Automatic delineation of organs at risk in non-small cell lung cancer radiotherapy based on deep learning networks

Objective To introduce an end-to-end automatic segmentation method for organs at risk(OARs)in chest computed tomography(CT)images based on dense connection deep learning and to provide an accurate auto-segmentation model to reduce the workload on radiation oncologists.Methods CT images of 36 lung cancer cases were included in this study.Of these,27 cases were randomly selected as the training set,six cases as the validation set,and nine cases as the testing set.The left and right lungs,cord,and heart were auto-segmented,and the training time was set to approximately 5 h.The testing set was evaluated using geometric metrics including the Dice similarity coefficient(DSC),95%Hausdorff distance(HD95),and average surface distance(ASD).Thereafter,two sets of treatment plans were optimized based on manually contoured OARs and automatically contoured OARs,respectively.Dosimetric parameters including Dmax and Vx of the OARs were obtained and compared.Results The proposed model was superior to U-Net in terms of the DSC,HD95,and ASD,although there was no significant difference in the segmentation results yielded by both networks(P>0.05).Compared to manual segmentation,auto-segmentation significantly reduced the segmentation time by nearly 40.7%(P<0.05).Moreover,the differences in dose-volume parameters between the two sets of plans were not statistically significant(P>0.05).Conclusion The bilateral lung,cord,and heart could be accurately delineated using the DenseNet-based deep learning method.Thus,feature map reuse can be a novel approach to medical image auto-segmentation.

Multi-Scale Network for Thoracic Organs Segmentation

Medical Imaging Segmentation is an essential technique for modern medical applications.It is the foundation of many aspects of clinical diagnosis,oncology,and computer-integrated surgical intervention.Although significant successes have been achieved in the segmentation of medical images,DL(deep learning)approaches.Manual delineation of OARs(organs at risk)is vastly dominant but it is prone to errors given the complex irregularities in shape,low texture diversity between tissues and adjacent blood area,patientwide location of organisms,and weak soft tissue contrast across adjacent organs in CT images.Till now several models have been implemented onmulti organs segmentation but not caters to the problemof imbalanced classes some organs have relatively small pixels as compared to others.To segment OARs in thoracic CT images,we proposed the model based on the encoder-decoder approach using transfer learning with the efficientnetB7 DL model.We have built a fully connected CNN(Convolutional Neural network)having 5 layers of encoding and 5 layers of decoding with efficientnetB7 specifically to tackle imbalance class pixels in an accurate way for the segmentation of OARs.Proposed methodology achieves 0.93405 IOU score,0.95138 F1 score and class-wise dice score for esophagus 0.92466,trachea 0.94257,heart 0.95038,aorta 0.9351 and background 0.99891.The results showed that our proposed framework can be segmented organs accurately.

The Impact of Variation in Bladder Volume on the Doses of Target and Organ-at-Risk in Intensity-Modulated Radiation Therapy for Localized Prostate Cancer

Intensity-modulated radiation therapy (IMRT) has become the mainstay of treatment for localized prostate cancer. In IMRT, minimizing differences between the conditions used during planning CT and daily treatment is important to prevent adverse events in normal tissues. In the present study, we evaluated the impact of variation in bladder volume on the doses to various organs. A total of 35 patients underwent definitive radiotherapy at Saitama Medical Center. A Light Speed RT16 (GE Healthcare) was used for planning and to obtain examination CT images. Such images were acquired after 4 - 6 days of planning CT image acquisition. The IMRT plans were optimized using the planning CT data to satisfy the dose constraints set by our in-house protocols for the PTV and the OARs. The dose distributions were then re-calculated using the same IMRT beams, and checked on examination CT images. It was clear that bladder volume affected the doses to certain organs. We focused on the prostate, bladder, rectum, small bowel, and large bowel. Regression coefficients were calculated for variables that correlated strongly with bladder volume (p < 0.05). We found that variation in bladder volume [cm3] predicted deviations in the bladder V70Gy, V50Gy, and V30Gy [%];the maximum dose to the small bowel [cGy];and the maximum dose to the large bowel [cGy]. The regression coefficients were -0.065, -0.125, -0.180, -10.22, and -9.831, respectively. We evaluated the impacts of such variation on organ doses. These may be helpful when checking a patient’s bladder volume before daily IMRT for localized prostate cancer.

Oral contrast agents lead to underestimation of dose calculation in volumetric-modulated arc therapy planning for pelvic irradiation

Background:The effects of oral contrast agents(OCAs)on dosimetry have not been studied in detail.Therefore,this study aimed to examine the influence of OCAs on dose calculation in volumetric-modulated arc therapy plans for rectal cancer.Methods:From 2008 to 2016,computed tomography(CT)images were obtained from 33 rectal cancer patients administered OCA with or without intravenous contrast agent(ICA)and 14 patients who received no contrast agent.CT numbers of organs at risk were recorded and converted to electronic densities.Volumetric-modulated arc therapy plans were designed before and after the original densities were replaced with non-enhanced densities.Doses to the planned target volume(PTV)and organs at risk were compared between the plans.Results:OCA significantly increased the mean and maximum densities of the bowels,while the effects of ICA on these parameters depended on the blood supply of the organs.With OCA,the actual doses for PTV were significantly higher than planned and doses to the bowel increased significantly although moderately.However,the increase in the volume receiving a high-range doses was substantial the absolute change of intestine volume receiving≥52 Gy:1.46[0.05-3.99,cubic centimeter range:-6.74 to 128.12],the absolute change of colon volume receiving≥50 Gy:0.34[0.01-1.53 cc,range:-0.08 to 3.80 cc].Dose changes due to ICA were insignificant.Pearson correlation showed that dose changes were significantly correlated with a high intestinal volume within or near the PTV(ρ>0.5,P<0.05)and with the density of enhanced intestine(ρ>0.3,P<0.05).Conclusions:Contrast agents applied in simulation cause underestimation of doses in actual treatment.The overdose due to ICA was slight,while that due to OCA was moderate.The bowel volume receiving≥50Gy was dramatically increased when OCA within the bowel was absent.Physicians should be aware of these issues if the original plan is barely within clinical tolerance or if a considerable volume of enhanced intestine is within or near the PTV.