Towards clinical implementation of ultrafast combined kV-MV CBCT for IGRT of lung cancer : Evaluation of registration accuracy based on phantom study

Deep learning-based bone suppression in chest radiographs using CT-derived features: a feasibility study

Background

Bone suppression of chest X-ray holds the potential to improve the accuracy of target localization in image-guided radiation therapy (IGRT). However, the training dataset for bone suppression is limited because of the scarcity of bone-free radiographs. This study aims to develop a deep learning-based bone suppression method using CT-derived features to reduce the reliance on the bone-free dataset.

Methods

In this study, 59 high-resolution lung CT scans were processed to generate the lung digital radiographs (DRs), bone DRs, and bone-free DRs, for the training and internal validation of the proposed cascade convolutional neural network (CCNN). A three-stage image processing framework (CT segmentation, DR simulation, and feature expansion) was developed to expand simulated lung DRs with different weightings of bone intensity. The CCNN consists of a bone detection network and a bone suppression network. In external validation, the trained CCNN was evaluated using 30 chest radiographs. The synthesized bone-suppressed radiographs were compared with the bone-suppressed reference in terms of peak signal-to-noise ratio (PSNR), mean absolute error (MAE), structural similarity index measure (SSIM), and Spearman's correlation coefficient. Furthermore, the effectiveness of the proposed feature expansion method and CCNN model were assessed via the ablation experiment and replacement experiment, respectively.

Results

Evaluation on real chest radiographs showed that the bone-suppressed chest radiographs closely matched with the bone-suppressed reference, achieving an accuracy of MAE =0.0087±0.0030, SSIM =0.8458±0.0317, correlation of 0.9554±0.0170, and PNSR of 20.86±1.60. After removing the feature expansion from the CCNN model, the performance decreased in terms of MAE (0.0294±0.0093, -237.9%), SSIM (0.7747±0.0.0416, -8.4%), correlation (0.8772±0.0271, -8.2%), and PSNR (15.53±1.42, -25.5%) metrics.

Conclusions

We successfully demonstrated a novel deep learning-based bone suppression method using CT-derived features to reduce the reliance on the bone-free dataset. Implementation of the feature expansion procedures resulted in a remarkable reinforcement of the model performance. For the application of target localization in IGRT, the clinical testing of the proposed method in the context of radiation therapy is a necessary procedure to move from theory into practice.

Comparative analysis of pathophysiological parameters between emphysematous smokers and emphysematous patients with COPD

Emphysematous smokers with normal spirometry form a considerable proportion of the clinical population. However, despite presenting with respiratory symptoms and activity limitation, they cannot be diagnosed with chronic obstructive lung disease (COPD) according to current criteria. Thus, we aimed to determine whether emphysema in smokers has a different pathogenesis from that in patients with COPD. We compared 12 pairs of lung tissue samples from emphysematous patients with normal spirometry and COPD, and determined the degree of emphysema using computed tomography. With a focus on COPD-related pathogenesis, we independently assessed inflammatory response, protease-antiprotease balance, oxidative stress, and apoptosis in both groups. Both groups showed similar pathological changes at a comparable degree of emphysema; the expression of inflammatory factors was comparable, with overexpression of proteases and decreased levels of antiproteases. Moreover, there was no significant difference in the activities of glutathione and superoxide dismutase, and expression of apoptosis-related factors. In conclusion, emphysema in smokers with normal spirometry and in patients with COPD had similar pathogenesis. Forced expiratory volume in 1 second cannot be used as the sole diagnostic criterion in patients with COPD; early intervention is of great importance to such patients.

Cardiac toxicity of lung cancer radiotherapy

Radical radiotherapy of lung cancer with dose escalation has been associated with increased tumor control. However, these attempts to continually improve local control through dose escalation, have met mixed results culminating in the findings of the RTOG trial 0617, where the heart dose was associated with a worse overall survival, indicating a significant contribution to radiation-induced cardiac morbidity. It is, therefore, very likely that poorly understood cardiac toxicity may have offset any potential improvement in overall survival derived from dose escalation and may be an obstacle that limits disease control and survival of patients. The manifestations of cardiac toxicity are relatively common after high dose radiotherapy of advanced lung cancers and are independently associated with both heart dose and baseline cardiac risk. Toxicity following the treatment may occur earlier than previously thought and, therefore, heart doses should be minimized. In patients with lung cancer, who not only receive substantial heart dose, but are also older with more comorbidities, all cardiac events have the potential to be clinically significant and life-threatening. Sophisticated radiation treatment planning techniques, charged particle therapy, and modern imaging methods in radiotherapy planning, may lead to reduction of the heart dose, which could potentially improve the clinical outcomes in patients with lung cancer. Efforts should be made to minimize heart radiation exposure whenever possible even at doses lower than those generally recommended. Heart doses should be limited as much as possible. A heart dosimetry as a whole is important for patient outcomes, rather than emphasizing just one parameter.

Progress in image-guided radiotherapy for the treatment of non-small cell lung cancer

Lung cancer is one of the most common malignant tumors. It has the highest incidence and mortality rate of all cancers worldwide. Late diagnosis of non-small cell lung cancer (NSCLC) is very common in clinical practice, and most patients miss the chance for radical surgery. Thus, radiotherapy plays an indispensable role in the treatment of NSCLC. Radiotherapy technology has evolved from the classic two-dimensional approach to three-dimensional conformal and intensity-modulated radiotherapy. However, how to ensure delivery of an accurate dose to the tumor while minimizing the irradiation of normal tissues remains a huge challenge for radiation oncologists, especially due to the positioning error between fractions and the autonomous movement of organs. In recent years, image-guided radiotherapy (IGRT) has greatly increased the accuracy of tumor irradiation while reducing the irradiation dose delivered to healthy tissues and organs. This paper presents a brief review of the definition of IGRT and the various technologies and applications of IGRT. IGRT can help ensure accurate dosing of the target area and reduce radiation damage to the surrounding normal tissue. IGRT may increase the local control rate of tumors and reduce the incidence of radio-therapeutic complications.

The Dosimetric Comparisons of CRT, IMRT, ARC, CRT+IMRT, and CRT+ARC of Postoperative Radiotherapy in IIIA-N2 Stage Non-Small-Cell Lung Cancer Patients

Currently, studies about PORT in stage IIIA-N2 NSCLC patients in recent years have mostly adopted the conformal radiation therapy (CRT) technique, while other modern techniques such as intensity modulated radiation therapy (IMRT), volumetric modulated arc therapy (VMAT, hereinafter referred to as ARC), helical tomotherapy (HT), and so forth are also developing quickly. In this paper, we intended to compare the dosimetric characteristics of CRT, IMRT, ARC, CRT+IMRT, and CRT+ARC of PORT in stage IIIA-N2 NSCLC patients. Ten patients with stage IIIA-N2 completely resected NSCLC, whom were treated by PORT in the radiotherapy department of our hospital from January 1, 2017, to January 1, 2018, were randomly selected in this study. For each patient, the CRT plan, IMRT plan, ARC plan, CRT+IMRT plan, and CRT+ARC plan were designed separately on the same set of CT images. The isodose distribution and dose-volume histogram (DVH) of the five plans were compared to determine the dosimetric parameters of the targets, OAR (organs at risk), and the normal tissue (defined as body subtracted to PTV (planning target volume), B-P). No plan had absolute dosimetry advantages than any other plans. In clinical practice, the plans could be chosen according to their dosimetry characteristics.

Automated ultrafast kilovoltage-megavoltage cone-beam CT for image guided radiotherapy of lung cancer: System description and real-time results

PURPOSE

To establish a fully automated kV-MV CBCT imaging method on a clinical linear accelerator that allows image acquisition of thoracic targets for patient positioning within one breath-hold (∼15s) under realistic clinical conditions.

METHODS AND MATERIALS

Our previously developed FPGA-based hardware unit which allows synchronized kV-MV CBCT projection acquisition is connected to a clinical linear accelerator system via a multi-pin switch; i.e. either kV-MV imaging or conventional clinical mode can be selected. An application program was developed to control the relevant linac parameters automatically and to manage the MV detector readout as well as the gantry angle capture for each MV projection. The kV projections are acquired with the conventional CBCT system. GPU-accelerated filtered backprojection is performed separately for both data sets. After appropriate grayscale normalization both modalities are combined and the final kV-MV volume is re-imported in the CBCT system to enable image matching. To demonstrate adequate geometrical accuracy of the novel imaging system the Penta-Guide phantom QA procedure is performed. Furthermore, a human plastinate and different tumor shapes in a thorax phantom are scanned. Diameters of the known tumor shapes are measured in the kV-MV reconstruction.

RESULTS

An automated kV-MV CBCT workflow was successfully established in a clinical environment. The overall procedure, from starting the data acquisition until the reconstructed volume is available for registration, requires ∼90s including 17s acquisition time for 100° rotation. It is very simple and allows target positioning in the same way as for conventional CBCT. Registration accuracy of the QA phantom is within ±1mm. The average deviation from the known tumor dimensions measured in the thorax phantom was 0.7mm which corresponds to an improvement of 36% compared to our previous kV-MV imaging system.

CONCLUSIONS

Due to automation the kV-MV CBCT workflow is speeded up by a factor of >10 compared to the manual approach. Thus, the system allows a simple, fast and reliable imaging procedure and fulfills all requirements to be successfully introduced into the clinical workflow now, enabling single-breath-hold volume imaging.

Phantom-based evaluation of dose exposure of ultrafast combined kV-MV-CBCT towards clinical implementation for IGRT of lung cancer

PURPOSE

Combined ultrafast 90°+90° kV-MV-CBCT within single breath-hold of 15s has high clinical potential for accelerating imaging for lung cancer patients treated with deep inspiration breath-hold (DIBH). For clinical feasibility of kV-MV-CBCT, dose exposure has to be small compared to prescribed dose. In this study, kV-MV dose output is evaluated and compared to clinically-established kV-CBCT.

METHODS

Accurate dose calibration was performed for kV and MV energy; beam quality was determined. For direct comparison of MV and kV dose output, relative biological effectiveness (RBE) was considered. CT dose index (CTDI) was determined and measurements in various representative locations of an inhomogeneous thorax phantom were performed to simulate the patient situation.

RESULTS

A measured dose of 20.5mGE (Gray-equivalent) in the target region was comparable to kV-CBCT (31.2mGy for widely-used, and 9.1mGy for latest available preset), whereas kV-MV spared healthy tissue and reduced dose to 6.6mGE (30%) due to asymmetric dose distribution. The measured weighted CTDI of 12mGE for kV-MV lay in between both clinical presets.

CONCLUSIONS

Dosimetric properties were in agreement with established imaging techniques, whereas exposure to healthy tissue was reduced. By reducing the imaging time to a single breath-hold of 15s, ultrafast combined kV-MV CBCT shortens patient time at the treatment couch and thus improves patient comfort. It is therefore usable for imaging of hypofractionated lung DIBH patients.

IGRT and motion management during lung SBRT delivery

Patient motion can cause misalignment of the tumour and toxicities to the healthy lung tissue during lung stereotactic body radiation therapy (SBRT). Any deviations from the reference setup can miss the target and have acute toxic effects on the patient with consequences onto its quality of life and survival outcomes. Correction for motion, either immediately prior to treatment or intra-treatment, can be realized with image-guided radiation therapy (IGRT) and motion management devices. The use of these techniques has demonstrated the feasibility of integrating complex technology with clinical linear accelerator to provide a higher standard of care for the patients and increase their quality of life.