Novel in-house knowledge-based automated planning system for lung cancer treated with intensity-modulated radiotherapy.

PURPOSE: The goal of this study was to propose a knowledge-based planning system which could automatically design plans for lung cancer patients treated with intensity-modulated radiotherapy (IMRT). METHODS AND MATERIALS: From May 2018 to June 2020, 612 IMRT treatment plans of lung cancer patients were retrospectively selected to construct a planning database. Knowledge-based planning (KBP) architecture named αDiar was proposed in this study. It consisted of two parts separated by a firewall. One was the in-hospital workstation, and the other was the search engine in the cloud. Based on our previous study, A­Net in the in-hospital workstation was used to generate predicted virtual dose images. A search engine including a three-dimensional convolutional neural network (3D CNN) was constructed to derive the feature vectors of dose images. By comparing the similarity of the features between virtual dose images and the clinical dose images in the database, the most similar feature was found. The optimization parameters (OPs) of the treatment plan corresponding to the most similar feature were assigned to the new plan, and the design of a new treatment plan was automatically completed. After αDiar was developed, we performed two studies. The first retrospective study was conducted to validate whether this architecture was qualified for clinical practice and involved 96 patients. The second comparative study was performed to investigate whether αDiar could assist dosimetrists in improving the quality of planning for the patients. Two dosimetrists were involved and designed plans for only one trial with and without αDiar; 26 patients were involved in this study. RESULTS: The first study showed that about 54% (52/96) of the automatically generated plans would achieve the dosimetric constraints of the Radiation Therapy Oncology Group (RTOG) and about 93% (89/96) of the automatically generated plans would achieve the dosimetric constraints of the National Comprehensive Cancer Network (NCCN). The second study showed that the quality of treatment planning designed by junior dosimetrists was improved with the help of αDiar. CONCLUSIONS: Our results showed that αDiar was an effective tool to improve planning quality. Over half of the patients' plans could be designed automatically. For the remaining patients, although the automatically designed plans did not fully meet the clinical requirements, their quality was also better than that of manual plans.

Dosimetric effect of intensity-modulated radiation therapy for postoperative non-small cell lung cancer with and without air cavity in the planning target volume.

To evaluate the dosimetric effect of intensity-modulated radiation therapy (IMRT) for postoperative non-small cell lung cancer (NSCLC), with and without the air cavity in the planning target volume (PTV). Two kinds of IMRT plans were made for 21 postoperative NSCLC patients. In Plan-0: PTV included the tracheal air cavity, and in Plan-1: the air cavity was subtracted from the PTV. For PTV, the dosimetric parameters, including Dmean, D98, D95, D2, D0.2, conformity index (CI), and homogeneity index (HI) were evaluated. For organs at risk (OARs), the evaluation indexes, included the V5, V20 and the mean lung dose (MLD) of total lung, the V30, V40, and the mean heart dose (MHD) of heart, the spinal cord Dmax, and the V35 and the mean esophageal dose (MED) of esophagus. The number of segments and MUs were also recorded. Additionally, the correlation between the Plan-1 dosimetric change value relative to Plan-0, the size of air cavity, and the volume proportion of the cavity in the PTV was also analyzed. The Dmean of PTV, D2, D0.2, HI and CI in Plan-1 decreased compared with those in Plan-0. For OARs, the V30, MHD, and MED also decreased. The CI change value of Plan-1 relative to Plan-0 had a significantly negative correlation with the size and the volume proportion of air cavity, and the MED change value also had a significantly negative correlation with the air cavity size. The IMRT plans for patients with postoperative NSCLC can achieve a better target dose distribution and offer an improved sparing of the heart and esophagus by removing the PTV air cavity, while reducing the target conformity. The change value of CI and MED had a significantly negative correlation with the air cavity size.

A 2D-3D hybrid convolutional neural network for lung lobe auto-segmentation on standard slice thickness computed tomography of patients receiving radiotherapy.

BACKGROUND: Accurate segmentation of lung lobe on routine computed tomography (CT) images of locally advanced stage lung cancer patients undergoing radiotherapy can help radiation oncologists to implement lobar-level treatment planning, dose assessment and efficacy prediction. We aim to establish a novel 2D-3D hybrid convolutional neural network (CNN) to provide reliable lung lobe auto-segmentation results in the clinical setting. METHODS: We retrospectively collected and evaluated thorax CT scans of 105 locally advanced non-small-cell lung cancer (NSCLC) patients treated at our institution from June 2019 to August 2020. The CT images were acquired with 5 mm slice thickness. Two CNNs were used for lung lobe segmentation, a 3D CNN for extracting 3D contextual information and a 2D CNN for extracting texture information. Contouring quality was evaluated using six quantitative metrics and visual evaluation was performed to assess the clinical acceptability. RESULTS: For the 35 cases in the test group, Dice Similarity Coefficient (DSC) of all lung lobes contours exceeded 0.75, which met the pass criteria of the segmentation result. Our model achieved high performances with DSC as high as 0.9579, 0.9479, 0.9507, 0.9484, and 0.9003 for left upper lobe (LUL), left lower lobe (LLL), right upper lobe (RUL), right lower lobe (RLL), and right middle lobe (RML), respectively. The proposed model resulted in accuracy, sensitivity, and specificity of 99.57, 98.23, 99.65 for LUL; 99.6, 96.14, 99.76 for LLL; 99.67, 96.13, 99.81 for RUL; 99.72, 92.38, 99.83 for RML; 99.58, 96.03, 99.78 for RLL, respectively. Clinician's visual assessment showed that 164/175 lobe contours met the requirements for clinical use, only 11 contours need manual correction. CONCLUSIONS: Our 2D-3D hybrid CNN model achieved accurate automatic segmentation of lung lobes on conventional slice-thickness CT of locally advanced lung cancer patients, and has good clinical practicability.

Automatic segmentation of lung tumors on CT images based on a 2D & 3D hybrid convolutional neural network.

OBJECTIVE: A stable and accurate automatic tumor delineation method has been developed to facilitate the intelligent design of lung cancer radiotherapy process. The purpose of this paper is to introduce an automatic tumor segmentation network for lung cancer on CT images based on deep learning. METHODS: In this paper, a hybrid convolution neural network (CNN) combining 2D CNN and 3D CNN was implemented for the automatic lung tumor delineation using CT images. 3D CNN used V-Net model for the extraction of tumor context information from CT sequence images. 2D CNN used an encoder-decoder structure based on dense connection scheme, which could expand information flow and promote feature propagation. Next, 2D features and 3D features were fused through a hybrid module. Meanwhile, the hybrid CNN was compared with the individual 3D CNN and 2D CNN, and three evaluation metrics, Dice, Jaccard and Hausdorff distance (HD), were used for quantitative evaluation. The relationship between the segmentation performance of hybrid network and the GTV volume size was also explored. RESULTS: The newly introduced hybrid CNN was trained and tested on a dataset of 260 cases, and could achieve a median value of 0.73, with mean and stand deviation of 0.72 ± 0.10 for the Dice metric, 0.58 ± 0.13 and 21.73 ± 13.30 mm for the Jaccard and HD metrics, respectively. The hybrid network significantly outperformed the individual 3D CNN and 2D CNN in the three examined evaluation metrics (p < 0.001). A larger GTV present a higher value for the Dice metric, but its delineation at the tumor boundary is unstable. CONCLUSIONS: The implemented hybrid CNN was able to achieve good lung tumor segmentation performance on CT images. ADVANCES IN KNOWLEDGE: The hybrid CNN has valuable prospect with the ability to segment lung tumor.

A novel lung-avoidance planning strategy based on 4DCT ventilation imaging and CT density characteristics for stage III non-small-cell lung cancer patients.

BACKGROUND: Functional planning based merely on 4DCT ventilation imaging has limitations. In this study, we proposed a radiotherapy planning strategy based on 4DCT ventilation imaging and CT density characteristics. MATERIALS AND METHODS: For 20 stage III non-small-cell lung cancer (NSCLC) patients, clinical plans and lung-avoidance plans were generated. Through deformable image registration (DIR) and quantitative image analysis, a 4DCT ventilation map was calculated. High-, medium-, and low-ventilation regions of the lung were defined based on the ventilation value. In addition, the total lung was also divided into high-, medium-, and low-density areas according to the HU threshold. The lung-avoidance plan aimed to reduce the dose to functional and high-density lungs while meeting standard target and critical structure constraints. Standard and dose-function metrics were compared between the clinical and lung-avoidance plans. RESULTS: Lung avoidance plans led to significant reductions in high-function and high-density lung doses, without significantly increasing other organ at risk (OAR) doses, but at the expense of a significantly degraded homogeneity index (HI) and conformity index (CI; p < 0.05) of the planning target volume (PTV) and a slight increase in monitor units (MU) as well as in the number of segments (p > 0.05). Compared with the clinical plan, the mean lung dose (MLD) in the high-function and high-density areas was reduced by 0.59 Gy and 0.57 Gy, respectively. CONCLUSION: A lung-avoidance plan based on 4DCT ventilation imaging and CT density characteristics is feasible and implementable, with potential clinical benefits. Clinical trials will be crucial to show the clinical relevance of this lung-avoidance planning strategy.

Geometric and Dosimetric Changes in Tumor and Lung Tissue During Radiotherapy for Lung Cancer With Atelectasis.

BACKGROUND AND PURPOSE: This article retrospectively characterized the geometric and dosimetric changes in target and normal tissues during radiotherapy for lung cancer patients with atelectasis. MATERIALS AND METHODS: A total of 270 cone beam computed tomography (CBCT) scans of 18 lung patients with atelectasis were collected. The degree and time of resolution or expansion of the atelectasis were recorded. The geometric, dosimetric, and biological changes in the target and lung tissue were also quantified. RESULTS: There were two patients with expansion, four patients with complete regression, six patients with partial regression, and six patients with no change. The time of resolution or expansion varied. The tumor volume increased by 3.8% in the first seven fractions, then decreased from the 9th fraction, and by 33.4% at the last CBCT. In the LR direction, the average center of mass (COM), boundaries of the tumors gradually shifted mediastinally. In the AP direction, the COM of the tumors was shifted slightly in the posterior direction and then gradually shifted to the anterior direction; the boundaries of the tumors all moved mediastinally. In the SI direction, the COM of the tumors on the right side of the body was substantially shifted toward the head direction. The boundaries of the tumors varied greatly. D2, D98, Dmean, V95, V107, and TCP of the PTV were reduced during radiotherapy and were reduced to their lowest values during the last two fractions. The volume of the ipsilateral lung tended to increase gradually. The V5, V10, V20, V30, V40, and NTCP of the total lung gradually increased with the fraction. CONCLUSIONS: For most patients, regression of the atelectasis occurred, and the volume of the ipsilateral lung tended to increase while the tumor volume decreased, and the COM and boundary of the tumors shifted toward mediastinum, which caused an insufficient dose to the target and an overdose to the lungs. Regression or expansion may occur for any fraction, and it is therefore recommended that CBCT be performed at least every other day.

Regression models for predicting physical and EQD2 plan parameters of two methods of hybrid planning for stage III NSCLC.

BACKGROUND/PURPOSE: To establish regression models of physical and equivalent dose in 2 Gy per fraction (EQD2) plan parameters of two kinds of hybrid planning for stage III NSCLC. METHODS: Two kinds of hybrid plans named conventional fraction radiotherapy & stereotactic body radiotherapy (C&S) and conventional fraction radiotherapy & simultaneous integrated boost (C&SIB) were retrospectively made for 20 patients with stage III NSCLC. Prescription dose of C&S plans was 2 Gy × 30f for planning target volume of lymph node (PTVLN) and 12.5 Gy × 4f for planning target volume of primary tumor (PTVPT), while prescription dose of C&SIB plans was 2 Gy × 26f for PTVLN and sequential 2 Gy × 4f for PTVLN combined with 12.5 Gy × 4f for PTVPT. Regression models of physical and EQD2 plan parameters were established based on anatomical geometry features for two kinds of hybrid plans. The features were mainly characterized by volume ratio, min distance and overlapping slices thickness of two structures. The possibilities of regression models of EQD2 plan parameters were verified by spearman's correlation coefficients between physical and EQD2 plan parameters, and the influence on the consistence of fitting goodness between physical and EQD2 models was investigated by the correlations between physical and EQD2 plan parameters. Finally, physical and EQD2 models predictions were compared with plan parameters for two new patients. RESULTS: Physical and EQD2 plan parameters of PTVLN CI60Gy have shown strong positive correlations with PTVLN volume and min distance(PT to LN), and strong negative correlations with PTVPT volume for two kinds of hybrid plans. PTV(PT+LN) CI60Gy is not only correlated with above three geometry features, but also negatively correlated with overlapping slices thickness(PT and LN). When neck lymph node metastasis was excluded from PTVLN volume, physical and EQD2 total lung V20 showed a high linear correlation with corrected volume ratio(LN to total lung). Meanwhile, physical total lung mean dose (MLD) had a high linear correlation with corrected volume ratio(LN to total lung), while EQD2 total lung MLD was not only affected by corrected volume ratio(LN to total lung) but also volume ratio(PT to total lung). Heart D5, D30 and mean dose (MHD) would be more susceptible to overlapping structure(heart and LN). Min distance(PT to ESO) may be an important feature for predicting EQD2 esophageal max dose for hybrid plans. It's feasible for regression models of EQD2 plan parameters, and the consistence of the fitting goodness of physical and EQD2 models had a positive correlation with spearman's correlation coefficients between physical and EQD2 plan parameters. For total lung V20, ipsilateral lung V20, and ipsilateral lung MLD, the models could predict that C&SIB plans were higher than C&S plans for two new patients. CONCLUSION: The regression models of physical and EQD2 plan parameters were established with at least moderate fitting goodness in this work, and the models have a potential to predict physical and EQD2 plan parameters for two kinds of hybrid planning.

On the optimal number of dose-limiting shells in the SBRT auto-planning design for peripheral lung cancer.

PURPOSE: The number of dose-limiting shells in the optimization process is one of the key factors determining the quality of stereotactic body radiotherapy (SBRT) auto-planning in the Pinnacle treatment planning system (TPS). This study attempted to derive the optimal number of shells by evaluating the auto-plans designed with different number of shells for peripheral lung cancer patients treated with SBRT. METHODS: Identical treatment technique, optimization process, constraints, and dose calculation algorithm in the Pinnacle TPS were retrospectively applied to 50 peripheral lung cancer patients who underwent SBRT in our center. For each of the patients, auto-plans were optimized based on two shells, three shells, four shells, five shells, six shells, seven shells, eight shells, respectively. The optimal number of shells for the SBRT auto-planning was derived through the evaluations and comparisons of various dosimetric parameters of planning target volume (PTV) and organs at risk (OARs), monitor units (MU), and optimization time of the plans. RESULTS: The conformity index (CI) and the gradient index (GI) of PTV, the maximum dose outside the 2 cm of PTV (D2cm ), Dmax of spinal cord (SCmax ), the percentage of volume of total lung excluding ITV receiving 20 Gy (V20) and 10 Gy (V10), and the mean lung dose (MLD) were improved when the number of shell increased, but the improvement became not significant as the number of shell reached six. The monitor units (MUs) varied little among different plans where no statistical differences were found. However, as the number of shell increased, the auto-plan optimization time increased significantly. CONCLUSIONS: It appears that for peripheral lung SBRT plan using six shells can yield satisfactory plan quality with acceptable beam MUs and optimization time in the Pinnacle TPS.