A deep learning model for translating CT to ventilation imaging: analysis of accuracy and impact on functional avoidance radiotherapy planning.

PURPOSE: Radiotherapy planning incorporating functional lung images has the potential to reduce pulmonary toxicity. Free-breathing 4DCT-derived ventilation image (CTVI) may help quantify lung function. This study introduces a novel deep-learning model directly translating planning CT images into CTVI. We investigated the accuracy of generated images and the impact on functional avoidance planning. MATERIALS AND METHODS: Paired planning CT and 4DCT scans from 48 patients with NSCLC were randomized to training (n = 41) and testing (n = 7) data sets. The ventilation maps were generated from 4DCT using a Jacobian-based algorithm to provide ground truth labels (CTVI4DCT). A 3D U-Net-based model was trained to map CT to synthetic CTVI (CTVISyn) and validated using fivefold cross-validation. The highest-performing model was applied to the testing set. Spearman's correlation (rs) and Dice similarity coefficient (DSC) determined voxel-wise and functional-wise concordance between CTVI4DCT and CTVISyn. Three plans were designed per patient in the testing set: one clinical plan without CTVI and two functional avoidance plans combined with CTVI4DCT or CTVISyn, aimed at sparing high-functional lungs defined as the top 50% of the percentile ventilation ranges. Dose-volume (DVH) parameters regarding the planning target volume (PTV) and organs at risk (OARs) were recorded. Radiation pneumonitis (RP) risk was estimated using a dose-function (DFH)-based normal tissue complication probability (NTCP) model. RESULTS: CTVISyn showed a mean rs value of 0.65 ± 0.04 compared to CTVI4DCT. Mean DSC values over the top 50% and 60% of ventilation ranges were 0.41 ± 0.07 and 0.52 ± 0.10, respectively. In the test set (n = 7), all patients' RP-risk benefited from CTVI4DCT-guided plans (Riskmean_4DCT_vs_Clinical: 29.24% vs. 49.12%, P = 0.016), and six patients benefited from CTVISyn-guided plans (Riskmean_Syn_vs_Clinical: 31.13% vs. 49.12%, P = 0.022). There were no significant differences in DVH and DFH metrics between CTVISyn and CTVI4DCT-guided plan (P > 0.05). CONCLUSION: Using deep-learning techniques, CTVISyn generated from planning CT exhibited a moderate-to-high correlation with CTVI4DCT. The CTVISyn-guided plans were comparable to the CTVI4DCT-guided plans, effectively reducing pulmonary toxicity in patients while maintaining acceptable plan quality. Further prospective trials are needed to validate these findings.

Clinical evaluation of deep learning-based automatic clinical target volume segmentation: a single-institution multi-site tumor experience.

PURPOSE: The large variability in tumor appearance and shape makes manual delineation of the clinical target volume (CTV) time-consuming, and the results depend on the oncologists' experience. Whereas deep learning techniques have allowed oncologists to automate the CTV delineation, multi-site tumor analysis is often lacking in the literature. This study aimed to evaluate the deep learning models that automatically contour CTVs of tumors at various sites on computed tomography (CT) images from objective and subjective perspectives. METHODS AND MATERIALS: 577 patients were selected for the present study, including nasopharyngeal (n = 34), esophageal (n = 40), breast-conserving surgery (BCS) (left-sided, n = 71; right-sided, n = 71), breast-radical mastectomy (BRM) (left-sided, n = 43; right-sided, n = 37), cervical (radical radiotherapy, n = 45; postoperative, n = 85), prostate (n = 42), and rectal (n = 109) carcinomas. Manually delineated CTV contours by radiation oncologists are served as ground truth. Four models were evaluated: Flexnet, Unet, Vnet, and Segresnet, which are commercially available in the medical product "AccuLearning AI model training platform". The data were divided into the training, validation, and testing set at a ratio of 5:1:4. The geometric metrics, including Dice Similarity Coefficient (DSC) and Hausdorff Distance (HD), were calculated for objective evaluation. For subjective assessment, oncologists rated the segmentation contours of the testing set visually. RESULTS: High correlations were observed between automatic and manual contours. Based on the results of the independent test group, most of the patients achieved satisfactory quantitative results (DSC > 0.8), except for patients with esophageal carcinoma (DSC: 0.62-0.64). The subjective review indicated that 82.65% of predicted CTVs scored either as clinically accepting (8.68%) or requiring minor revision (73.97%), and no patients were scored as rejected. CONCLUSION: This experimental work demonstrated that auto-generated contours could serve as an initial template to help oncologists save time in CTV delineation. The deep learning-based auto-segmentations achieve acceptable accuracy and show the potential to improve clinical efficiency for radiotherapy of a variety of cancer.

CT ventilation image-guided helical Tomotherapy at sparing functional lungs for locally advanced lung cancer: analysis of dose-function metrics and the impact on pulmonary toxicity.

PURPOSE: CT ventilation image (CTVI)-guided radiotherapy that selectively avoids irradiating highly-functional lung regions has potential to reduce pulmonary toxicity. Considering Helical TomoTherapy (HT) has higher modulation capabilities, we investigated the capability and characteristic of HT at sparing functional lungs for locally advanced lung cancer. METHODS AND MATERIALS: Pretreatment 4DCT scans were carried out for 17 patients. Local lung volume expansion (or contraction) during inspiration is related to the volume change at a given lung voxel and is used as a surrogate for ventilation. The ventilation maps were generated from two sets of CT images (peak-exhale and peak-inhale) by deformable registration and a Jacobian-based algorithm. Each ventilation map was normalized to percentile images. Six plans were designed for each patient: one anatomical plan without ventilation map and five functional plans incorporating ventilation map which designed to spare varying degrees of high-functional lungs that were defined as the top 10%, 20%, 30%, 40%, and 50% of the percentile ventilation ranges, respectively. The dosimetric and evaluation factors were recorded regarding planning target volume (PTV) and other organs at risk (OARs), with particular attention to the dose delivered to total lung and functional lungs. An established dose-function-based normal tissue complication probability (NTCP) model was used to estimate risk of radiation pneumonitis (RP) for each scenario. RESULTS: Patients were divided into a benefit group (8 patients) and a non-benefit group (9 patients) based on whether the RP-risk of functional plan was lower than that of anatomical plan. The distance between high-ventilated region and PTV, as well as tumor volume had significant differences between the two groups (P < 0.05). For patients in the benefit group, the mean value of fV5, fV10, fV20, and fMLD (functional V5, V10, V20, and mean lung dose, respectively) were significantly lower starting from top 30% functional plan than in anatomical plan (P < 0.05). With expand of avoidance region in functional plans, the dose coverage of PTV is not sacrificed (P > 0.05) but at the cost of increased dose received by OARs. CONCLUSION: Ventilation image-guided HT plans can reduce the dose received by highly-functional lung regions with a range up to top 50% ventilated area. The spatial distribution of ventilation and tumor size were critical factors to better select patients who could benefit from the functional plan.

Benefits of intraoral stents for sparing normal tissue in radiotherapy of nasopharyngeal carcinoma: a radiobiological model-based quantitative analysis.

BACKGROUND: To determine the value of individualized intraoral stent for normal tissue sparing in radiotherapy of nasopharyngeal carcinoma (NPC) using quantitative analysis of radiobiological model. METHODS: Sixteen patients with NPC who used intraoral stent and 17 patients without intraoral stent were enrolled in this study. All patients underwent Helical Tomotherapy (HT) in our center. Based on the patient's dose volume histogram (DVH), the modified Webb-Nahum model was used to predict tumor control probability (TCP), and the parallel architecture model and Lyman-Kutcher-Burman (LKB) model were used to estimate the normal tissue complications probability (NTCP). The differences of TCP, NTCP and dosimetric parameters between the two groups were compared and analyzed. RESULTS: The mean dose metrics of oral cavity, mandible, left and right parotid gland in patients with intraoral stent was significantly decreased by 11.6%, 12.2%, 15.4%, and 8.7% on average, respectively (P<0.05), while the conformity index (CI, P=0.056) and homogeneity index (HI, P=0.676) of the tumor target showed no statistically different. Quantitative assessment of radiobiological model revealed that the NTCP of oral cavity and parotid glands were both significantly lower in patients with intraoral stent than those without intraoral stent (P<0.001), without compromising TCP of the tumor target (P=0.056). For example, patients using intraoral stent significantly reduced oral mucositis and xerostomia complication probability by 2.52% and 10.11% on average compared to unused ones, respectively. CONCLUSIONS: The custom-made intraoral stents showed promising value at sparing normal tissue during radiotherapy for NPC without affecting target dose coverage or tumor control.

Image-guided hypofractionated simultaneous integrated boost and adaptive radiotherapy for spine metastasis.

BACKGROUND: (I) Dosimetric comparison between Helical tomotherapy (HT) and intensity-modulated radiation therapy (IMRT) with hypofractionated simultaneously integrated boost (HF-SIB) technique for spine metastasis. (II) Mega-voltage CT (MVCT) imaging can be used to monitor changes and calculate the true spinal cord dose over multiple fractions, and compare it with the planned dose by deformable image registration. METHODS: We retrospectively analyzed 57 patients with spine metastasis receiving HT at our institution between March 2016 and March 2019. (I) The dose/fraction to planning gross tumor volume (PGTV) and planning target volume (PTV) were 40 or 50 Gy/10 fractions and 30 or 40 Gy/10 fractions. Plans were compared according to PTV coverage and OARs sparing. (II) The daily MVCT were strictly registered with plan CT according to the rigid structures. Contours of the target and critical organs were then deformed from plan CT to MVCT using MIM deformable registration algorithm. After rigid imaging registration, the planning dose matrix was mapped to the MVCT images. The total doses of 10 fractions to the deformed structures were compared with the planned doses. RESULTS: Comparing HT and IMRT plan averaged over all patients, differences were observed for both homogeneity index (HI) (P<0.001) and conformity index (CI) (P=0.032). The maximum dose of the spinal cord was significantly lower in the HT group than the IMRT group (P=0.037). The actual dose of the spinal cord obtained based on the MVCT modified adaptive radiation therapy (ART) was slightly higher than the initially planned dose, while no significant difference was observed (P=0.083). CONCLUSIONS: HF-SIB with HT was able to achieve lower spinal cord doses, which was dosimetrically superior to IMRT, especially in those cases where strict control of the spinal cord dose is required. With deformable contours from the planning CT, daily MVCT along with rigid dose transformation can provide quantitative dose guidance for replanning.

Dosimetric comparison of helical tomotherapy and conventional Linac-based X-knife stereotactic body radiation therapy for primary lung cancer or pulmonary metastases.

BACKGROUND: To compare helical tomotherapy (HT) with X-knife stereotactic body radiation therapy (HT-SBRT/X-SBRT) for primary or metastatic lung cancer regarding planning target volume (PTV) coverage, such as homogeneity index (HI), conformity index (CI) and dose-volume histogram (DVH) of organs at risk (OARs). METHODS: We retrospectively analyzed 21 patients receiving definitive radiation treatment for non-small cell lung cancer (NSCLC) or pulmonary metastases at our institution between March 2015 and October 2016. Tumors were irradiated with 4-10 Gy per fraction in 5-15 fractions. Plans were compared according to PTV coverage and OARs sparing. RESULTS: Significant differences between HT and X-knife were observed for both HI (P=0.003) and CI (P<0.001). The V5 (P=0.001), V10 (P=0.009), V20 (P=0.001), the mean lung dose (P=0.005) of total lung and maximum dose of the spinal cord (P=0.010) were significantly lower in the X-SBRT group than the HT-SBRT group. There were no significant differences for the V30 (P=0.075) and the mean heart dose (P=0.584) between the two groups. CONCLUSIONS: X-SBRT was dosimetrically superior to HT-SBRT, when applied in these tumors' maximum diameters <5 cm. As HT resulted in increased low-dose volume, it is essential to optimize the patient selection in order to avoid severe radiation pneumonitis in HT-SBRT.

Use of simultaneous radiation boost achieves high treatment response rate in patients with metastatic gastric cancer.

OBJECTIVE: Intensity-modulated radiation therapy (IMRT) with a simultaneous integrated boost (SIB) could improve local control rates at different anatomic sites. However, little is known for its use in metastatic gastric cancer. Our study aimed to compare the treatment response rates of IMRI-SIB and conformal radiotherapy (CRT) in patients with metastatic gastric cancer. MATERIALS AND METHODS: We retrospectively identified twenty patients with metastatic gastric cancer from 2013 to 2015, 12 given IMRT-SIB, and eight given CRT. Treatment response and toxicities were evaluated for all patients. The radiation target included peritoneal lymph nodes. RECIST criteria were used to assess the treatment response. Three patients of eight in the CRT group died before the end of treatment due to the progression of diseases in the field. RESULTS: For the IMRT-SIB group, the median dose of high dose field was 60.8 Gy (50-64.4 Gy), and the median dose of low-dose field was 45 Gy (36-50.4 Gy). For the CRT group, the median dose of the total dose was 50 Gy (41.4-60 Gy). IMRT-SIB could elevate local dose significantly, compared to the CRT group. One patient of 12 in the IMRT-SIB group achieved complete response, and nine patients achieved partial response (PR), whereas no patient achieved CR in the CRT group. Two of five patients achieved PR (40%) in the CRT group. IMRT-SIB improved the treatment response rate significantly (odds ratio 8.33, 95% confidence interval: 1.03-67.14, P = 0.046). Two patients of 12 in the IMRT-SIB group developed enteritis, whereas two patients of five developed enteritis in the CRT group. CONCLUSIONS: IMRT-SIB could escalate the local dose and improve the treatment response rates in patients with metastatic gastric cancer and with acceptable toxicities. Further study with a larger population to validate our data is underway.

Use of Pulsed Low-Dose Rate Radiotherapy in Refractory Malignancies.

BACKGROUND: Most tumor cell lines exhibited low-dose hyperradiosensitivity (LDHRS) to radiation doses lower than 0.3 Gy. Pulsed low-dose rate radiotherapy (PLDR) took advantage of LDHRS and maximized the tumor control process. In this study, we retrospectively analyzed patients receiving PLDR for refractory malignancies. PATIENTS AND METHODS: In total, 22 patients were included in our study: 9 females and 13 males. The median age was 61 years old. All the patients previously received multiline treatments and failed with an estimated survival less than 6 months. Thus, palliative PLDR was given. The PLDR was delivered using 10 fractions of 2 Gy/day, with an interval of 3 minutes, for 5 days per week. The dose rate was 6.67 cGy/min. The median follow-up was 1 year (range 8-30 months). Nine patients underwent PLDR for reirradiation due to locally recurrent diseases. The time interval from last irradiation was 11 to 168 months. Ten patients received PLDR due to poor performance status. Three patients were given PLDR for bulky tumor. The irradiated sites included primary disease (seven patients), locally recurrent disease (nine patients), and retroperitoneal adenopathy (six patients). RESULTS: Five patients developed grade 3 or 4 toxicities. No grade 5 toxicities occurred. All the toxicities recovered after treatments. In general, the 1-year local-regional control rate was approximately 40%, and almost all the patients developed progression at the second year after PLDR. The 6-month survival rate was 76%, and the 1-year survival rate was 69%. For the three patients given PLDR for bulky tumor, all of them achieved partial remission 1 month after the PLDR, and one patient achieved complete response at the fourth month. CONCLUSION: PLDR is an effective and safe option not only for reirradiation but also for patients with poor performance status or bulky tumors. A prospective clinical trial (NCT03061162) is ongoing to validate our results.

Use of Radiomics Combined With Machine Learning Method in the Recurrence Patterns After Intensity-Modulated Radiotherapy for Nasopharyngeal Carcinoma: A Preliminary Study.

Objective: To analyze the recurrence patterns and reasons in patients with nasopharyngeal carcinoma (NPC) treated with intensity-modulated radiotherapy (IMRT) and to investigate the feasibility of radiomics for analysis of radioresistance. Methods: We analyzed 306 NPC patients treated with IMRT from Jul-2009 to Aug-2016, 20 of whom developed with recurrence. For the NPCs with recurrence, CT, MR, or PET/CT images of recurrent disease were registered with the primary planning CT for dosimetry analysis. The recurrences were defined as in-field, marginal or out-of-field, according to dose-volume histogram (DVH) of the recurrence volume. To explore the predictive power of radiomics for NPCs with in-field recurrences (NPC-IFR), 16 NPCs with non-progression disease (NPC-NPD) were used for comparison. For these NPC-IFRs and NPC-NPDs, 1117 radiomic features were quantified from the tumor region using pre-treatment spectral attenuated inversion-recovery T2-weighted (SPAIR T2W) magnetic resonance imaging (MRI). Intraclass correlation coefficients (ICC) and Pearson correlation coefficient (PCC) was calculated to identify influential feature subset. Kruskal-Wallis test and receiver operating characteristic (ROC) analysis were employed to assess the capability of each feature on NPC-IFR prediction. Principal component analysis (PCA) was performed for feature reduction. Artificial neural network (ANN), k-nearest neighbor (KNN), and support vector machine (SVM) models were trained and validated by using stratified 10-fold cross validation. Results: The median follow up was 26.5 (range 8-65) months. 9/20 (45%) occurred in the primary tumor, 8/20 (40%) occurred in regional lymph nodes, and 3/20 (15%) patients developed a primary and regional failure. Dosimetric and target volume analysis of the recurrence indicated that there were 18 in-field, and 1 marginal as well as 1 out-of-field recurrence. With pre-therapeutic SPAIR T2W MRI images available, 11 NPC-IFRs (11 of 18 NPC-IFRs who had available pre-therapeutic MRI) and 16 NPC-NPDs were subsequently employed for radiomic analysis. Results showed that NPC-IFRs vs. NPC-NPDs could be differentiated by 8 features (AUCs: 0.727-0.835). The classification models showed potential in prediction of NPC-IFR with higher accuracies (ANN: 0.812, KNN: 0.775, SVM: 0.732). Conclusion: In-field and high-dose region relapse were the main recurrence patterns which may be due to the radioresistance. After integration in the clinical workflow, radiomic analysis can be served as imaging biomarkers to facilitate early salvage for NPC patients who are at risk of in-field recurrence.

The effect of Mepitel Film on acute radiation-induced skin reactions in head and neck cancer patients: a feasibility study.

OBJECTIVE: Mepitel Film significantly decreases acute radiation-induced skin reactions in breast cancer patients. Here we investigated the feasibility of using Mepitel Film in head and neck cancer patients (ACTRN12614000932662). METHODS: Out of a total of 36 head and neck cancer patients from New Zealand (NZ) (n = 24) and China (n = 12) recruited between June 2015 and December 2016, 33 patients complied with protocol. Of these, 11 NZ patients followed a management protocol; 11 NZ patients and 11 Chinese patients followed a prophylactic protocol. An area of the neck receiving a homogenous radiation dose of > 35 Gy was divided into two equal halves; one half was randomized to Film and the other to either Sorbolene cream (NZ) or Biafine cream (China). Skin reaction severity was measured by Radiation Induced Skin Reaction Assessment Scale and expanded Radiation Therapy Oncology Group toxicity criteria. Skin dose was measured by thermoluminescent dosimeters or gafchromic film. RESULTS: Film decreased overall skin reaction severity (combined Radiation Induced Skin Reaction Assessment Scale score) by 29% and moist desquamation rates by 37% in the Chinese cohort and by 27 and 28%, respectively in the NZ cohort. Mepitel Film did not affect head movements but did not adhere well to the skin, particularly in males with heavy beard stubble, and caused itchiness, particularly in Chinese patients. CONCLUSION: Mepitel Film reduced acute radiation-induced skin reactions in our head and neck cancer patients, particularly in patients without heavy stubble. Advances in knowledge: This is the first study to confirm the feasibility of using Mepitel Film in head and neck cancer patients.