ABSTRACT
Objective: Applying radiomics and dosimetry features to establish machine learning models, which is used to predict the 2-year survival of esophageal patients with radiotherapy. Methods: Retrospective analysis of 579 esophageal cancer patients who underwent radiotherapy from January 2013 to December 2017 in Tianjin Medical University Cancer Institute and Hospital. Radiomics and dosimetry features were extracted from the GTV of the radiotherapy plan for patients with esophageal cancer. The maximum correlation and minimum redundancy and manual methods were used to reduce the feature vector. A total of 14 radiomics and 14 dosimetry features were selected, then normalized to the range [0,1]. The machine learning models such as support vector machines (SVM), Logistic regression (LR), and random forest (RF) were used to train and test the radiomics and dosimetry features, respectively, then to predict the 2-year survival of esophageal cancer patients treated with radiotherapy. Results: When only the radiomics features were used to predict the 2-year survival after radiotherapy, the accuracy of SVM, LR and RF models were 84.98%, 85.92% and 84.51%, respectively. Furthermore, when the combined features of radiomics and dosimetry were used for prediction, the accuracy of the SVM, LR and RF models were 86.32%, 83.02% and 90.01%, respectively. Using the radiomics and dosimetry features, the predictive accuracy of SVM and RF models are effectively improved. Conclusions: For the SVM and RF models, the radiomics and dosimetry features can effectively improve the accuracy of predicting 2-year survival for esophageal cancer patients after radiation therapy.
ABSTRACT
Objective:An automatic analysis system for radiotherapy planning was developed to realize the automatic analysis of dose parameters of radiotherapy planning through the intelligent analysis of the underlying data of Pinnacle 3 treatment planning system (TPS). Methods:The radiotherapy plans of 12 patients with esophagus cancer were analyzed. The automatic analysis system automatically retrieved the Pinnacle 3 TPS database, obtained the raw data of 12 cases of treatment plan, and automatically analyzed the underlying raw data, reconstructed contours, radiation fields, and dose parameters, and recalculated dose distribution and dose-volume histograms. The accuracy of the recalculation of the volume and dose data of the new system was evaluated by comparing with volume and dose data from the original plans of online TPS. Results:The automatic analysis system successfully parsed the underlying data of the treatment plan and reconstructed the parameters of the treatment plan. The volume deviation between the contour calculated by the new system and the original plans was ≤0.1%; Compared with the reference dose of the original plans, the deviations of dose parameters (D max, D mean, D 95, and D 50 for GTV, PGTV, CTV, and PTV) recalculated by the new analysis system were ≤1.0%; The deviations of D max and D mean of recalculated ROIs from the original plans were <5%. Conclusions:The automatic analysis system can directly analyze the underlying data of the Pinnacle 3 TPS treatment plan, reconstruct the treatment plan, calculate the contour volume and dose parameters, and the dose deviations from the original plans meet clinical requirements
ABSTRACT
Objective To evaluate the performance of progressive optimization algorithm-based Auto-Planning module in automated volumetric-modulated arc therapy(VMAT)planning for nasopharyngeal carcinoma. Methods Thirteen treated VMAT plans of nasopharyngeal carcinoma were re-planed with Auto-Planning module. Only one cycle of automated optimization of the Auto-Planning module was performed for each plan without any manual intervention. The dosimetric parameters of the automated treatment plans were compared with those of the manual plans. Paired t-test was used for statistical analysis. The time required for automated planning using the Auto-Planning module was also measured. Results All plans generated with the Auto-Planning module met the routine dosimetric requirements and were acceptable for clinical use. The homogeneity index of targets was superior in the automated plans than in manual plans(P= 0.000).In addition,the automated plans had significantly improved protection for some organs at risk than the manual plans. The mean dose to the left and right parotids were reduced by 7.75 Gy(P=0.000)and 5.79 Gy(P=0.000)in the automated plans,respectively. Furthermore,the V60(0.58% vs. 3.12%,P=0.000)and Dmean(34.11 Gy vs. 40.78 Gy,P= 0.000)of the mandible were also significantly lower with Auto-Planning than with manual planning. Conclusions Auto-Planning module can improve the overall quality and consistency of treatment plans,and reduce the workload and time of treatment planning,resulting in substantially enhanced treatment planning efficiency.
ABSTRACT
Objective To analyze the impact of dose calculation resolution on Gamma pass rate for planar dose distribution verification of intensity-modulated radiotherapy (IMRT),and to find out the appropriate calculation resolution value with specific applied criteria.Methods 25 IMRT plans were selected for the planar dose verification.The Gamma pass rates with 3 mm/3％,2 mm/2％ and 1 mm/1％ criteria were compared and analyzed.The planar dose distributions were calculatedand exported from TPS with different resolutions of 0.5,1,2,3,4 and 5 mm.The pass rates of Gamma analysis between the computed dose distributions with different resolutions and the dose distributions measured by Mapcheck device were compared and analyzed.Results The average Gamma pass rates of all the 225 fields decreased with the increment of calculation resolution.When the pass rates of Gamma analysis were calculated using 3 mm/3％ criteria,the gamma pass rate with 1 mm and 0.5 mm calculation resolution were (98.3±1.3)％ and (98.3±1.2)％,respectively.The results were almost the same and the difference wasn't statistically significant (P＞0.05).The gamma pass rates with greater than or equal to3 mm calculation resolution were significantly less than the pass rate with 0.5 mm calculation resolution (P＜0.05).When the pass rates of Gamma analysis were calculated using 2 mm/2％ criteria,all the pass rates with greater than 0.5 mm calculation resolution were significantly less than the pass rate with 0.5 mm calculation resolution (P＜0.05).Conclusions The Gamma pass rates of planar dose verification reduce significantly with the increment of calculation resolution.In order to reduce the effect of calculation resolution on Gamma pass rate,the lower calculation resolution should be used.When using the 3 mm/3％ criteria for the Gamma analysis,1 mm calculation resolution is recommended.When using the 2 mn/2％ criteria for the Gamma analysis,0.5 mm calculation resolution is recommended.The appropriate calculation resolution will ensure the reliability of planar dose verification.
ABSTRACT
Objective To analyze the anatomic changes and dosimetric variations of patients with head-and-neck cancer during intensity-modulated radiotherapy and to assess the necessity of re-planning the treatment course.Methods Twenty-one nasopharyngeal carcinoma patients were selected to receive the second CT scan in the course of intensity-modulated radiotherapy,targets and organs at risk were re-contoured on the new CT images with the help of deformable registration software.Actual dose distribution delivered by the original treatment plan was calculated on the second CT image,and then the volume and doses of targets and organs at risk were compared between the planning CT and second CT.Results The volume of patient's head-and-neck outlines and GTVnd decreased significantly,the volume of cord and brainstem didn't change much,while the volume of left and right parotids decreased significantly by (24.23 ± 12.15)％ and (25.82± 10.46)％,respectively.The parameters D95％ and Dmean of PGTVnx kept stable,but the parameters D1cc of spinal cord PRV and brainstem PRV increased by (8.12± 10.32)％ and (14.60±18.85)％ respectively.The mean dose of the left and right parotids increased significantly by (27.43±17.67)％ and (26.76±12.46)％,respectively.Conclusion The anatomical changes of patients undergoing intensity-modulated radiotherapy will cause significant dosimetric variations of the cord,brainstem and parotids,so it's meaningful and necessary to re-design the treatment plan in the course of radiotherapy.Repeat CT imaging and replanning during the course of IMRT is essential to ensure adequate doses to target volumes and safe doses to normal tissue.
ABSTRACT
Objective To quantify the incidental irradiation dose (ⅡD) to lymph node stations of esophagus when treating patients with T1-4N0 M0 thoracic esophageal squamous cell carcinoma (ESCC) with a dose of 60 Gy/30f.Methods Twenty-nine patients with medically inoperable T1-4N0M0 thoracic ESCC were treated with three-dimensional radiotherapy on involved-field.The conformal CTV was re-created using a 3 cm margin in the proximal and distal direction (following the course of the esophagus) beyond the barium esophagogram,endoscopic examination and CT defined GTV and a 0.5 cm margin in the lateral and anteroposterior directions of the CT defined GTV.The PTV encompassed 1 cm proximal and distal margins,0.5 cm radiaI margin on the basis of CTV.Cervical,mediastinal and abdominal lymph nodes were delineated respectively.Equivalent uniform dose (EUD) and other dosimetric paraneters were calculated for each nodal station.Nodal region whose metastasis rate is greater than 5％ was considered a high risk lymph node subgroups.Results Under a 60 Gy dose prescription,the median Dmean and EUD,V40 and V50 were ≥40 Gy,≥85％ and ≥75％ in most of the high risk nodal regions.For the subgroups whose EUD were less than 40 Gy,most of the ⅡD of these regions was significantly associated with the length and location of esophageal tumor (r =0.892,P =0.000).Conclusions Lymph node stations nearby of ESCC received considerable ⅡD with involved-field irradiation which could control subclinical lesions.But more clinical studies should be needed.
ABSTRACT
ObjectiveTo investigate the application of non-contrast and contrast-enhanced 18FDG PET/CT in the delineation of gross tumor volume ( GTV ) of pancreatic cancer.MethodsBetween Jan.2008 and Dec.2009,twenty-one patients with unresectable locally advanced pancreatic cancer or recurrent pancreatic cancer after surgery in our hospital had both non-contrast CT and PET images acquired at the same body position.Among the whole group,eleven patients also had contrast CT images.The image data sets were transferred to the treatment planning workstation for registration.Then gross tumor volumes ( GTV )were delineated independently using the information of PET images,contrast/non-contrast CT scan and contrast/non-contrast PET-CT fusion images.The differences of mean volume in these different sets of GTV were analyzed.Results For the whole group,the mean volume of non-contrast GTVCT,GTVPET,noncontrast GTVPET-CT were 76.9 cm3,47.0 cm3 and 44.5 cm3,respectively.The mean volume of non-contrast GTVPET-CT was significantly smaller than non-contrast GTVCT ( z =-3.91,P =0.000 ).For the eleven patients with contrast CT,the mean volume of contrast GTVCT,GTVPET,contrast GTVPET-CT were 64.1 cm3,45.1 cm3 and 49.3 cm3,respectively.The mean volume of contrast GTVPET-CT was significantly smaller than contrast GTVCT (z =-2.13,P =0.033 ).No significant differences were found between contrast PET-CT and non-contrast PET-CT (z =-0.80,P =0.424).ConclusionsCo-registration of PET and contrast/noncontrast CT information in pancreatic cancer may improve the accuracy of GTV delineation,and possibly reduce the adverse effect of irradiation.