Summary Lung cancer is the most frequent cause of cancer-related deaths worldwide. Every year, 1·8 million people are diagnosed with lung cancer, and 1·6 million people die as a result of the ...disease. 5-year survival rates vary from 4–17% depending on stage and regional differences. In this Seminar, we discuss existing treatment for patients with lung cancer and the promise of precision medicine, with special emphasis on new targeted therapies. Some subgroups, eg—patients with poor performance status and elderly patients—are not specifically addressed, because these groups require special treatment considerations and no frameworks have been established in terms of new targeted therapies. We discuss prevention and early detection of lung cancer with an emphasis on lung cancer screening. Although we acknowledge the importance of smoking prevention and cessation, this is a large topic beyond the scope of this Seminar.
This paper presents a review of deep learning (DL)-based medical image registration methods. We summarized the latest developments and applications of DL-based registration methods in the medical ...field. These methods were classified into seven categories according to their methods, functions and popularity. A detailed review of each category was presented, highlighting important contributions and identifying specific challenges. A short assessment was presented following the detailed review of each category to summarize its achievements and future potential. We provided a comprehensive comparison among DL-based methods for lung and brain registration using benchmark datasets. Lastly, we analyzed the statistics of all the cited works from various aspects, revealing the popularity and future trend of DL-based medical image registration.
Whole-breast irradiation after breast-conserving surgery for patients with early-stage breast cancer decreases ipsilateral breast-tumour recurrence (IBTR), yielding comparable results to mastectomy. ...It is unknown whether accelerated partial breast irradiation (APBI) to only the tumour-bearing quadrant, which shortens treatment duration, is equally effective. In our trial, we investigated whether APBI provides equivalent local tumour control after lumpectomy compared with whole-breast irradiation.
We did this randomised, phase 3, equivalence trial (NSABP B-39/RTOG 0413) in 154 clinical centres in the USA, Canada, Ireland, and Israel. Adult women (>18 years) with early-stage (0, I, or II; no evidence of distant metastases, but up to three axillary nodes could be positive) breast cancer (tumour size ≤3 cm; including all histologies and multifocal breast cancers), who had had lumpectomy with negative (ie, no detectable cancer cells) surgical margins, were randomly assigned (1:1) using a biased-coin-based minimisation algorithm to receive either whole-breast irradiation (whole-breast irradiation group) or APBI (APBI group). Whole-breast irradiation was delivered in 25 daily fractions of 50 Gy over 5 weeks, with or without a supplemental boost to the tumour bed, and APBI was delivered as 34 Gy of brachytherapy or 38·5 Gy of external bream radiation therapy in 10 fractions, over 5 treatment days within an 8-day period. Randomisation was stratified by disease stage, menopausal status, hormone-receptor status, and intention to receive chemotherapy. Patients, investigators, and statisticians could not be masked to treatment allocation. The primary outcome of invasive and non-invasive IBTR as a first recurrence was analysed in the intention-to-treat population, excluding those patients who were lost to follow-up, with an equivalency test on the basis of a 50% margin increase in the hazard ratio (90% CI for the observed HR between 0·667 and 1·5 for equivalence) and a Cox proportional hazard model. Survival was assessed by intention to treat, and sensitivity analyses were done in the per-protocol population. This trial is registered with ClinicalTrials.gov, NCT00103181.
Between March 21, 2005, and April 16, 2013, 4216 women were enrolled. 2109 were assigned to the whole-breast irradiation group and 2107 were assigned to the APBI group. 70 patients from the whole-breast irradiation group and 14 from the APBI group withdrew consent or were lost to follow-up at this stage, so 2039 and 2093 patients respectively were available for survival analysis. Further, three and four patients respectively were lost to clinical follow-up (ie, survival status was assessed by phone but no physical examination was done), leaving 2036 patients in the whole-breast irradiation group and 2089 in the APBI group evaluable for the primary outcome. At a median follow-up of 10·2 years (IQR 7·5–11·5), 90 (4%) of 2089 women eligible for the primary outcome in the APBI group and 71 (3%) of 2036 women in the whole-breast irradiation group had an IBTR (HR 1·22, 90% CI 0·94–1·58). The 10-year cumulative incidence of IBTR was 4·6% (95% CI 3·7–5·7) in the APBI group versus 3·9% (3·1–5·0) in the whole-breast irradiation group. 44 (2%) of 2039 patients in the whole-breast irradiation group and 49 (2%) of 2093 patients in the APBI group died from recurring breast cancer. There were no treatment-related deaths. Second cancers and treatment-related toxicities were similar between the two groups. 2020 patients in the whole-breast irradiation group and 2089 in APBI group had available data on adverse events. The highest toxicity grade reported was: grade 1 in 845 (40%), grade 2 in 921 (44%), and grade 3 in 201 (10%) patients in the APBI group, compared with grade 1 in 626 (31%), grade 2 in 1193 (59%), and grade 3 in 143 (7%) in the whole-breast irradiation group.
APBI did not meet the criteria for equivalence to whole-breast irradiation in controlling IBTR for breast-conserving therapy. Our trial had broad eligibility criteria, leading to a large, heterogeneous pool of patients and sufficient power to detect treatment equivalence, but was not designed to test equivalence in patient subgroups or outcomes from different APBI techniques. For patients with early-stage breast cancer, our findings support whole-breast irradiation following lumpectomy; however, with an absolute difference of less than 1% in the 10-year cumulative incidence of IBTR, APBI might be an acceptable alternative for some women.
National Cancer Institute, US Department of Health and Human Services.
Radiotherapy with concomitant and adjuvant temozolomide is the standard of care for newly diagnosed glioblastoma (GBM). O(6)-methylguanine-DNA methyltransferase (MGMT) methylation status may be an ...important determinant of treatment response. Dose-dense (DD) temozolomide results in prolonged depletion of MGMT in blood mononuclear cells and possibly in tumor. This trial tested whether DD temozolomide improves overall survival (OS) or progression-free survival (PFS) in patients with newly diagnosed GBM.
This phase III trial enrolled patients older than age 18 years with a Karnofsky performance score of ≥ 60 with adequate tissue. Stratification included clinical factors and tumor MGMT methylation status. Patients were randomly assigned to standard temozolomide (arm 1) or DD temozolomide (arm 2) for 6 to 12 cycles. The primary end point was OS. Secondary analyses evaluated the impact of MGMT status.
A total of 833 patients were randomly assigned to either arm 1 or arm 2 (1,173 registered). No statistically significant difference was observed between arms for median OS (16.6 v 14.9 months, respectively; hazard ratio HR, 1.03; P = .63) or median PFS (5.5 v 6.7 months; HR, 0.87; P = .06). Efficacy did not differ by methylation status. MGMT methylation was associated with improved OS (21.2 v 14 months; HR, 1.74; P < .001), PFS (8.7 v 5.7 months; HR, 1.63; P < .001), and response (P = .012). There was increased grade ≥ 3 toxicity in arm 2 (34% v 53%; P < .001), mostly lymphopenia and fatigue.
This study did not demonstrate improved efficacy for DD temozolomide for newly diagnosed GBM, regardless of methylation status. However, it did confirm the prognostic significance of MGMT methylation. Feasibility of large-scale accrual, prospective tumor collection, and molecular stratification was demonstrated.
Purpose
Accurate and timely organs‐at‐risk (OARs) segmentation is key to efficient and high‐quality radiation therapy planning. The purpose of this work is to develop a deep learning‐based method to ...automatically segment multiple thoracic OARs on chest computed tomography (CT) for radiotherapy treatment planning.
Methods
We propose an adversarial training strategy to train deep neural networks for the segmentation of multiple organs on thoracic CT images. The proposed design of adversarial networks, called U‐Net‐generative adversarial network (U‐Net‐GAN), jointly trains a set of U‐Nets as generators and fully convolutional networks (FCNs) as discriminators. Specifically, the generator, composed of U‐Net, produces an image segmentation map of multiple organs by an end‐to‐end mapping learned from CT image to multiorgan‐segmented OARs. The discriminator, structured as an FCN, discriminates between the ground truth and segmented OARs produced by the generator. The generator and discriminator compete against each other in an adversarial learning process to produce the optimal segmentation map of multiple organs. Our segmentation results were compared with manually segmented OARs (ground truth) for quantitative evaluations in geometric difference, as well as dosimetric performance by investigating the dose‐volume histogram in 20 stereotactic body radiation therapy (SBRT) lung plans.
Results
This segmentation technique was applied to delineate the left and right lungs, spinal cord, esophagus, and heart using 35 patients’ chest CTs. The averaged dice similarity coefficient for the above five OARs are 0.97, 0.97, 0.90, 0.75, and 0.87, respectively. The mean surface distance of the five OARs obtained with proposed method ranges between 0.4 and 1.5 mm on average among all 35 patients. The mean dose differences on the 20 SBRT lung plans are −0.001 to 0.155 Gy for the five OARs.
Conclusion
We have investigated a novel deep learning‐based approach with a GAN strategy to segment multiple OARs in the thorax using chest CT images and demonstrated its feasibility and reliability. This is a potentially valuable method for improving the efficiency of chest radiotherapy treatment planning.
Purpose
The incorporation of cone‐beam computed tomography (CBCT) has allowed for enhanced image‐guided radiation therapy. While CBCT allows for daily 3D imaging, images suffer from severe artifacts, ...limiting the clinical potential of CBCT. In this work, a deep learning‐based method for generating high quality corrected CBCT (CCBCT) images is proposed.
Methods
The proposed method integrates a residual block concept into a cycle‐consistent adversarial network (cycle‐GAN) framework, called res‐cycle GAN, to learn a mapping between CBCT images and paired planning CT images. Compared with a GAN, a cycle‐GAN includes an inverse transformation from CBCT to CT images, which constrains the model by forcing calculation of both a CCBCT and a synthetic CBCT. A fully convolution neural network with residual blocks is used in the generator to enable end‐to‐end CBCT‐to‐CT transformations. The proposed algorithm was evaluated using 24 sets of patient data in the brain and 20 sets of patient data in the pelvis. The mean absolute error (MAE), peak signal‐to‐noise ratio (PSNR), normalized cross‐correlation (NCC) indices, and spatial non‐uniformity (SNU) were used to quantify the correction accuracy of the proposed algorithm. The proposed method is compared to both a conventional scatter correction and another machine learning‐based CBCT correction method.
Results
Overall, the MAE, PSNR, NCC, and SNU were 13.0 HU, 37.5 dB, 0.99, and 0.05 in the brain, 16.1 HU, 30.7 dB, 0.98, and 0.09 in the pelvis for the proposed method, improvements of 45%, 16%, 1%, and 93% in the brain, and 71%, 38%, 2%, and 65% in the pelvis, over the CBCT image. The proposed method showed superior image quality as compared to the scatter correction method, reducing noise and artifact severity. The proposed method produced images with less noise and artifacts than the comparison machine learning‐based method.
Conclusions
The authors have developed a novel deep learning‐based method to generate high‐quality corrected CBCT images. The proposed method increases onboard CBCT image quality, making it comparable to that of the planning CT. With further evaluation and clinical implementation, this method could lead to quantitative adaptive radiation therapy.
Summary Background We aimed to compare overall survival after standard-dose versus high-dose conformal radiotherapy with concurrent chemotherapy and the addition of cetuximab to concurrent ...chemoradiation for patients with inoperable stage III non-small-cell lung cancer. Methods In this open-label randomised, two-by-two factorial phase 3 study in 185 institutions in the USA and Canada, we enrolled patients (aged ≥18 years) with unresectable stage III non-small-cell lung cancer, a Zubrod performance status of 0–1, adequate pulmonary function, and no evidence of supraclavicular or contralateral hilar adenopathy. We randomly assigned (1:1:1:1) patients to receive either 60 Gy (standard dose), 74 Gy (high dose), 60 Gy plus cetuximab, or 74 Gy plus cetuximab. All patients also received concurrent chemotherapy with 45 mg/m2 paclitaxel and carboplatin once a week (AUC 2); 2 weeks after chemoradiation, two cycles of consolidation chemotherapy separated by 3 weeks were given consisting of paclitaxel (200 mg/m2 ) and carboplatin (AUC 6). Randomisation was done with permuted block randomisation methods, stratified by radiotherapy technique, Zubrod performance status, use of PET during staging, and histology; treatment group assignments were not masked. Radiation dose was prescribed to the planning target volume and was given in 2 Gy daily fractions with either intensity-modulated radiation therapy or three-dimensional conformal radiation therapy. The use of four-dimensional CT and image-guided radiation therapy were encouraged but not necessary. For patients assigned to receive cetuximab, 400 mg/m2 cetuximab was given on day 1 followed by weekly doses of 250 mg/m2 , and was continued through consolidation therapy. The primary endpoint was overall survival. All analyses were done by modified intention-to-treat. The study is registered with ClinicalTrials.gov , number NCT00533949. Findings Between Nov 27, 2007, and Nov 22, 2011, 166 patients were randomly assigned to receive standard-dose chemoradiotherapy, 121 to high-dose chemoradiotherapy, 147 to standard-dose chemoradiotherapy and cetuximab, and 110 to high-dose chemoradiotherapy and cetuximab. Median follow-up for the radiotherapy comparison was 22·9 months (IQR 27·5–33·3). Median overall survival was 28·7 months (95% CI 24·1–36·9) for patients who received standard-dose radiotherapy and 20·3 months (17·7–25·0) for those who received high-dose radiotherapy (hazard ratio HR 1·38, 95% CI 1·09–1·76; p=0·004). Median follow-up for the cetuximab comparison was 21·3 months (IQR 23·5–29·8). Median overall survival in patients who received cetuximab was 25·0 months (95% CI 20·2–30·5) compared with 24·0 months (19·8–28·6) in those who did not (HR 1·07, 95% CI 0·84–1·35; p=0·29). Both the radiation-dose and cetuximab results crossed protocol-specified futility boundaries. We recorded no statistical differences in grade 3 or worse toxic effects between radiotherapy groups. By contrast, the use of cetuximab was associated with a higher rate of grade 3 or worse toxic effects (205 86% of 237 vs 160 70% of 228 patients; p<0·0001). There were more treatment-related deaths in the high-dose chemoradiotherapy and cetuximab groups (radiotherapy comparison: eight vs three patients; cetuximab comparison: ten vs five patients). There were no differences in severe pulmonary events between treatment groups. Severe oesophagitis was more common in patients who received high-dose chemoradiotherapy than in those who received standard-dose treatment (43 21% of 207 patients vs 16 7% of 217 patients; p<0·0001). Interpretation 74 Gy radiation given in 2 Gy fractions with concurrent chemotherapy was not better than 60 Gy plus concurrent chemotherapy for patients with stage III non-small-cell lung cancer, and might be potentially harmful. Addition of cetuximab to concurrent chemoradiation and consolidation treatment provided no benefit in overall survival for these patients. Funding National Cancer Institute and Bristol-Myers Squibb.
Purpose
Automated synthetic computed tomography (sCT) generation based on magnetic resonance imaging (MRI) images would allow for MRI‐only based treatment planning in radiation therapy, eliminating ...the need for CT simulation and simplifying the patient treatment workflow. In this work, the authors propose a novel method for generation of sCT based on dense cycle‐consistent generative adversarial networks (cycle GAN), a deep‐learning based model that trains two transformation mappings (MRI to CT and CT to MRI) simultaneously.
Methods and materials
The cycle GAN‐based model was developed to generate sCT images in a patch‐based framework. Cycle GAN was applied to this problem because it includes an inverse transformation from CT to MRI, which helps constrain the model to learn a one‐to‐one mapping. Dense block‐based networks were used to construct generator of cycle GAN. The network weights and variables were optimized via a gradient difference (GD) loss and a novel distance loss metric between sCT and original CT.
Results
Leave‐one‐out cross‐validation was performed to validate the proposed model. The mean absolute error (MAE), peak signal‐to‐noise ratio (PSNR), and normalized cross correlation (NCC) indexes were used to quantify the differences between the sCT and original planning CT images. For the proposed method, the mean MAE between sCT and CT were 55.7 Hounsfield units (HU) for 24 brain cancer patients and 50.8 HU for 20 prostate cancer patients. The mean PSNR and NCC were 26.6 dB and 0.963 in the brain cases, and 24.5 dB and 0.929 in the pelvis.
Conclusion
We developed and validated a novel learning‐based approach to generate CT images from routine MRIs based on dense cycle GAN model to effectively capture the relationship between the CT and MRIs. The proposed method can generate robust, high‐quality sCT in minutes. The proposed method offers strong potential for supporting near real‐time MRI‐only treatment planning in the brain and pelvis.
This paper reviewed the deep learning‐based studies for medical imaging synthesis and its clinical application. Specifically, we summarized the recent developments of deep learning‐based methods in ...inter‐ and intra‐modality image synthesis by listing and highlighting the proposed methods, study designs, and reported performances with related clinical applications on representative studies. The challenges among the reviewed studies were then summarized with discussion.
•Comprehensive review of deep learning-based multi-organ segmentation.•Categorization of pixel-wise classification and end-to-end segmentation.•Pixel-wise classification includes AE and ...CNN.•End-to-end segmentation includes FCN, R-FCN, GAN and synthetic image-aided.•Benchmark of algorithms’ performances for thoracic and head-neck CT segmentation.
Deep learning has revolutionized image processing and achieved the-state-of-art performance in many medical image segmentation tasks. Many deep learning-based methods have been published to segment different parts of the body for different medical applications. It is necessary to summarize the current state of development for deep learning in the field of medical image segmentation. In this paper, we aim to provide a comprehensive review with a focus on multi-organ image segmentation, which is crucial for radiotherapy where the tumor and organs-at-risk need to be contoured for treatment planning. We grouped the surveyed methods into two broad categories which are ‘pixel-wise classification’ and ‘end-to-end segmentation’. Each category was divided into subgroups according to their network design. For each type, we listed the surveyed works, highlighted important contributions and identified specific challenges. Following the detailed review, we discussed the achievements, shortcomings and future potentials of each category. To enable direct comparison, we listed the performance of the surveyed works that used thoracic and head-and-neck benchmark datasets.