent both cylinder and ovoids application alternatively on weekly basis. Ovoids size ranged from 2 to 3 cm diameter. Diameters of cylinder ranged between 2.5 and 3.5 cm. Bladder, rectum, urethra, and clinical target volume (CTV) were contoured on CT simulation images. Prescribed dose was 6-7 Gy in 2-3 fractions at 0.5 cm from the surface of applicator. The mean values of D90, D50, V150, V100, V90, and V50 of CTV were comparable between cylinder and ovoids plans. The mean dose of CTV was significantly higher with cylinder than with ovoids, and D100 was significantly higher with ovoids (mean = 15.63 Gy vs. 14.64 Gy, p = 0.016, and D100 = 37.82% vs. 42.86%, p = 0.042, for cylinder vs. ovoids). In the dosimetry of the vault, D90, D50, V100, V90, V50, and mean of the vault did not show any significant difference between cylinder and ovoids. The V150 was significantly higher with cylinder plans than ovoids, and D100 of the vault was significantly higher with ovoids plans (V150 = 14.81% vs. 6.86%, p = 0.02, and D100 = 37.77% vs. 44.80%, p = 0.029, for cylinder vs. ovoids). D0.1cc, D1cc, D2cc, and mean for the bladder, rectum, and urethra were comparable between the cylinder and ovoid plans. The present study showed that the dose to organs at risk, most of the dosimetric parameters of CTV, and vault were comparable between the cylinder and ovoid plans. Both applicators provide good reproducibility. The choice of applicator will ultimately depend on the institutional policies and oncologist decision. However, in patients with dog-ear configuration of the vagina, ovoids may be preferred as per ABS guidelines.

Adaptive radiation therapy (art) for patients with limited-stage small cell lung cancer (LS-SCLC): A dosimetric evaluation.

Adaptive radiation therapy (ART) refers to redesigning of radiation therapy (RT) treatment plans with respect to dynamic changes in tumor size and location throughout the treatment course. In this study, we performed a comparative volumetric and dosimetric analysis to investigate the impact of ART for patients with limited-stage small cell lung cancer (LS-SCLC). Twenty-four patients with LS-SCLC receiving ART and concomitant chemotherapy were included in the study. ART was performed by replanning of patients based on a mid-treatment computed tomography (CT)-simulation which was routinely scheduled for all patients 20-25 days after the initial CT-simulation. While the first 15 RT fractions were planned using the initial CT-simulation images, the latter 15 RT fractions were planned using the mid-treatment CT-simulation images acquired 20-25 days after the initial CT-simulation. In order to document the impact of ART, target and critical organ dose-volume parameters acquired from this adaptive radiation treatment planning (RTP) were compared with the RTP based solely on the initial CT-simulation to deliver the whole RT dose of 60 Gy. Statistically significant reduction was detected in gross tumor volume (GTV) and planning target volume (PTV) during the conventionally fractionated RT course along with statistically significant reduction in critical organ doses with incorporation of ART. One-third of the patients in our study who were otherwise ineligible for curative intent RT due to violation of critical organ dose constraints could be treated with full dose irradiation by use of ART. Our results suggest significant benefit of ART for patients with LS-SCLC.

Thyroid volume changes following adjuvant radiation therapy for breast cancer

Background and purposeIncidental thyroid gland irradiation frequently occurs in breast cancer patients who receive regional nodal irradiation (RNI) to the supraclavicular (SCV) region. Recent studies suggest hypothyroidism (HT) is a complication of radiation therapy (RT) that includes SCV fields. We retrospectively analyzed patients who received RNI to evaluate thyroid gland evolution following RT as well as its association with the development of HT. Materials and methods61 breast cancer patients received SCV-directed RT between 2007 and 2019 and met inclusion criteria. Thyroid glands were retrospectively contoured on CT simulation and follow-up images. Individual dose-volume histograms were analyzed to determine thyroid volume within and outside specific isodose lines. Relative thyroid volume changes based on different radiation doses were estimated by fusing post-RT scans with CT simulation. Logistic regression was performed to assess thyroid volume changes as a factor in the development of HT. ResultsMedian pre-treatment thyroid volume was 11.8 cc (range: 6.3–74.1 cc) with a median of 42.2 % within the 20 Gy and 23.2 % within the 40 Gy isodose lines. A significant decrease in thyroid volume was noted by 1-year post-treatment (p < 0.0001) and thereafter. By 4 years post-treatment, average thyroid volume was decreased by 29.7 % (range: 2.3–64.4 %). Thyroid volume receiving 40 Gy or higher demonstrated a greater decrease compared to those receiving lower irradiation dosage. HT occurred in 17 patients (27.9 %). Patients who developed HT displayed a larger decrease in the thyroid volume receiving between 20 and 40 Gy at 12 months (p = 0.033). ConclusionOur study demonstrates for the first time that a reduction in thyroid volume may be seen as early as 6 months after SCV-directed RT for breast cancer, which correlates with development of clinical and subclinical HT. Furthermore, a dose-dependent correlation exists between thyroid subvolume reduction and SCV-directed RT in breast cancer patients. As feasible, efforts should be made to reduce the dose to the thyroid in patients who undergo RNI for breast cancer.

Dosimetric evaluation of bone marrow sparing in proton radiotherapy for cervical cancer guided by MR functional imaging

BackgroundTo segment the pelvic active bone marrow (PABM) using magnetic resonance (MR) functional imaging and investigate the feasibility and dosimetric characteristics of cervical cancer proton radiotherapy for active bone marrow (ABM) sparing.MethodsWe collected CT and MR simulation images of 33 patients with cervical cancer retrospectively. The PBM was contoured on the MRI FatFrac images; the PBM was divided into high-active bone marrow (ABM-high) and low-active bone marrow based on the fat content of the PBM. Four radiotherapy plans were created for each patient, which included intensity-modulated photon therapy (IMRT), bone marrow sparing IMRT (IMRT-BMS), intensity-modulated proton therapy (IMPT), and bone marrow sparing IMPT (IMPT-BMS). The dosimetric differences among the four plans were compared.ResultsThe ABM-high volume in the enrolled patients accounted for 45.2% of the total ABM volume. The target coverage was similar among the four radiotherapy plans. IMRT-BMS, IMPT, and IMPT-BMS reduced the Dmean of ABM-high by 16.6%, 14.2%, and 44.5%, respectively, compared to the Dmean of IMRT (p < 0.05). IMPT-BMS had the best protective effect on the bone marrow. Compared to IMRT, the volume of ABM-high receiving an irradiation dose of 5–40 Gy decreased by 10.2%, 36.8%, 58.8%, 67.4%, 64.9%, and 44.5%, respectively (p < 0.001).ConclusionsThe MR functional imaging technique helped in the grading and segmentation of PABM. MR functional image-guided proton radiotherapy for cervical cancer can achieve optimal BMS.

A novel approach for dose painting radiotherapy of brain metastases guided by mr perfusion images.

To investigate the feasibility and dosimetric index features of dose painting guided by perfusion heterogeneity for brain metastasis (BMs) patients. A total of 50 patients with single BMs were selected for this study. CT and MR simulation images were obtained, including contrast-enhanced T1-weighted images (T1WI+C) and cerebral blood flow (CBF) maps from 3D-arterial spin labeling (ASL). The gross tumor volume (GTV) was determined by fusion of CT and T1WI+C images. Hypoperfused subvolumes (GTVH) with less than 25% of the maximum CBF value were defined as the dose escalation region. The planning target volume (PTV) and PTVH were calculated from GTV and GTVH respectively. The PTVN was obtained by subtracting PTVH from PTV, and conventional dose was given. Three kinds of radiotherapy plans were designed based on the CBF values. Plan 1 was defined as the conventional plan with an arbitrary prescription dose of 60 Gy for PTV. For dose painting, Plan 2 and Plan 3 escalated the prescription dose for PTVH to 72 Gy based on Plan 1, but Plan 3 removed the maximum dose constraint. Dosimetric indices were compared among the three plans. The mean GTV volume was 34.5 (8.4-118.0) cm3, and mean GTVH volume was 17.0 (4.5-58.3) cm3, accounting for 49.3% of GTV. Both conventional plan and dose painting plans achieved 98% target coverage. The conformity index of PTVH were 0.44 (Plan1), 0.64 and 0.72 (Plan 2 and Plan 3, P<0.05). Compared to Plan 1, the D2%, D98% and Dmean values of the PTVH escalated by 20.50%, 19.32%, and 19.60% in Plan 2 and by 24.88%, 17.22% and 19.22% in Plan 3 respectively (P<0.05). In the three plans, the index of achievement value for PTVH was between 1.01 and 1.03 (P<0.05). The dose increment rates of Plan 2 and Plan 3 for each organs at risk (OARs) was controlled at 2.19% - 5.61% compared with Plan 1. The doses received by OARs did not significantly differ among the three plans (P >0.05). BMs are associated with significant heterogeneity, and effective escalation of the dose delivered to target subvolumes can be achieved with dose painting guided by 3D-ASL without extra doses to OARs.

Evaluating the relationship between vaginal apex “dog ears” and patterns of recurrence in endometrial cancer following adjuvant image guided vaginal cuff brachytherapy

PURPOSEThe aim of this investigation is to characterize vaginal apex “dog ears” and their association with patterns of treatment failure in patients with endometrial cancer treated with adjuvant high-dose-rate (HDR) single-channel vaginal cuff brachytherapy (VCB). METHODSA retrospective review of patients treated with HDR VCB from 2012 to 2021 for medically operable endometrial cancer at a single institution was conducted. Dog ears, defined as tissue at the apex extending at least 10 mm from the brachytherapy applicator were identified on CT simulation images. Fisher exact test and a multivariate logistic regression model evaluated the association between factors of interest with treatment failure. Vaginal cuff failure free survival (VCFFS) was calculated from first brachytherapy to vaginal cuff recurrence (VCR). RESULTSA total of 219 patients were reviewed. In this sample, 57.5% of patients met criteria for having dog ears. In total, 13 patients (5.9%) developed a VCR. There was no statistically significant difference in the rate of VCR between patients with and without dog ears (7.1% vs. 4.3%, p = 0.56). There was a trend toward increased risk of recurrence with higher grade histology identified in the multivariate logistic regression model (p = 0.085). The estimated 3-year probability of VCFFS was 86%. CONCLUSIONSVaginal apex dog ears are prevalent but are not found to statistically increase the risk of VCR after VCB in our single institution experience. However, while local failure remains low in this population, we report an absolute value of over twice as many VCRs in patients with dog ears, indicating that with improved dog ear characterization this may remain a relevant parameter for consideration in treatment planning.

Contouring lumbosacral plexus nerves with MR neurography and MR/CT deformable registration technique.

It is difficult to contour nerve structures with the naked eye due to poor differentiation between the nerve structures with other soft tissues on CT images. Magnetic resonance neurography (MRN) has the advantage in nerve visualization. The purpose of this study is to identify one MRN sequence to better assist the delineation of the lumbosacral plexus (LSP) nerves to assess the radiation dose to the LSP using the magnetic resonance (MR)/CT deformable coregistration technique. A total of 18 cases of patients with prostate cancer and one volunteer with radiation-induced lumbosacral plexopathy (RILSP) were enrolled. The data of simulation CT images and original treatment plans were collected. Two MRN sequences (Lr_NerveVIEW sequence and Cs_NerveVIEW sequence) were optimized from a published MRN sequence (3D NerveVIEW sequence). The nerve visualization ability of the Lr_NerveVIEW sequence and the Cs_NerveVIEW sequence was evaluated via a four-point nerve visualization score (NVS) scale in the first 10 patients enrolled to determine the better MRN sequence for assisting nerve contouring. Deformable registration was applied to the selected MRN sequence and simulation CT images to get fused MR/CT images, on which the LSP was delineated. The contouring of the LSP did not alter treatment planning. The dosimetric data of the LSP nerve were collected from the dose-volume histogram in the original treatment plans. The data of the maximal dose (Dmax) and the location of the maximal radiation point received by the LSP structures were collected. The Cs_NerveVIEW sequence gained lower NVS scores than the Lr_NerveVIEW sequence (Z=-2.887, p=0.004). The LSP structures were successfully created in 18 patients and one volunteer with MRN (Lr_NerveVIEW)/CT deformable registration techniques, and the LSP structures conformed with the anatomic distribution. In the patient cohort, the percentage of the LSP receiving doses exceeding 50, 55, and 60 Gy was 68% (12/18), 33% (6/18), and 17% (3/18), respectively. For the volunteer with RILSP, the maximum irradiation dose to his LSP nerves was 69 Gy. The Lr_NerveVIEW MRN sequence performed better than the Cs_NerveVIEW sequence in nerve visualization. The dose in the LSP needs to be measured to understand the potential impact on treatment-induced neuropathy.

Evaluating the accumulated dose distribution of organs at risk in combined radiotherapy for cervical carcinoma based on deformable image registration

ObjectiveTo evaluate the feasibility and value of deformable image registration (DIR) in calculating the cumulative doses of organs at risk (OARs) in the combined radiotherapy of cervical cancer. Patients and methodsThirty cervical cancer patients treated with external beam radiotherapy (EBRT) combined with intracavitary brachytherapy (ICBT) were reviewed. The simulation CT images of EBRT and ICBT were imported into Varian Velocity 4.1 for the DIR-based dose accumulation. Cumulative dose-volume parameters of D2cc for rectum and bladder were compared between the direct addition (DA) and DIR methods. The quantitative parameters were measured to evaluate the accuracy of DIR. ResultsThe three-dimensional cumulative dose distribution of the tumor and OARs were graphically well illustrated by composite isodose lines. In combined EBRT and ICBT, the mean cumulative bladder D2cc calculated by DIR and DA was 86.13 Gy and 86.27 Gy, respectively. The mean cumulative rectal D2cc calculated by DIR and DA was 72.97 Gy and 73.90 Gy, respectively. No significant differences were noted between these two methods (p > 0.05). As to the parameters used to evaluate the DIR accuracy, the mean DSC, Jacobian, MDA (mm) and Hausdorff distance (mm) were 0.79, 1.0, 3.84, and 22.01 respectively for the bladder and 0.53, 1.2, 7.31, and 29.58 respectively for the rectum. In this study, the DSC seemed to be slightly lower compared with previous studies. ConclusionDose accumulation based on DIR might be an alternative method to illustrate and evaluate the cumulative doses of the OARs in combined radiotherapy for cervical cancer. However, DIR should be used with caution before overcoming the relevant limitations.

GPP02 Presentation Time: 8:10 AM: Evaluating the Relationship Between Vaginal Apex ``Dog Ears'' and Patterns of Recurrence in Endometrial Cancer Following Adjuvant Image Guided Vaginal Cuff Brachytherapy

<h3>Purpose</h3> The vaginal cuff is a common site of tumor recurrence following surgery for endometrial cancer, and adjuvant vaginal cuff brachytherapy (VCB) is indicated to reduce the risk of local recurrence. Following hysterectomy, many patients have redundant tissue at the lateral apex of the vaginal cuff ("dog ears"). This extra tissue may extend irregularly beyond the typical cylindrical dose distribution of radiation delivered during single channel VCB and it is hypothesized that they are a potential site of local treatment failure even after VCB. The effects of the presence and size of vaginal cuff "dog ears" on tumor recurrence is not yet fully understood. The aim of this study is to evaluate the prevalence and size of extra tissue at the vaginal apex, and whether these factors predict disease recurrence in patients treated with adjuvant VCB for endometrial cancer. <h3>Materials and Methods</h3> We retrospectively reviewed the medical records of 219 patients with early-stage endometrial cancer treated with image-guided single channel VCB at our institution from 2012 - 2021 with available brachytherapy planning CT simulation images. "Dog ears" were defined as the presence of additional soft tissue extending at least 10 mm from the apex of the vaginal cuff perpendicular to the surface of the brachytherapy applicator as defined on a coronal or axial CT simulation. Patient characteristics and outcomes including stage, size and laterality of dog ears, presence of air gaps, and patterns of failure were collected. Vaginal cuff recurrence (VCR) failure free survival was calculated from time of first brachytherapy to VCR and was estimated by the product limit method of Kaplan and Meier. Fisher's exact test was used to compare categorical variables and a multivariate logistic regression model was developed to evaluate the association of overall treatment failure with various factors of interest. <h3>Results</h3> We found that 57.5% of patients (n = 126) met our criteria for having "dog ears," which, on average, extended 23.8 mm (SD = 6.4) from the brachytherapy applicator. In total, 13 patients (5.9%) developed a VCR, 12 (5.5%) developed a regional recurrence, and 16 (7.3%) developed distant recurrence. There was no statistically significant difference in VCRs between patients with dog ears and those without dog ears (7.1% vs 4.3%, p=0.56). The various surgical approaches (abdominal, laparoscopic, transvaginal) did not have an increased propensity to result in dog ears (p=0.74). The 3-year probability of VCR failure-free survival is 86%. On multivariate analysis (MVA), no factors predicted for overall treatment failure including the presence of dog ears, maximum dimension of dog ears, FIGO stage, presence of air gaps at the applicator surface on post insertion CT images, or brachytherapy duration. There was a trend toward increased risk of recurrence with higher grade histology on MVA (p=0.0805). <h3>Conclusions</h3> In this investigation of a single institution experience with single-channel image-guided VCB, the presence of "dog ears" at the vaginal apex is prevalent; however, VCR remains low with distant metastases being the predominant site of failure. While dog ears appear to have little impact on overall recurrence patterns, these may be a factor to consider when planning adjuvant brachytherapy including choosing alternate applicators.

Efficacy of Medical Image Transformations within an Open Sourced Head and Neck Deep Learning Auto-Contouring Framework

<h3>Purpose/Objective(s)</h3> Radiation therapy treatment plans have recently utilized deep learning methods to automatically contour organs at risk (OAR). In this study, we use an open-source framework, Medical Open Network for AI (MONAI), and the 3D U-Net network to automatically segment head and neck OAR structures. This study shows that using pre-processing medical image transformations improves performance of auto-contouring. <h3>Materials/Methods</h3> This study used the MONAI deep learning framework to create 28 distinct segmentation models for head and neck OAR structures from a dataset of manually contoured structure sets from CT simulation images. The MONAI framework integrated pre-processing transforms included image windowing, pixel resampling, foreground cropping, and random crop patching that was performed on the training sets prior to model training. Model training was performed using a 3D U-Net algorithm. Mean dice similarity coefficients were computed for each OAR deep learning model. <h3>Results</h3> The training was conducted on a single institution's dataset composed of 609 unique OAR structure sets. Select mean dice similarity coefficients achieved in the auto-contouring model's validation sets ranged as high as 0.906, 0.804, and 0.775 for the orbit, submandibular gland, and thyroid respectively. <h3>Conclusion</h3> We trained a series of deep convolutional neural network models that perform automated contouring of head and neck OAR with a high degree of accuracy, outperforming other models to date. The models were trained using the open-source MONAI framework that is easily accessible and customizable to anyone in radiation oncology. The utility of medically specific image transformations available within MONAI allow for improved model performance and generalizability, and similar techniques could be employed in future studies.

Pre-Radiation CT-Based Radiomic Features Predict Locoregional and Distant Failure in Locally Advanced Head and Neck Cancer

<h3>Purpose/Objective(s)</h3> Locally advanced head and neck cancer treatment often involves aggressive multimodal therapy. While outcomes have improved in certain groups (HPV-positive disease), locoregional (LRF) and distant failure (DF) remain persistent problems. Accurate prediction of these risks would provide vital clinical insight; for instance, a patient with low LRF risk could be de-escalated, while a patient at high-risk for DF could be offered systemic options in lieu of aggressive local therapy. In this study we explore the use of CT-based radiomic features (quantitative sub-visual cues) for this purpose. <h3>Materials/Methods</h3> We identified patients with primary or recurrent locally advanced head/neck cancer treated between 2014-2020 who received a definitive or adjuvant RT course (naïve to RT prior). Histologies included squamous cell carcinoma (SCC), poorly differentiated carcinoma (nasopharyngeal or salivary origin), high-risk parotid malignancies, and skin SCC metastatic to the parotid/neck. Patients had at least 12 months of follow-up and charts were assessed for LRF and DF. CT simulation images were extracted from our treatment planning system, and specific contoured regions of interest (ROIs) were selected including the peritumoral region or post-surgical tumor bed (high-risk clinical tumor volume, CTV HR) and involved/elective neck levels (CTV Necks). CT scans were resampled to be isotropic and 3D radiomic features (Gabor wavelet, Hara lick, and CoLlAGe) were extracted from within CTV HR and CTV Necks ROIs. Using a Wilcoxon rank sum test, differential expression for each feature was calculated between patients experiencing treatment failure (LRF and DF) and those with no evidence of disease progression. A similar process was done with baseline clinical features including performance status, tumor stage, age, p16 status and smoking history. Where required, a chi-squared test was used for categorical clinical variables. <h3>Results</h3> 95 patients were analyzed (45 definitive, 50 adjuvant). There were 17 LRF (17.9%) and 16 DF (16.8%) events. 304 radiomic features were extracted from CTV HR and CTV Necks (the latter was only performed for patients with bilateral neck contours). 2 features extracted from CTV HR ROIs were significantly associated with LRF and 29 features with DF (p < 0.05). 30 features extracted from CTV Necks ROIs were significantly associated with LRF and 0 features with DF (p < 0.05). There were no statistically significant clinical features associated with LRF or DF. <h3>Conclusion</h3> CT-based radiomic features derived from specific ROIs in simulation images were associated with both LRF and DF events, which was not the case for canonical clinical features. Interestingly, the CTV HR was more prominently associated with DF, whereas the CTV Necks was associated with LRF. Further studies with larger, more diverse datasets are needed to refine this model and provide a path for clinical validation.

Geometric and Dosimetric Evaluation of Using a Novel Tongue Positioning Device to Reduce Tongue Motions during Radiation Therapy for Head and Neck Cancer

<h3>Purpose/Objective(s)</h3> Inter- and intra-fractional organ motion in the head and neck region can reduce the precision of radiation therapy (RT). Among the organs, a tongue has substantial changes in its position and volume. Therefore, various types of tongue positioning devices have been used during RT for head and neck cancer. Despite of several clinical studies reporting an efficacy of using tongue positioning device, there are no studies investigating the inter-fractional reproducibility of position and shape of tongue while using a tongue positioning device. This study performs geometric and dosimetric evaluation of a novel tongue positioning device during RT for head and neck cancer. <h3>Materials/Methods</h3> Fifty-two patients with head and neck tumors using tongue positioning device were available for analysis. All patients were treated in 28 fractions using VMAT technique with cone-beam CT (CBCT) imaging. CBCT images were taken at 1^st, 7^th, 11^th, 15^th, 19^th, 23^th, and 27^th fractions. In order to evaluate the difference in geometry of the tongue between the CT simulation and the treatment at each fraction, tongues in simulation CT images and CBCT images were contoured in treatment planning system by a trained radiation oncologist. The geometric differences of the tongue between planning CT (pCT) and CBCT were assessed by Dice similarity coefficient (DSC) and averaged Hausdorff distance (AVD). In addition, 2D <i>in vivo</i> measurements using radiochromic films were performed for thirteen patients once a week during RT for 5 weeks. The films were attached to heads of the devices. The planned dose distributions at the surface of the head of device was compared to the measured dose distributions by using gamma analysis with criteria of 3%/3mm. <h3>Results</h3> For entire patients, DSC scored 0.77 to 0.80. The DSC at 1^st fraction (i.e., pCT vs. 1^st CBCT) was 0.80 while the DSC at 27^th fraction (i.e., pCT vs. 27^th CBCT) was 0.77. There was no statistical significance between the DSC at 1^st fraction and the DSC at 27^th fraction (p-value > 0.05). On the other hand, AVD ranged from 1.78 mm to 2.25 mm. The AVD was increased as the fractions were proceeded. There was difference between the AVD at 1^st fraction and the AVD at 23^th fraction and 27^th fraction with a statistical significance (p-value < 0.05). <i>In vivo</i> measurements showed averaged gamma passing rate of 90.54%. There was no significant difference between the measurements at 1^st week and those at other weeks. <h3>Conclusion</h3> Comparison of geometry of the tongue during RT for head and neck cancer was performed with pCT and CBCT. The AVD showed maximum 2.25 mm difference between pCT and CBCT. The AVD difference between the fractions was within 0.47 mm. DSC showed no significant inter-fractional differences. In conclusion, the novel tongue positioning device was found to minimize the inter-fractional variations of position and shape of the tongue.

ANAPLASTİK OLİGODENDROGLİOMADA GÖRÜNTÜLEME YÖNTEMLERİNİN RADYOTERAPİ TEDAVİ VOLÜMÜNÜ BELİRLEMEDEKİ ETKİSİ

Objective: Utility of radiation therapy (RT) has been well established for management of World Health Organization (WHO) grade 3 anaplastic oligodendroglioma. Nevertheless, optimal target definition for radiotherapeutic management of WHO grade 3 anaplastic oligodendroglioma requires elucidation. Within this context, treatment volume determination for WHO grade anaplastic oligodendroglioma based on multimodality imaging is assessed in this original research article. &#x0D; Materials and methods: Treatment volume determination for RT by incorporation of MRI or by CT-simulation images only was comparatively assessed for patients receiving RT for WHO grade 3 anaplastic oligodendroglioma in this original research article.&#x0D; Results: VersaHD (Elekta, UK) Linear Accelerator (LINAC) was used for treatment with 6 MV photons. IGRT techniques were utilized for treatment verification. Ground truth target volume was used as the reference for actual treatment and for comparison purposes, and it was determined after meticulous evaluation on an individual basis. Primary outcome measure of the study was comparative assessment of treatment volume determination by either CT-only imaging and by CT-MR fusion based imaging. This study revealed that the ground truth target volume was identical with treatment volume definition by CT-MR fusion based imaging. &#x0D; Conclusion: This study reveals improved treatment volume determination for WHO grade 3 anaplastic oligodendrogliomas by incorporation of MRI in the treatment planning procedure. Clearly, further studies are warranted to shed light on this issue.

Synthetic CT generation for MRI-guided adaptive radiotherapy in prostate cancer.

Current MRI-guided adaptive radiotherapy (MRgART) workflows require fraction-specific electron and/or mass density maps, which are created by deformable image registration (DIR) between the simulation CT images and daily MR images. Manual density overrides may also be needed where DIR-produced results are inaccurate. This approach slows the adaptive radiotherapy workflow and introduces additional dosimetric uncertainties, especially in the presence of the magnetic field. This study investigated a method based on a conditional generative adversarial network (cGAN) with a multi-planar method to generate synthetic CT images from low-field MR images to improve efficiency in MRgART workflows for prostate cancer. Fifty-seven male patients, who received MRI-guided radiation therapy to the pelvis using the ViewRay MRIdian Linac, were selected. Forty-five cases were randomly assigned to the training cohort with the remaining twelve cases assigned to the validation/testing cohort. All patient datasets had a semi-paired DIR-deformed CT-sim image and 0.35T MR image acquired using a true fast imaging with steady-state precession (TrueFISP) sequence. Synthetic CT images were compared with deformed CT images to evaluate image quality and dosimetric accuracy. To evaluate the dosimetric accuracy of this method, clinical plans were recalculated on synthetic CT images in the MRIdian treatment planning system. Dose volume histograms for planning target volumes (PTVs) and organs-at-risk (OARs) and dose distributions using gamma analyses were evaluated. The mean-absolute-errors (MAEs) in CT numbers were 30.1 ± 4.2 HU, 19.6 ± 2.3 HU and 158.5 ± 26.0 HU for the whole pelvis, soft tissue, and bone, respectively. The peak signal-to-noise ratio was 35.2 ± 1.7 and the structural index similarity measure was 0.9758 ± 0.0035. The dosimetric difference was on average less than 1% for all PTV and OAR metrics. Plans showed good agreement with gamma pass rates of 99% and 99.9% for 1%/1 mm and 2%/2 mm, respectively. Our study demonstrates the potential of using synthetic CT images created with a multi-planar cGAN method from 0.35T MRI TrueFISP images for the MRgART treatment of prostate radiotherapy. Future work will validate the method in a large cohort of patients and investigate the limitations of the method in the adaptive workflow.

Improving reproducibility and performance of radiomics in low-dose CT using cycle GANs.

BackgroundAs a means to extract biomarkers from medical imaging, radiomics has attracted increased attention from researchers. However, reproducibility and performance of radiomics in low‐dose CT scans are still poor, mostly due to noise. Deep learning generative models can be used to denoise these images and in turn improve radiomics’ reproducibility and performance. However, most generative models are trained on paired data, which can be difficult or impossible to collect.PurposeIn this article, we investigate the possibility of denoising low‐dose CTs using cycle generative adversarial networks (GANs) to improve radiomics reproducibility and performance based on unpaired datasets.Methods and materialsTwo cycle GANs were trained: (1) from paired data, by simulating low‐dose CTs (i.e., introducing noise) from high‐dose CTs and (2) from unpaired real low dose CTs. To accelerate convergence, during GAN training, a slice‐paired training strategy was introduced. The trained GANs were applied to three scenarios: (1) improving radiomics reproducibility in simulated low‐dose CT images and (2) same‐day repeat low dose CTs (RIDER dataset), and (3) improving radiomics performance in survival prediction. Cycle GAN results were compared with a conditional GAN (CGAN) and an encoder–decoder network (EDN) trained on simulated paired data.ResultsThe cycle GAN trained on simulated data improved concordance correlation coefficients (CCC) of radiomic features from 0.87 (95%CI, [0.833,0.901]) to 0.93 (95%CI, [0.916,0.949]) on simulated noise CT and from 0.89 (95%CI, [0.881,0.914]) to 0.92 (95%CI, [0.908,0.937]) on the RIDER dataset, as well improving the area under the receiver operating characteristic curve (AUC) of survival prediction from 0.52 (95%CI, [0.511,0.538]) to 0.59 (95%CI, [0.578,0.602]). The cycle GAN trained on real data increased the CCCs of features in RIDER to 0.95 (95%CI, [0.933,0.961]) and the AUC of survival prediction to 0.58 (95%CI, [0.576,0.596]).ConclusionThe results show that cycle GANs trained on both simulated and real data can improve radiomics’ reproducibility and performance in low‐dose CT and achieve similar results compared to CGANs and EDNs.

Deep learning versus iterative reconstruction on image quality and dose reduction in abdominal CT: a live animal study

Objective. While simulated low-dose CT images and phantom studies cannot fully approximate subjective and objective effects of deep learning (DL) denoising on image quality, live animal models may afford this assessment. This study is to investigate the potential of DL in CT dose reduction on image quality compared to iterative reconstruction (IR). Approach. The upper abdomen of a live 4 year old sheep was scanned on a CT scanner at different exposure levels. Images were reconstructed using FBP and ADMIRE with 5 strengths. A modularized DL network with 5 modules was used for image reconstruction via progressive denoising. Radiomic features were extracted from a region over the liver. Concordance correlation coefficient (CCC) was applied to quantify agreement between any two sets of radiomic features. Coefficient of variation was calculated to measure variation in a radiomic feature series. Structural similarity index (SSIM) was used to measure the similarity between any two images. Diagnostic quality, low-contrast detectability, and image texture were qualitatively evaluated by two radiologists. Pearson correlation coefficient was computed across all dose-reconstruction/denoising combinations. Results. A total of 66 image sets, with 405 radiomic features extracted from each, are analyzed. IR and DL can improve diagnostic quality and low-contrast detectability and similarly modulate image texture features. In terms of SSIM, DL has higher potential in preserving image structure. There is strong correlation between SSIM and radiologists’ evaluations for diagnostic quality (0.559) and low-contrast detectability (0.635) but moderate correlation for texture (0.313). There is moderate correlation between CCC of radiomic features and radiologists’ evaluation for diagnostic quality (0.397), low-contrast detectability (0.417), and texture (0.326), implying that improvement of image features may not relate to improvement of diagnostic quality. Conclusion. DL shows potential to further reduce radiation dose while preserving structural similarity, while IR is favored by radiologists and more predictably alters radiomic features.

Photon-counting CT versus conventional CT for COPD quantifications: intra-scanner optimization and inter-scanner assessments using virtual imaging trials.

Chronic obstructive pulmonary disease (COPD) is a chronic inflammatory lung disease and a major cause of death and disability worldwide. Quantitative CT is a powerful tool to better understand the heterogeneity and severity of this disease. Quantitative CT is being increasingly used in COPD research, and the recent advancements in CT technology have made it even more encouraging. One recent advancement has been the development of photon-counting detectors, offering higher spatial resolution, higher image contrast, and lower noise levels in the images. However, the quantification performance of this new technology compared to conventional scanners remains unknown. Additionally, different protocol settings (e.g., different dose levels, slice thicknesses, reconstruction kernels and algorithms) affect quantifications in an unsimilar fashion. This study investigates the potential advantages of photon-counting CT (PCCT) against conventional energy-integrating detector (EID) CT and explores the effects of protocol settings on lung density quantifications in COPD patients. This study was made possible using a virtual imaging platform, taking advantage of anthropomorphic phantoms with COPD (COPD-XCAT) and a scanner-specific CT simulator (DukeSim). Having the physical and geometrical properties of three Siemens commercial scanners (Flash, Force for EID and NAEOTOM Alpha for PCCT) modeled, we simulated CT images of ten COPD-XCAT phantoms at 0.63 and 3.17 mGy dose levels and reconstructed at three levels of kernel sharpness. The simulated CT images were quantified in terms of "Lung mean absolute error (MAE)," "LAA -950," "Perc 15," "Lung mass" imaging biomarkers and compared against the ground truth values of the phantoms. The intra-scanner assessment demonstrated the superior qualitative and quantitative performance of the PCCT scanner over the conventional scanners (21.01% and 22.74% mean lung MAE improvement, and 53.97% and 68.13% mean LAA -950 error improvement compared to Flash and Force). The results also showed that higher mAs, thinner slices, smoother kernels, and iterative reconstruction could lead to more accurate and precise quantification scores. This study underscored the qualitative and quantitative benefits of PCCT against conventional EID scanners as well as the importance of optimal protocol choice within scanners for more accurate quantifications.

Dosimetric Comparison and TCP-NTCP Modeling for Lung, Heart, Left Anterior Descending, and Right Coronary Artery in Left-sided Breast Cancer Conventional and Hypofractionated Radiotherapy

Background: The aim of this study was to evaluate the dose distribution, as well as tumor control probability (TCP) and normal tissue complications probability (NTCP) models for females with left-sided breast cancer for 3D-CRT, 6, and 9 fields intensity modulated radiotherapy (IMRT) and hypofractionated tangential plans. Methods: Eighty females with left-sided breast cancer (N1T1 – N3T3) were included in this study. The patients were divided into breast conserving surgery (BCS) (n= 50) and mastectomy (n=30) patients. CT simulation images of the patients were imported on the treatment planning software (TiGRT, LinaTech, China) and the tangential treatment plans of the mentioned methods were done, using TPS. Dose assessments were performed, employing the TPS, and TCP-NTCP models of stated modalities were done, using Poisson linear-quadatric (PLQ) and Lyman-Kutcher-Burman (LKB) models on MATLAB and R software. Results: For the BCS and post-mastectomy patients, 6FIMRT imposed lower doses to ipsilateral lung, heart, LAD, RCA, and contralateral breast compared to 9FIMRT, hypofractionated RT, and 3D-CRT (P &lt; 0.005). Also, the NTCPs of the mentioned normal tissues for 6FIMRT were lower than other methods for both BCS and post-mastectomy groups. Moreover, it was found that the V20Gy for Ipsilateral lung and the V25Gy for heart, LAD, and RCA of 6FIMRT, 9FIMRT, and hypofractionated RT was significantly lower compared to 3D-CRT (P &lt; 0.005) for both BCS and mastectomy groups, while there were no significant differences among them for the 6FIMRT and 9FIMRT with hypofractionated RT (P &gt; 0.005). The TCP values of 9FIMRT, 6FIMRT, and hypofractionated RT were not considerably different; however, the TCP values of 3D-CRT were lower compared to other stated methods. Conclusions: 6FIMRT is a suitable choice for RT of patients with left-sided breast cancer compared to other mentioned modalities, as a result of providing adequate PTV dose coverage and TCPs. Also, it may impose lower doses and NTCPs for OARs. Hypofractionated RT is a good alternative to reduce treatment time for patients with breast cancer.