PGC-PTSD

The impact of respiratory motion and CT pitch on the robustness of radiomics feature extraction in 4DCT lung imaging

Orthogonally constrained matrix factorization for robust unsupervised feature selection with local preserving

PurposeThis study aimed to select robust features against lung motion in a phantom study and use them as input to feature selection algorithms and machine learning classifiers in a clinical study to predict the lymphovascular invasion (LVI) of non-small cell lung cancer (NSCLC). The results of robust features were also compared with conventional techniques without considering the robustness of radiomic features.MethodsAn in-house developed lung phantom was developed with two 22mm lesion sizes based on a clinical study. A specific motor was built to simulate motion in two orthogonal directions. Lesions of both clinical and phantom studies were segmented using a Fuzzy C-means-based segmentation algorithm. After inducing motion and extracting 105 radiomic features in 4 feature sets, including shape, first-, second-, and higher-order statistics features from each region of interest (ROI) of the phantom image, statistical analyses were performed to select robust features against motion. Subsequently, these robust features and a total of 105 radiomic features were extracted from 126 clinical data. Various feature selection (FS) and multiple machine learning (ML) classifiers were implemented to predict the LVI of NSCLC, followed by comparing the results of predicting LVI using robust features with common conventional techniques not considering the robustness of radiomic features.ResultsOur results demonstrated that selecting robust features as input to FS algorithms and ML classifiers surges the sensitivity, which has a gentle negative effect on the accuracy and the area under the curve (AUC) of predictions compared with commonly used methods in 12 of 15 outcomes. The top performance of the LVI prediction was achieved by the NB classifier and RFE FS without considering the robustness of radiomic features with 95% area under the curve of AUC, 67% accuracy, and 100% sensitivity. Moreover, the top performance of the LVI prediction using robust features belonged to the NB classifier and Boruta feature selection with 92% AUC, 86% accuracy, and 100% sensitivity.ConclusionRobustness over various influential factors is critical and should be considered in a radiomic study. Selecting robust features is a solution to overcome the low reproducibility of radiomic features. Although setting robust features against motion in a phantom study has a minor negative impact on the accuracy and AUC of LVI prediction, it boosts the sensitivity of prediction to a large extent.

Robust feature selection using label enhancement and β-precision fuzzy rough sets for multilabel fuzzy decision system

Individualized altered fractionation as a more effective radiotherapy for non-small cell lung cancer.

: Pulmonary resection remains the accepted standard of care for early stage non-small cell lung cancer (NSCLC) in patients with adequate cardiopulmonary reserve. Radiation therapy, and in particular stereotactic body radiation therapy (SBRT), has traditionally been recommended as curative therapy for medically unfit patients who harbor early stage lung cancer. As a consequence of the rising success of SBRT in patients with medically inoperable early stage NSCLC (evidenced by Radiation Therapy Oncology Group trials 0236 and 0915), there has been keen interest in exploring this modality in operable patients. SBRT to lung is often very well tolerated, thus making it very attractive to patients. The risk of series complications in carefully selected patients is low. Equipoise among surgeons and the medical community for the use of SBRT as curative therapy for patients eligible for surgery has been limited, thus contributing to the poor accrual and early closure of 3 randomized trials attempting to directly compare SBRT to surgery in operable patients. These trials included the STARS trial (Study of Anastrozole and Radiotherapy Sequencing), the ROSEL trial [Either Surgery or Stereotactic Radiotherapy for Early-Stage (IA) Lung Cancer] and the American College of Surgeons Oncology Group (ACOSOG) Z4099 trial. No preliminary data is available from the ACOSOG trial. The STARS and ROSEL trial randomized operable patients with tumors <4 cm (cT1–T2a) and required invasive mediastinal staging in patients harboring radiographically suspicious lymph nodes. Sublobar resection was not permitted in these trials. Since patient accrual was limited (n=58 total patients) the two trials were combined in a controversial publication that showed improved overall survival and decreased toxicity with SBRT compared to surgery. This has spawned additional randomized trials including the STABLE-MATES trial comparing sublobar resection and SBRT in patients with stage I NSCLC (NCT02468024) with an estimated enrollment of 258 patients, a Veterans Affairs Lung Cancer or Stereotactic Radiotherapy (VALOR) trial (NCT02984761) with estimated enrollment of 670 patients, and lastly, a phase II United Kingdom sponsored SABRTooth study (NCT02629458) that is enrolling up to 54 patients. The lack of randomized data has stimulated multiple publications that retrospectively examine these modalities in single center case series and more recently in larger data sets derived from the National Cancer Database (NCDB), the Surveillance, Epidemiology, and End Results (SEER)-Medicare linked database and the United States Veterans Affairs Health Care System. These retrospective analyses are clearly limited by patient selection bias and different levels of staging accuracy. Propensity matched retrospective series showed that overall survival was no different at 1 year but favored surgery at 3 years. Cancer-specific survival (clearly a better outcome measure particularly in elderly patients receiving SBRT) was no different compared to surgery, nor was local failure, or the risk of developing distant metastases. Collectively, the studies that used large data with statistical matching demonstrated a survival advantage of surgery (namely lobectomy) over SBRT for stage I NSCLC but raised awareness of the increased post procedure morbidity and mortality for surgery compared to SBRT. Interestingly, sublobar resection versus SBRT in these large data sets had mixed results with some showing survival improvement with surgery and some showing no survival difference. Surgery appears superior at later time points. In a phase II ablate and resect human clinical trial (MISSILE-NSCLC, NCT02136355) for early stage (T1–T2) NSCLC, 35 patients underwent neoadjuvant SBRT using 54–60 Gy in 3–8 fractions followed by surgery at an interval of 10 weeks. The authors reported a 60% pathologic complete response rate, invoking many questions about local control of SBRT. Assessment of local control after SBRT remains the single most challenging real-world clinical problem. The lung injury post ablation is heterogenous and requires very careful observation with imaging modalities that lack sensitivity and specificity for recurrence. At present, the American Society of Radiation Oncology does not recommend SBRT for operable patients with early stage NSCLC outside of a clinical trial.

High-dimensional data expands the spatial dimension, leading to increased computational complexity and reduced generalization performance. Microarray data classification, such as diagnosing diseases like cancer, involves complex dimensions due to their genetic and biological information. To address this issue, dimension reduction is essential for these data sets. The main goal of this chapter is to provide a method for dimension reduction and classification of genetic data sets. The proposed approach comprises multiple stages. Initially, various feature ranking methods are combined to improve the robustness and stability of the feature selection process. A hybrid ranking method, which incorporates gene interactions, is integrated with a wrapper method. Subsequently, a support vector machine (SVM) is employed for classification. To address class imbalance in the training data, a solution is implemented before feeding the data into the SVM classifier. The experimental outcomes of the proposed approach, tested on five microarray databases, indicate robust feature selection with a metric ranging from 0.70 to 0.88. Additionally, the classification accuracy falls within the range of 91-96%.

8538 Background: Patients undergoing surgery for early-stage non-small cell lung cancer (NSCLC) may be at high-risk for post-operative mortality. Access to stereotactic body radiation therapy (SBRT) offers a less invasive alternative for this population that may facilitate more appropriate patient selection for surgery. Methods: An analysis of all patients with early-stage NSCLC reported to the National Cancer Database between 2004-2015 was performed. Post-operative mortality rates were derived using vital status data. Utilization of SBRT was defined by each facility’s SBRT Experience in years and SBRT-to-Surgery volume ratios, defined by quartiles. Multivariable logistic regression with backward elimination was used to test for independence of associations between exposures of interest and post-operative mortality. Interaction testing was performed to assess the statistical relationship of covariates found to have independent associations. Results: The study cohort consisted of 202,542 patients who underwent surgical resection of clinical stage T1-T2 NSCLC (AJCC 7th edition). The 90-day post-operative mortality rate declined significantly during the study period from 4.6% to 2.6% (p &lt; 0.001). During this period, the proportion of facilities that utilized SBRT increased from 3.3% to 77.5% (p &lt; 0.001) and the proportion of patients treated with SBRT increased significantly from 0.7% to 15.4% (p &lt; 0.001). Lower 90-day post-operative mortality rates were observed at facilities with greater than six years of SBRT experience (OR 0.84, CI 0.76-0.94, p = 0.003) and SBRT-to-Surgery volume ratios above 17% (OR 0.85, CI 0.79-0.92, p &lt; 0.001). Additional covariates associated with 90-day mortality included higher surgical volume, geographic region, year of diagnosis, age, sex, race, insurance status, facility type, Charlson-Deyo score, clinical T stage, histology, anatomic location, surgery type, and prior malignancy. Interaction testing between these covariates was negative, demonstrating that higher SBRT Experience and SBRT-to-Surgery volume ratios were independently associated with lower 90-day surgical mortality. Conclusions: Patients who underwent surgery for early-stage NSCLC at facilities with higher SBRT Experience and SBRT-to-Surgery volume ratios had lower rates of post-operative mortality. These findings suggest that the availability of SBRT may be a surrogate for a more comprehensive and safer approach to matching patients to surgery or SBRT. The observation of higher post-operative mortality rates at facilities without an SBRT program deserves further study.

Previous chapter Next chapter Full AccessProceedings Proceedings of the 2016 SIAM International Conference on Data Mining (SDM)Robust Unsupervised Feature Selection on Networked DataJundong Li, Xia Hu, Liang Wu, and Huan LiuJundong Li, Xia Hu, Liang Wu, and Huan Liupp.387 - 395Chapter DOI:https://doi.org/10.1137/1.9781611974348.44PDFBibTexSections ToolsAdd to favoritesExport CitationTrack CitationsEmail SectionsAboutAbstract Feature selection has shown its effectiveness to prepare high-dimensional data for many data mining and machine learning tasks. Traditional feature selection algorithms are mainly based on the assumption that data instances are independent and identically distributed. However, this assumption is invalid in networked data since instances are not only associated with high dimensional features but also inherently interconnected with each other. In addition, obtaining label information for networked data is time consuming and labor intensive. Without label information to direct feature selection, it is difficult to assess the feature relevance. In contrast to the scarce label information, link information in networks are abundant and could help select relevant features. However, most networked data has a lot of noisy links, resulting in the feature selection algorithms to be less effective. To address the above mentioned issues, we propose a robust unsupervised feature selection framework NetFS for networked data, which embeds the latent representation learning into feature selection. Therefore, content information is able to help mitigate the negative effects from noisy links in learning latent representations, while good latent representations in turn can contribute to extract more meaningful features. In other words, both phases could cooperate and boost each other. Experimental results on real-world datasets demonstrate the effectiveness of the proposed framework. Previous chapter Next chapter RelatedDetails Published:2016eISBN:978-1-61197-434-8 https://doi.org/10.1137/1.9781611974348Book Series Name:ProceedingsBook Code:PRDT16Book Pages:1-867

Consecutive Daily vs. Every Other Day SBRT Scheduling for Stage I NSCLC

Robust microarray data feature selection using a correntropy based distance metric learning approach

Cost-Utility Analysis of Stereotactic Body Radiation Therapy Versus Surgery for Patients With Stage I Non-small Cell Lung Cancer in Japan.

e21036 Background: The oligometastatic state is one of limited metastases, amenable to local therapies of curative potential. Stereotactic body radiation therapy (SBRT) is a non-surgical treatment with a primary goal of rapid local tumor control and low toxicity allowing prompt return to systemic therapy. Breast (BRC), colorectal (CRC) and non-small cell lung cancers (NSCLC) have oligometastases as first presentation. In these patients, consolidative SBRT offers effective local control and low toxicity, with the potential for improved survival benefit without affecting quality of life. We report our experience on such patients. Methods: 31 patients with BRC (9), CRC (16) and NSCLC (6) with ≤ 5 unresectable metastases to one to three organ sites were identified. Data were obtained from chart review. Primary end point was infield local control. Lesions with at least 6 months of radiographic follow-up were assessable. Secondary end points were toxicity and survival. Results: All patients completed SBRT to 38 lesions (≤2/patient). 29 received prior chemotherapy for metastatic disease. 14 were treated with at least 1 prior regimen and 15 at least 2 (0-8). At a median follow-up of 16 months (4-32), 22 patients (8 BRC, 10 CRC and 4 NSCLC) were assessable. At 6 months, 20 patients had infield local control and 3 demonstrated distant metastases. At 1 year, this was seen in 19 and 7 patients respectively. The most common toxicity was grade ≤2 fatigue. All BRC had infield local control at 6 months and 1 year. Among CRC, 9 had infield local control at 6 months and 7 at 1 year. 3 NSCLC patients had infield local control at 6 months and 1 year. No BRC demonstrated distant progression at 6 months; at 1 year, it was seen in 2 patients. 3 CRC had evidence of distant progression at 6 months and 1 year. 1 NSCLC patient had distant progression at 6 months and 2 at 1 year. 6 patients died from their underlying malignancy; the median survival for 5 of these patients was 20 months (5-34). Conclusions: The combination of systemic chemotherapy and consolidative SBRT is safe and effective for patients with oligometastatic disease

BackgroundLung resections and stereotactic body radiation therapy (SBRT) are primary treatments for clinical stage I non-small cell lung cancer (NSCLC). The aim of this network meta-analysis is to compare the differences in effectiveness between SBRT and specific lung resections, including lobectomy, segmentectomy, and wedge resection for clinical stage I NSCLC.MethodsPubMed, EMBASE, Cochrane Library, and the ClinicalTrials.gov registry were searched. Random-effects model was conducted to assess differences in survival outcomes and treatment-related complication incidence between SBRT and three specific lung resections (lobectomy, segmentectomy, wedge resection). Subgroup analyses were performed according to clinical stages, regions, and publication years.ResultsA total of 30 studies were enrolled for network meta-analysis. All three lung resections demonstrated superior overall survival (OS) compared with SBRT [lobectomy: hazard ratio (HR) 0.65, 95% confidence interval: 0.53–0.79; segmentectomy: HR 0.64, 95% confidence interval: 0.50–0.82); wedge resection: HR 0.72, 95% confidence interval: 0.55–0.93]. The recurrence-free survival (RFS) of patients in lobectomy and segmentectomy groups was significantly better than that of patients in the SBRT group, with no significant difference between the wedge resection and SBRT groups. Patients with clinical stage IA NSCLC showed no significant difference in OS between SBRT and three lung resections (lobectomy: HR 0.99, 95% confidence interval: 0.85–1.16; segmentectomy: HR 0.98, 95% confidence interval: 0.83–1.16; wedge resection: HR 1.71, 95% confidence interval: 0.91–3.26).ConclusionsFor clinical stage I NSCLC patients, lobectomy and segmentectomy are superior to SBRT. Wedge resection is associated with similar RFS but better OS to SBRT. In clinical stage IA NSCLC, SBRT may provide comparable OS to lung resections.

Comparison of Long-Term Survival Outcomes Between Stereotactic Body Radiation Therapy and Sublobar Resection for Stage I Non-Small Cell Lung Cancer in High-Risk Operable Patients: Propensity Score Match Analysis