A distributed feature selection pipeline for survival analysis using radiomics in non-small cell lung cancer patients.

Predictive modelling of cancer outcomes using radiomics faces dimensionality problems and data limitations, as radiomics features often number in the hundreds, and multi-institutional data sharing is ()often unfeasible. Federated learning (FL) and feature selection (FS) techniques combined can help overcome these issues, as one provides the means of training models without exchanging sensitive data, while the other identifies the most informative features, reduces overfitting, and improves model interpretability. Our proposed FS pipeline based on FL principles targets data-driven radiomics FS in a multivariate survival study of non-small cell lung cancer patients. The pipeline was run across datasets from three institutions without patient-level data exchange. It includes two FS techniques, Correlation-based Feature Selection and LASSO regularization, and Cox Proportional-Hazard regression with Overall Survival as endpoint. Trained and validated on 828 patients overall, our pipeline yielded a radiomic signature comprising "intensity-based energy" and "mean discretised intensity". Validation resulted in a mean Harrell C-index of 0.59, showcasing fair efficacy in risk stratification. In conclusion, we suggest a distributed radiomics approach that incorporates preliminary feature selection to systematically decrease the feature set based on data-driven considerations. This aims to address dimensionality challenges beyond those associated with data constraints and interpretability concerns.

Epidemiology, time course, and risk factors for hospital-acquired bloodstream infections in a cohort of 14,884 patients before and during the COVID-19 pandemic.

OBJECTIVE: COVID-19 pandemic has changed in-hospital care and was linked to superimposed infections. Here, we described epidemiology and risk factors for hospital-acquired bloodstream infections (HA-BSIs), before and during COVID-19 pandemic. METHODS: This retrospective, observational, single-center real-life study included 14,884 patients admitted to hospital wards and intensive care units (ICUs) with at least one blood culture, drawn 48 h after admission, either before (pre-COVID, N = 7382) or during pandemic (N = 7502, 1203 COVID-19+ and 6299 COVID-19-). RESULTS: Two thousand two hundred and forty-five HA-BSI were microbiologically confirmed in 14,884 patients (15.1%), significantly higher among COVID-19+ (22.9%; ptrend < .001). COVID-19+ disclosed a significantly higher mortality rate (33.8%; p < .001) and more ICU admissions (29.7%; p < .001). Independent HAI-BSI predictors were: COVID-19 (OR: 1.43, 95%CI: 1.21-1.69; p < .001), hospitalization length (OR: 1.04, 95%CI: 1.03-1.04; p < .001), ICU admission (OR: 1.38, 95%CI: 1.19-1.60; p < .001), neoplasms (OR:1.48, 95%CI: 1.34-1.65; p < .001) and kidney failure (OR: 1.81, 95%CI: 1.61-2.04; p < .001). Of note, HA-BSI IRs for Acinetobacter spp. (0.16 × 100 patient-days) and Staphylococcus aureus (0.24 × 100 patient-days) peaked during the interval between first and second pandemic waves in our National context. CONCLUSIONS: Patients with HA-BSI admitted before and during pandemic substantially differed. COVID-19 represented a risk factor for HA-BSI, though not confirmed in the sole pandemic period. Some etiologies emerged between pandemic waves, suggesting potential COVID-19 long-term effect on HA-BSIs.

Fib-4 score is able to predict intra-hospital mortality in 4 different SARS-COV2 waves.

Increased values of the FIB-4 index appear to be associated with poor clinical outcomes in COVID-19 patients. This study aimed to develop and validate predictive mortality models, using data upon admission of hospitalized patients in four COVID-19 waves between March 2020 and January 2022. A single-center cohort study was performed on consecutive adult patients with Covid-19 admitted at the Fondazione Policlinico Gemelli IRCCS (Rome, Italy). Artificial intelligence and big data processing were used to retrieve data. Patients and clinical characteristics of patients with available FIB-4 data derived from the Gemelli Generator Real World Data (G2 RWD) were used to develop predictive mortality models during the four waves of the COVID-19 pandemic. A logistic regression model was applied to the training and test set (75%:25%). The model's performance was assessed by receiver operating characteristic (ROC) curves. A total of 4936 patients were included. Hypertension (38.4%), cancer (12.15%) and diabetes (16.3%) were the most common comorbidities. 23.9% of patients were admitted to ICU, and 12.6% had mechanical ventilation. During the study period, 762 patients (15.4%) died. We developed a multivariable logistic regression model on patient data from all waves, which showed that the FIB-4 score > 2.53 was associated with increased mortality risk (OR = 4.53, 95% CI 2.83-7.25; p ≤ 0.001). These data may be useful in the risk stratification at the admission of hospitalized patients with COVID-19.

A process mining approach for clinical guidelines compliance: real-world application in rectal cancer.

In the era of evidence-based medicine, several clinical guidelines were developed, supporting cancer management from diagnosis to treatment and aiming to optimize patient care and hospital resources. Nevertheless, individual patient characteristics and organizational factors may lead to deviations from these standard recommendations during clinical practice. In this context, process mining in healthcare constitutes a valid tool to evaluate conformance of real treatment pathways, extracted from hospital data warehouses as event log, to standard clinical guidelines, translated into computer-interpretable formats. In this study we translate the European Society of Medical Oncology guidelines for rectal cancer treatment into a computer-interpretable format using Pseudo-Workflow formalism (PWF), a language already employed in pMineR software library for Process Mining in Healthcare. We investigate the adherence of a real-world cohort of rectal cancer patients treated at Fondazione Policlinico Universitario A. Gemelli IRCCS, data associated with cancer diagnosis and treatment are extracted from hospital databases in 453 patients diagnosed with rectal cancer. PWF enables the easy implementation of guidelines in a computer-interpretable format and visualizations that can improve understandability and interpretability of physicians. Results of the conformance checking analysis on our cohort identify a subgroup of patients receiving a long course treatment that deviates from guidelines due to a moderate increase in radiotherapy dose and an addition of oxaliplatin during chemotherapy treatment. This study demonstrates the importance of PWF to evaluate clinical guidelines adherence and to identify reasons of deviations during a treatment process in a real-world and multidisciplinary setting.

pCR and 2-Year Disease-Free Survival: A Combination of the Two Endpoints as a New Classification for Locally Advanced Rectal Cancer Patients-An Updated Pooled Analysis of Eleven International Randomized Trials.

LARC is managed by multimodal treatments whose intensity can be highly modulated. In this context, we need surrogate endpoints to help predict long-term outcomes and better personalize treatments. A previous study identified 2yDFS as a stronger predictor of OS than pCR in LARC patients undergoing neoadjuvant RT. The aim of this pooled analysis was to assess the role of pCR and 2yDFS as surrogate endpoints for OS in a larger cohort. The pooled and subgroup analyses were performed on large rectal cancer randomized trial cohorts who received long-course RT. Our analysis focused on the evaluation of OS in relation to the pCR and 2-year disease status. A total of 4600 patients were analyzed. Four groups were identified according to intermediate outcomes: 12% had both pCR and 2yDFS (the better); 67% achieved 2yDFS but not pCR (the good); 1% had pCR but not 2yDFS; and 20% had neither pCR nor 2yDFS (the bad). The pCR and 2yDFS were favorably associated with OS in the univariate analysis, and 2yDFS maintained a statistically significant association in the multivariate analysis independently of the pCR status. The combination of the pCR and 2yDFS results in a strong predictor of OS, whereas failure to achieve 2yDFS carries a poor prognosis regardless of the pCR status. This new stratification of LARC patients could help design predictive models where the combination of 2yDFS and pCR should be employed as the primary outcome.

Bayesian network structure for predicting local tumor recurrence in rectal cancer patients treated with neoadjuvant chemoradiation followed by surgery.

Background and Purpose: Tumor recurrence, a characteristic of malignant tumors, is the biggest concern for rectal cancer survivors. The epidemiology of the disease calls for a pressing need to improve healthcare quality and patient outcomes. Prediction models such as Bayesian networks, which can probabilistically reason under uncertainty, could assist caregivers with patient management. However, some concerns are associated with the standard approaches to developing these structures in medicine. Therefore, this study aims to compare Bayesian network structures that stem from these two techniques. Patients and Methods: A retrospective analysis was performed on 6754 locally advanced rectal cancer (LARC) patients enrolled in 14 international clinical trials. Local tumor recurrence at 2, 3, and 5-years was defined as the endpoints of interest. Five rectal cancer treating physicians from three countries elicited the expert structure. The algorithmic structure was inferred from the data with the hill-climbing algorithm. Structural performance was assessed with calibration plots and area under the curve values. Results: The area under the curve for the expert structure on the training and validation data was above 0.9 and 0.8, respectively, for all the time points. However, the algorithmic structure had superior predictive performance over the expert structure for all time points of interest. Conclusion: We have developed and internally validated a Bayesian networks structure from experts' opinions, which can predict the risk of a LARC patient developing a tumor recurrence at 2, 3, and 5 years. Our result shows that the algorithmic-based structures are more performant and less interpretable than expert-based structures.

The Assisi Think Tank Meeting Breast Large Database for Standardized Data Collection in Breast Cancer-ATTM.BLADE.

Background: During the 2016 Assisi Think Tank Meeting (ATTM) on breast cancer, the panel of experts proposed developing a validated system, based on rapid learning health care (RLHC) principles, to standardize inter-center data collection and promote personalized treatments for breast cancer. Material and Methods: The seven-step Breast LArge DatabasE (BLADE) project included data collection, analysis, application, and evaluation on a data-sharing platform. The multidisciplinary team developed a consensus-based ontology of validated variables with over 80% agreement. This English-language ontology constituted a breast cancer library with seven knowledge domains: baseline, primary systemic therapy, surgery, adjuvant systemic therapies, radiation therapy, follow-up, and toxicity. The library was uploaded to the BLADE domain. The safety of data encryption and preservation was tested according to General Data Protection Regulation (GDPR) guidelines on data from 15 clinical charts. The system was validated on 64 patients who had undergone post-mastectomy radiation therapy. In October 2018, the BLADE system was approved by the Ethical Committee of Fondazione Policlinico Gemelli IRCCS, Rome, Italy (Protocol No. 0043996/18). Results: From June 2016 to July 2019, the multidisciplinary team completed the work plan. An ontology of 218 validated variables was uploaded to the BLADE domain. The GDPR safety test confirmed encryption and data preservation (on 5000 random cases). All validation benchmarks were met. Conclusion:BLADE is a support system for follow-up and assessment of breast cancer care. To successfully develop and validate it as the first standardized data collection system, multidisciplinary collaboration was crucial in selecting its ontology and knowledge domains. BLADE is suitable for multi-center uploading of retrospective and prospective clinical data, as it ensures anonymity and data privacy.

Predicting Radiotherapy Impact on Late Bladder Toxicity in Prostate Cancer Patients: An Observational Study.

BACKGROUND AND PURPOSE: The aim of our study was to elaborate a suitable model on bladder late toxicity in prostate cancer (PC) patients treated by radiotherapy with volumetric technique. MATERIALS AND METHODS: PC patients treated between September 2010 and April 2017 were included in the analysis. An observational study was performed collecting late toxicity data of any grade, according to RTOG and CTCAE 4.03 scales, cumulative dose volumes histograms were exported for each patient. Vdose, the value of dose to a specific volume of organ at risk (OAR), impact was analyzed through the Mann-Whitney rank-sum test. Logistic regression was used as the final model. The model performance was estimated by taking 1000 samples with replacement from the original dataset and calculating the AUC average. In addition, the calibration plot (Hosmer-Lemeshow goodness-of-fit test) was used to evaluate the performance of internal validation. RStudio Software version 3.3.1 and an in house developed software package "Moddicom" were used. RESULTS: Data from 175 patients were collected. The median follow-up was 39 months (min-max 3.00-113.00). We performed Mann-Whitney rank-sum test with continuity correction in the subset of patients with late bladder toxicity grade ≥ 2: a statistically significant p-value with a Vdose of 51.43 Gy by applying a logistic regression model (coefficient 4.3, p value 0.025) for the prediction of the development of late G ≥ 2 GU toxicity was observed. The performance for the model's internal validation was evaluated, with an AUC equal to 0.626. Accuracy was estimated through the elaboration of a calibration plot. CONCLUSIONS: Our preliminary results could help to optimize treatment planning procedures and customize treatments.

GENERATOR Breast DataMart-The Novel Breast Cancer Data Discovery System for Research and Monitoring: Preliminary Results and Future Perspectives.

BACKGROUND: Artificial Intelligence (AI) is increasingly used for process management in daily life. In the medical field AI is becoming part of computerized systems to manage information and encourage the generation of evidence. Here we present the development of the application of AI to IT systems present in the hospital, for the creation of a DataMart for the management of clinical and research processes in the field of breast cancer. MATERIALS AND METHODS: A multidisciplinary team of radiation oncologists, epidemiologists, medical oncologists, breast surgeons, data scientists, and data management experts worked together to identify relevant data and sources located inside the hospital system. Combinations of open-source data science packages and industry solutions were used to design the target framework. To validate the DataMart directly on real-life cases, the working team defined tumoral pathology and clinical purposes of proof of concepts (PoCs). RESULTS: Data were classified into "Not organized, not 'ontologized' data", "Organized, not 'ontologized' data", and "Organized and 'ontologized' data". Archives of real-world data (RWD) identified were platform based on ontology, hospital data warehouse, PDF documents, and electronic reports. Data extraction was performed by direct connection with structured data or text-mining technology. Two PoCs were performed, by which waiting time interval for radiotherapy and performance index of breast unit were tested and resulted available. CONCLUSIONS: GENERATOR Breast DataMart was created for supporting breast cancer pathways of care. An AI-based process automatically extracts data from different sources and uses them for generating trend studies and clinical evidence. Further studies and more proof of concepts are needed to exploit all the potentials of this system.

Multidisciplinary Tumor Board Smart Virtual Assistant in Locally Advanced Cervical Cancer: A Proof of Concept.

AIM: The first prototype of the "Multidisciplinary Tumor Board Smart Virtual Assistant" is presented, aimed to (i) Automated classification of clinical stage starting from different free-text diagnostic reports; (ii) Resolution of inconsistencies by identifying controversial cases drawing the clinician's attention to particular cases worthy for multi-disciplinary discussion; (iii) Support environment for education and knowledge transfer to junior staff; (iv) Integrated data-driven decision making and standardized language and interpretation. PATIENTS AND METHOD: Data from patients affected by Locally Advanced Cervical Cancer (LACC), FIGO stage IB2-IVa, treated between 2015 and 2018 were extracted. Magnetic Resonance (MR), Gynecologic examination under general anesthesia (EAU), and Positron Emission Tomography-Computed Tomography (PET-CT) performed at the time of diagnosis were the items from the Electronic Health Records (eHRs) considered for analysis. An automated extraction of eHR that capture the patient's data before the diagnosis and then, through Natural Language Processing (NLP), analysis and categorization of all data to transform source information into structured data has been performed. RESULTS: In the first round, the system has been used to retrieve all the eHR for the 96 patients with LACC. The system has been able to classify all patients belonging to the training set and - through the NLP procedures - the clinical features were analyzed and classified for each patient. A second important result was the setup of a predictive model to evaluate the patient's staging (accuracy of 94%). Lastly, we created a user-oriented operational tool targeting the MTB who are confronted with the challenge of large volumes of patients to be diagnosed in the most accurate way. CONCLUSION: This is the first proof of concept concerning the possibility of creating a smart virtual assistant for the MTB. A significant benefit could come from the integration of these automated methods in the collaborative, crucial decision stages.

A field strength independent MR radiomics model to predict pathological complete response in locally advanced rectal cancer.

PURPOSE: Aim of this study was to develop a generalised radiomics model for predicting pathological complete response after neoadjuvant chemo-radiotherapy in locally advanced rectal cancer patients using pre-CRT T2-weighted images acquired at a 1.5 T and a 3 T scanner. METHODS: In two institutions, 195 patients were scanned: 136 patients were scanned on a 1.5 T MR scanner, 59 patients on a 3 T MR scanner. Gross tumour volumes were delineated on the MR images and 496 radiomic features were extracted, applying the intensity-based (IB) filter. Features were standardised with Z-score normalisation and an initial feature selection was carried out using Wilcoxon-Mann-Whitney test: The most significant features at 1.5 T and 3 T were selected as main features. Several logistic regression models combining the main features with a third one selected by those resulting significant were elaborated and evaluated in terms of area under curve (AUC). A tenfold cross-validation was repeated 300 times to evaluate the model robustness. RESULTS: Three features were selected: maximum fractal dimension with IB = 0-50, energy and grey-level non-uniformity calculated on the run-length matrix with IB = 0-50. The AUC of the model applied to the whole dataset after cross-validation was 0.72, while values of 0.70 and 0.83 were obtained when 1.5 T and 3 T patients were considered, respectively. CONCLUSIONS: The model elaborated showed good performance, even when data from patients scanned on 1.5 T and 3 T were merged. This shows that magnetic field intensity variability can be overcome by means of selecting appropriate image features.

Timing to achieve the highest rate of pCR after preoperative radiochemotherapy in rectal cancer: a pooled analysis of 3085 patients from 7 randomized trials.

PURPOSE: Optimal timing of surgery after neoadjuvant chemoradiotherapy (Nad-CRT) is still controversial in locally advanced rectal cancer (LARC). The primary goal of this study was to determine the best surgical interval (SI) to achieve the highest rate of pathological complete response (pCR) and secondly to evaluate the effect on survival outcomes according to the SI. PATIENTS AND METHODS: Patients data were extracted from the international randomized trials: Accord12/0405, EORTC22921, FFCD9203, CAO/ARO/AIO-94, CAO-ARO-AIO-04, INTERACT and TROG01.04. Inclusion criteria were: age≥ 18, cT3-T4 and cN0-2, no clinical evidence of distant metastasis at diagnosis, Nad-CRT followed by surgery. Pearson's Chi-squared test with Yates' continuity correction for categorical variables, the Mann-Whitney test for continuous variables, Mann-Kendall test, Kaplan-Meier curves with log-rank test, univariate and multivariate logistic regression model was used for data analysis. RESULTS: 3085 patients met the inclusion criteria. Overall, the pCR rate was 14% at a median SI of 6 weeks (range 1-31). The cumulative pCR rate increased significantly when SI lengthened, with 95% of pCR events within 10 weeks from Nad-CRT. At univariate and multivariate logistic regression analysis, lengthening of SI (p< 0.01), radiotherapy dose (p< 0.01), and the addition of oxaliplatin to Nad-CRT (p< 0.01) had a favorable impact on pCR. Furthermore, lengthening of SI was not impact on local recurrences, distance metastases, and overall survival. CONCLUSION: This pooled analysis suggests that the best time to achieve pCR in LARC is at 10 weeks, considering that the lengthening of SI is not detrimental concerning survival outcomes.

Delta Radiomics Can Predict Distant Metastasis in Locally Advanced Rectal Cancer: The Challenge to Personalize the Cure.

PURPOSE: Distant metastases are currently the main cause of treatment failure in locally advanced rectal cancer (LARC) patients. The aim of this research is to investigate a correlation between the variation of radiomics features using pre- and post-neoadjuvant chemoradiation (nCRT) magnetic resonance imaging (MRI) with 2 years distant metastasis (2yDM) rate in LARC patients. METHODS AND MATERIALS: Diagnostic pre- and post- nCRT MRI of LARC patients, treated in a single institution from May 2008 to June 2015 with an adequate follow-up time, were retrospectively collected. Gross tumor volumes (GTV) were contoured by an abdominal radiologist and blindly reviewed by a radiation oncologist expert in rectal cancer. The dataset was firstly randomly split into 90% training data, for features selection, and 10% testing data, for the validation. The final set of features after the selection was used to train 15 different classifiers using accuracy as target metric. The models' performance was then assessed on the testing data and the best performing classifier was then selected, maximising the confusion matrix balanced accuracy (BA). RESULTS: Data regarding 213 LARC patients (36% female, 64% male) were collected. Overall 2yDM was 17%. A total of 2,606 features extracted from the pre- and post- nCRT GTV were tested and 4 features were selected after features selection process. Among the 15 tested classifiers, logistic regression proved to be the best performing one with a testing set BA, sensitivity and specificity of 78.5%, 71.4% and 85.7%, respectively. CONCLUSIONS: This study supports a possible role of delta radiomics in predicting following occurrence of distant metastasis. Further studies including a consistent external validation are needed to confirm these results and allows to translate radiomics model in clinical practice. Future integration with clinical and molecular data will be mandatory to fully personalized treatment and follow-up approaches.

Intraluminal Brachytherapy in Unresectable Extrahepatic Biliary Duct Cancer: An Italian Pooled Analysis.

BACKGROUND/AIM: To evaluate the outcome of patients with unresectable extrahepatic cholangiocarcinoma (CC) treated with external-beam radiotherapy (EBRT) and concurrent chemotherapy (CT) with or without intraluminal brachytherapy (ILBT) boost or with definitive ILBT. PATIENTS AND METHODS: A pooled analysis of patients with non-metastatic unresectable CC was performed. They were treated in three different institution with EBRT plus CT with or without an ILBT boost. Some patients received only ILBT with curative dose. RESULTS: Seventy-three patients were included in the analysis. Thirty-nine patients (53%) received EBRT treatment with ILBT boost (18 patients with CT during EBRT), while 28 patients (38%) were treated with EBRT (CT in 26 patients) and 6 patients (8.2%) with definitive ILBT (2 patients with CT). CT was administered including either the use of gemcitabine or 5-fluorouracil. With a median follow-up of 16 month (range=1-94 months), median overall survival (OS) was 16 months. Overall median LC was 16 months and patients who underwent ILBT had a better local control (LC) (p=0.018). CONCLUSION: The role of ILBT in unresectable CC is not yet supported by robust evidence in the literature. However, within this limit, preliminary results seem to suggest an improved local control in patients treated with ILBT, almost comparable to the ones of standard chemo-radiotherapy (CRT).

SKIN-COBRA (Consortium for Brachytherapy data Analysis) ontology: The first step towards interdisciplinary standardized data collection for personalized oncology in skin cancer.

PURPOSE: The primary objective of the SKIN-COBRA (Consortium for Brachytherapy data Analysis) ontology is to define a specific terminological system to standardize data collection for non-melanoma skin cancer patients treated with brachytherapy (BT, interventional radiotherapy). Through ontological characterization of information, it is possible to find, isolate, organize, and integrate its meaning. MATERIAL AND METHODS: SKIN-COBRA is a standardized data collection consortium for non-melanoma skin patients treated with BT, including 8 cancer centers. Its ontology was firstly defined by a multicentric and multidisciplinary working group and evaluated by the consortium, followed by a multi-professional technical commission involving a mathematician, an engineer, a physician with experience in data storage, a programmer, and a software expert. RESULTS: Two hundred and ninety variables were defined in 10 input forms. There are 3 levels, with each offering a specific type of analysis: 1. Registry level (epidemiology analysis); 2. Procedures level (standard oncology analysis); 3. Research level (radiomics analysis). The ontology was approved by the technical commission and consortium, and an ad-hoc software system was defined to be implemented in the SKIN-COBRA consortium. CONCLUSIONS: Large databases are natural extension of traditional statistical approaches, a valuable and increasingly necessary tool for modern healthcare system. Future analysis of the collected multinational and multicenter data will show whether the use of the system can produce high-quality evidence to support multidisciplinary management of non-melanoma skin cancer and utilizing this information for personalized treatment decisions.

Exploring technical issues in personalized medicine: NSCLC survival prediction by quantitative image analysis-usefulness of density correction of volumetric CT data.

The aim of this study was to apply density correction method to the quantitative image analysis of non-small cell lung cancer (NSCLC) computed tomography (CT) images, determining its influence on overall survival (OS) prediction of surgically treated patients. Clinicopathological (CP) data and preoperative CT scans, pre- and post-contrast medium (CM) administration, of 57 surgically treated NSCLC patients, were retrospectively collected. After CT volumetric density measurement of primary gross tumour volume (GTV), aorta and tracheal air, density correction was conducted on GTV (reference values: aortic blood and tracheal air). For each resulting data set (combining CM administration and normalization), first-order statistical and textural features were extracted. CP and imaging data were correlated with patients 1-, 3- and 5-year OS, alone and combined (uni-/multivariate logistic regression and Akaike information criterion). Predictive performance was evaluated using the ROC curves and AUC values and compared among non-normalized/normalized data sets (DeLong test). The best predictive values were obtained when combining CP and imaging parameters (AUC values: 1 year 0.72; 3 years 0.82; 5 years 0.78). After normalization resulted an improvement in predicting 1-year OS for some of the grey level size zonebased features (large zone low grey level emphasis) and for the combined CP-imaging model, a worse performance for grey level co-occurrence matrix (cluster prominence and shade) and first-order statistical (range) parameters for 1- and 5-year OS, respectively. The negative performance of cluster prominence in predicting 1-year OS was the only statistically significant result (p value 0.05). Density corrections of volumetric CT data showed an opposite influence on the performance of imaging quantitative features in predicting OS of surgically treated NSCLC patients, even if no statistically significant for almost all predictors.

Distributed learning on 20 000+ lung cancer patients - The Personal Health Train.

BACKGROUND AND PURPOSE: Access to healthcare data is indispensable for scientific progress and innovation. Sharing healthcare data is time-consuming and notoriously difficult due to privacy and regulatory concerns. The Personal Health Train (PHT) provides a privacy-by-design infrastructure connecting FAIR (Findable, Accessible, Interoperable, Reusable) data sources and allows distributed data analysis and machine learning. Patient data never leaves a healthcare institute. MATERIALS AND METHODS: Lung cancer patient-specific databases (tumor staging and post-treatment survival information) of oncology departments were translated according to a FAIR data model and stored locally in a graph database. Software was installed locally to enable deployment of distributed machine learning algorithms via a central server. Algorithms (MATLAB, code and documentation publicly available) are patient privacy-preserving as only summary statistics and regression coefficients are exchanged with the central server. A logistic regression model to predict post-treatment two-year survival was trained and evaluated by receiver operating characteristic curves (ROC), root mean square prediction error (RMSE) and calibration plots. RESULTS: In 4 months, we connected databases with 23 203 patient cases across 8 healthcare institutes in 5 countries (Amsterdam, Cardiff, Maastricht, Manchester, Nijmegen, Rome, Rotterdam, Shanghai) using the PHT. Summary statistics were computed across databases. A distributed logistic regression model predicting post-treatment two-year survival was trained on 14 810 patients treated between 1978 and 2011 and validated on 8 393 patients treated between 2012 and 2015. CONCLUSION: The PHT infrastructure demonstrably overcomes patient privacy barriers to healthcare data sharing and enables fast data analyses across multiple institutes from different countries with different regulatory regimens. This infrastructure promotes global evidence-based medicine while prioritizing patient privacy.

Deep Learning: A Review for the Radiation Oncologist.

Introduction: Deep Learning (DL) is a machine learning technique that uses deep neural networks to create a model. The application areas of deep learning in radiation oncology include image segmentation and detection, image phenotyping, and radiomic signature discovery, clinical outcome prediction, image dose quantification, dose-response modeling, radiation adaptation, and image generation. In this review, we explain the methods used in DL and perform a literature review using the Medline database to identify studies using deep learning in radiation oncology. The search was conducted in April 2018, and identified studies published between 1997 and 2018, strongly skewed toward 2015 and later. Methods: A literature review was performed using PubMed/Medline in order to identify important recent publications to be synthesized into a review of the current status of Deep Learning in radiation oncology, directed at a clinically-oriented reader. The search strategy included the search terms "radiotherapy" and "deep learning." In addition, reference lists of selected articles were hand searched for further potential hits of relevance to this review. The search was conducted in April 2018, and identified studies published between 1997 and 2018, strongly skewed toward 2015 and later. Results: Studies using DL for image segmentation were identified in Brain (n = 2), Head and Neck (n = 3), Lung (n = 6), Abdominal (n = 2), and Pelvic (n = 6) cancers. Use of Deep Learning has also been reported for outcome prediction, such as toxicity modeling (n = 3), treatment response and survival (n = 2), or treatment planning (n = 5). Conclusion: Over the past few years, there has been a significant number of studies assessing the performance of DL techniques in radiation oncology. They demonstrate how DL-based systems can aid clinicians in their daily work, be it by reducing the time required for or the variability in segmentation, or by helping to predict treatment outcomes and toxicities. It still remains to be seen when these techniques will be employed in routine clinical practice.

Delta radiomics for rectal cancer response prediction with hybrid 0.35 T magnetic resonance-guided radiotherapy (MRgRT): a hypothesis-generating study for an innovative personalized medicine approach.

The aim of this study was to evaluate the variation of radiomics features, defined as "delta radiomics", in patients undergoing neoadjuvant radiochemotherapy (RCT) for rectal cancer treated with hybrid magnetic resonance (MR)-guided radiotherapy (MRgRT). The delta radiomics features were then correlated with clinical complete response (cCR) outcome, to investigate their predictive power. A total of 16 patients were enrolled, and 5 patients (31%) showed cCR at restaging examinations. T2*/T1 MR images acquired with a hybrid 0.35 T MRgRT unit were considered for this analysis. An imaging acquisition protocol of 6 MR scans per patient was performed: the first MR was acquired at first simulation (t0) and the remaining ones at fractions 5, 10, 15, 20 and 25. Radiomics features were extracted from the gross tumour volume (GTV), and each feature was correlated with the corresponding delivered dose. The variations of each feature during treatment were quantified, and the ratio between the values calculated at different dose levels and the one extracted at t0 was calculated too. The Wilcoxon-Mann-Whitney test was performed to identify the features whose variation can be predictive of cCR, assessed with a MR acquired 6 weeks after RCT and digital examination. The most predictive feature ratios in cCR prediction were the L_least and glnu ones, calculated at the second week of treatment (22 Gy) with a p value = 0.001. Delta radiomics approach showed promising results and the quantitative analysis of images throughout MRgRT treatment can successfully predict cCR offering an innovative personalized medicine approach to rectal cancer treatment.

Towards a modular decision support system for radiomics: A case study on rectal cancer.

Following the personalized medicine paradigm, there is a growing interest in medical agents capable of predicting the effect of therapies on patients, by exploiting the amount of data that is now available for each patient. In disciplines like oncology, where images and scans are available, the exploitation of medical images can provide an additional source of potentially useful information. The study and analysis of features extracted by medical images, exploited for predictive purposes, is termed radiomics. A number of tools are available for supporting some of the steps of the radiomics process, but there is a lack of approaches which are able to deal with all the steps of the process. In this paper, we introduce a medical agent-based decision support system capable of handling the whole radiomics process. The proposed system is tested on two independent data sets of patients treated for rectal cancer. Experimental results indicate that the system is able to generate highly performant centre-specific predictive model, and show the issues related to differences in data sets collected by different centres, and how such issues can affect the performance of the generated predictive models.