Search

Background/Objectives: We assessed the influence of local patients and clinical characteristics on the performance of commercial deep learning (DL) segmentation models for head-and-neck (HN), breast, and prostate cancers. Methods: Clinical computed tomography (CT) scans and clinically approved contours of 210 patients (53 HN, 49 left breast, 55 right breast, and 53 prostate cancer) were used to train and validate segmentation models integrated within a vendor-supplied DL training toolkit and to assess the performance of both vendor-pretrained and custom-trained models. Four custom models (HN, left breast, right breast, and prostate) were trained and validated with 30 (training)/5 (validation) HN, 34/5 left breast, 39/5 right breast, and 30/5 prostate patients to auto-segment a total of 24 organs at risk (OARs). Subsequently, both vendor-pretrained and custom-trained models were tested on the remaining patients from each group. Auto-segmented contours were evaluated by comparing them with clinically approved contours via the Dice similarity coefficient (DSC) and mean surface distance (MSD). The performance of the left and right breast models was assessed jointly according to ipsilateral/contralateral locations. Results: The average DSCs for all structures in vendor-pretrained and custom-trained models were as follows: 0.81 ± 0.12 and 0.86 ± 0.11 in HN; 0.67 ± 0.16 and 0.80 ± 0.11 in the breast; and 0.87 ± 0.09 and 0.92 ± 0.06 in the prostate. The corresponding average MSDs were 0.81 ± 0.76 mm and 0.76 ± 0.56 mm (HN), 4.85 ± 2.44 mm and 2.42 ± 1.49 mm (breast), and 2.17 ± 1.39 mm and 1.21 ± 1.00 mm (prostate). Notably, custom-trained models showed significant improvements over vendor-pretrained models for 14 of 24 OARs, reflecting the influence of data/contouring variations in segmentation performance. Conclusions: These findings underscore the substantial impact of institutional preferences and clinical practices on the implementation of vendor-pretrained models. We also found that a relatively small amount of institutional data was sufficient to train customized segmentation models with sufficient accuracy. [ABSTRACT FROM AUTHOR]

Simple Summary: Radiation therapy is becoming increasingly important in the treatment of liver and pancreatic tumors, particularly in situations where high doses of radiation can be delivered safely. However, there are several challenges to treating tumors in the abdomen, including poor visibility of the tumor and movements due to breathing and digestion. The typical imaging available at the time of treatment makes it difficult to see both the tumor and nearby portions of the digestive tract, which is particularly sensitive to radiation damage. This paper describes the workflow involved when using high-quality computed tomography imaging at the time of treatment, to ensure that the tumor is accurately targeted and normal tissues are avoided. This study shows that by using these images and the planned dose distribution, the dose to normal structures can be maintained below specified targets. With this technology and workflow, more patients can benefit from high-dose radiation treatment to the liver and pancreas. Background/Objectives: Stereotactic body radiation therapy (SBRT) for abdominal targets faces a variety of challenges, including motion caused by the respiration and digestion and a relatively poor level of contrast between the tumor and the surrounding tissues. Breath-hold treatments with computed tomography-on-rails (CTOR) image guidance is one way of addressing these challenges, allowing for both the tumor and normal tissues to be well-visualized. Using isodose lines (IDLs) from CT simulations as a guide, the anatomical information can be used to shift the alignment or trigger a replan, such that normal tissues receive acceptable doses of radiation. Methods: This study aims to describe the workflow involved when using CTOR for pancreas and liver SBRT and demonstrates its effectiveness through several case studies. Results: In these case studies, using the anatomical information gained through diagnostic-quality CT guidance to make slight adjustments to the alignment, resulted in reductions in the maximum dose to the stomach. Conclusions: High-quality imaging, such as CTOR, and the use of IDLs to estimate the doses to OARs, enable the safe delivery of SBRT, without the added complexity and resource commitment required by daily online adaptive planning. [ABSTRACT FROM AUTHOR]

Purpose: To estimate the effects of interfractional anatomic changes on dose to organs at risk (OARs) and tumors, as measured with cone beam computed tomography (CBCT) image guidance for pancreatic stereotactic body radiation therapy.Methods and Materials: We evaluated 11 patients with pancreatic cancer whom were treated with stereotactic body radiation therapy (33-40 Gy in 5 fractions) using daily CT-on-rails (CTOR) image guidance immediately before treatment with breath-hold motion management. CBCT alignment was simulated in the treatment planning software by aligning the original planning CT to each fractional CTOR image set via fiducial markers. CTOR data sets were used to calculate fractional doses after alignment by applying the rigid shift of the planning CT and CTOR image sets to the planning treatment isocenter and recalculating the fractional dose. Accumulated dose to the gross tumor volume (GTV), tumor vessel interface, duodenum, small bowel, and stomach were calculated by summing the 5 fractional absolute dose-volume histograms into a single dose-volume histogram for comparison with the original planned dose.Results: Four patients had a GTV D100% of at least 1.5 Gy less than the fractional planned value in several fractions; 4 patients had fractional underestimation of duodenum dose by 1.0 Gy per fraction. The D1.0 cm3 <35 Gy constraint was violated for at least 1 OAR in 3 patients, with either the duodenum (n = 2) or small bowel (n = 1) D1.0 cm3 being higher on the accumulated dose distribution (P = .01). D100% was significantly lower according to accumulated dose GTV (P = .01) and tumor vessel interface (P = .02), with 4 and 2 patients having accumulated D100%  ≥4 Gy lower than the planned value for the GTV and tumor vessel interface, respectively.Conclusions: For some patients, CBCT image guidance based on fiducial alignment may cause large dosimetric uncertainties for OARs and target structures, according to accumulated dose. [ABSTRACT FROM AUTHOR]

Purpose: To study radiation-induced esophageal expansion as an objective measure of radiation esophagitis in patients with non-small cell lung cancer (NSCLC) treated with intensity modulated radiation therapy.Methods and Materials: Eighty-five patients had weekly intra-treatment CT imaging and esophagitis scoring according to Common Terminlogy Criteria for Adverse Events 4.0, (24 Grade 0, 45 Grade 2, and 16 Grade 3). Nineteen esophageal expansion metrics based on mean, maximum, spatial length, and volume of expansion were calculated as voxel-based relative volume change, using the Jacobian determinant from deformable image registration between the planning and weekly CTs. An anatomic variability correction method was validated and applied to these metrics to reduce uncertainty. An analysis of expansion metrics and radiation esophagitis grade was conducted using normal tissue complication probability from univariate logistic regression and Spearman rank for grade 2 and grade 3 esophagitis endpoints, as well as the timing of expansion and esophagitis grade. Metrics' performance in classifying esophagitis was tested with receiver operating characteristic analysis.Results: Expansion increased with esophagitis grade. Thirteen of 19 expansion metrics had receiver operating characteristic area under the curve values >0.80 for both grade 2 and grade 3 esophagitis endpoints, with the highest performance from maximum axial expansion (MaxExp1) and esophageal length with axial expansion ≥30% (LenExp30%) with area under the curve values of 0.93 and 0.91 for grade 2, 0.90 and 0.90 for grade 3 esophagitis, respectively.Conclusions: Esophageal expansion may be a suitable objective measure of esophagitis, particularly maximum axial esophageal expansion and esophageal length with axial expansion ≥30%, with 2.1 Jacobian value and 98.6 mm as the metric value for 50% probability of grade 3 esophagitis. The uncertainty in esophageal Jacobian calculations can be reduced with anatomic correction methods. [ABSTRACT FROM AUTHOR]