Radiomics Beyond the Hype: A Critical Evaluation Toward Oncologic Clinical Use.

Radiomics is a promising and fast-developing field within oncology that involves the mining of quantitative high-dimensional data from medical images. Radiomics has the potential to transform cancer management, whereby radiomics data can be used to aid early tumor characterization, prognosis, risk stratification, treatment planning, treatment response assessment, and surveillance. Nevertheless, certain challenges have delayed the clinical adoption and acceptability of radiomics in routine clinical practice. The objectives of this report are to (a) provide a perspective on the translational potential and potential impact of radiomics in oncology; (b) explore frequent challenges and mistakes in its derivation, encompassing study design, technical requirements, standardization, model reproducibility, transparency, data sharing, privacy concerns, quality control, as well as the complexity of multistep processes resulting in less radiologist-friendly interfaces; (c) discuss strategies to overcome these challenges and mistakes; and (d) propose measures to increase the clinical use and acceptability of radiomics, taking into account the different perspectives of patients, health care workers, and health care systems. Keywords: Radiomics, Oncology, Cancer Management, Artificial Intelligence © RSNA, 2024.

The ras-related protein RAB22A interacts with hypoxia-inducible factor 1-alpha (HIF-1α) in MDA-MB-231 breast cancer cells in hypoxia.

BACKGROUND: Recent studies suggest that hypoxia-inducible factor 1-alpha (HIF-1α) and the small GTPase protein Ras-related protein Rab-22 A (RAB22A) may be colocalized in the cytoplasm and that as a conequence they may enhance the formation of microvesicles in breast cancer cells under hypoxia. Therefore, we sought to determine whether these two proteins are present in intracellular complexes in breast carcinoma cells. METHODS AND RESULTS: Evaluation using molecular docking indicated that HIF-1α and RAB22A interact with each other. Co-immunoprecipitation of endogenous or ectopically expressed HIF-1α and RAB22A proteins in MDA-MB-231 breast cancer cells or HEK-293T cells demonstrated that endogenous HIF-1α and RAB22A can form an intracellular complex; however, transiently expressed HIF-1α and RAB22A failed to interact. Investigating RAB22A and HIF-1α interactions in various cancer cell lines under hypoxia may shed light on their roles in cancer cell survival and progression through regulation of intracellular trafficking by HIF-1α under hypoxic conditions. CONCLUSIONS: Our study is the first to reveal the potential involvement of HIF-1α in intracellular trafficking through physical interactions with the small GTPase protein RAB22A. We discuss the implications of our work on the role of exosomes and microvesicles in tumor invasiveness.

METhodological RadiomICs Score (METRICS): a quality scoring tool for radiomics research endorsed by EuSoMII.

PURPOSE: To propose a new quality scoring tool, METhodological RadiomICs Score (METRICS), to assess and improve research quality of radiomics studies. METHODS: We conducted an online modified Delphi study with a group of international experts. It was performed in three consecutive stages: Stage#1, item preparation; Stage#2, panel discussion among EuSoMII Auditing Group members to identify the items to be voted; and Stage#3, four rounds of the modified Delphi exercise by panelists to determine the items eligible for the METRICS and their weights. The consensus threshold was 75%. Based on the median ranks derived from expert panel opinion and their rank-sum based conversion to importance scores, the category and item weights were calculated. RESULT: In total, 59 panelists from 19 countries participated in selection and ranking of the items and categories. Final METRICS tool included 30 items within 9 categories. According to their weights, the categories were in descending order of importance: study design, imaging data, image processing and feature extraction, metrics and comparison, testing, feature processing, preparation for modeling, segmentation, and open science. A web application and a repository were developed to streamline the calculation of the METRICS score and to collect feedback from the radiomics community. CONCLUSION: In this work, we developed a scoring tool for assessing the methodological quality of the radiomics research, with a large international panel and a modified Delphi protocol. With its conditional format to cover methodological variations, it provides a well-constructed framework for the key methodological concepts to assess the quality of radiomic research papers. CRITICAL RELEVANCE STATEMENT: A quality assessment tool, METhodological RadiomICs Score (METRICS), is made available by a large group of international domain experts, with transparent methodology, aiming at evaluating and improving research quality in radiomics and machine learning. KEY POINTS: • A methodological scoring tool, METRICS, was developed for assessing the quality of radiomics research, with a large international expert panel and a modified Delphi protocol. • The proposed scoring tool presents expert opinion-based importance weights of categories and items with a transparent methodology for the first time. • METRICS accounts for varying use cases, from handcrafted radiomics to entirely deep learning-based pipelines. • A web application has been developed to help with the calculation of the METRICS score ( https://metricsscore.github.io/metrics/METRICS.html ) and a repository created to collect feedback from the radiomics community ( https://github.com/metricsscore/metrics ).

Cancer Marker Immunosensing through Surface-Enhanced Photoluminescence on Nanostructured Silver Substrates.

Nanostructured noble metal surfaces enhance the photoluminescence emitted by fluorescent molecules, permitting the development of highly sensitive fluorescence immunoassays. To this end, surfaces with silicon nanowires decorated with silver nanoparticles in the form of dendrites or aggregates were evaluated as substrates for the immunochemical detection of two ovarian cancer indicators, carbohydrate antigen 125 (CA125) and human epididymis protein 4 (HE4). The substrates were prepared by metal-enhanced chemical etching of silicon wafers to create, in one step, silicon nanowires and silver nanoparticles on top of them. For both analytes, non-competitive immunoassays were developed using pairs of highly specific monoclonal antibodies, one for analyte capture on the substrate and the other for detection. In order to facilitate the identification of the immunocomplexes through a reaction with streptavidin labeled with Rhodamine Red-X, the detection antibodies were biotinylated. An in-house-developed optical set-up was used for photoluminescence signal measurements after assay completion. The detection limits achieved were 2.5 U/mL and 3.12 pM for CA125 and HE4, respectively, with linear dynamic ranges extending up to 500 U/mL for CA125 and up to 500 pM for HE4, covering the concentration ranges of both healthy and ovarian cancer patients. Thus, the proposed method could be implemented for the early diagnosis and/or prognosis and monitoring of ovarian cancer.

Overcoming phenotypic switching: targeting protein-protein interactions in cancer.

Alternative protein-protein interactions (PPIs) arising from mutations or post-translational modifications (PTMs), termed phenotypic switching (PS), are critical for the transmission of alternative pathogenic signals and are particularly significant in cancer. In recent years, PPIs have emerged as promising targets for rational drug design, primarily because their high specificity facilitates targeting of disease-related signaling pathways. However, obstacles exist at the molecular level that arise from the properties of the interaction interfaces and the propensity of small molecule drugs to interact with more than one cleft surface. The difficulty in identifying small molecules that act as activators or inhibitors to counteract the biological effects of mutations raises issues that have not been encountered before. For example, small molecules can bind tightly but may not act as drugs or bind to multiple sites (interaction promiscuity). Another reason is the absence of significant clefts on protein surfaces; if a pocket is present, it may be too small, or its geometry may prevent binding. PS, which arises from oncogenic (alternative) signaling, causes drug resistance and forms the basis for the systemic robustness of tumors. In this review, the properties of PPI interfaces relevant to the design and development of targeting drugs are examined. In addition, the interactions between three tyrosine kinase inhibitors (TKIs) employed as drugs are discussed. Finally, potential novel targets of one of these drugs were identified in silico.

SERS Immunosensors for Cancer Markers Detection.

Early diagnosis and monitoring are essential for the effective treatment and survival of patients with different types of malignancy. To this end, the accurate and sensitive determination of substances in human biological fluids related to cancer diagnosis and/or prognosis, i.e., cancer biomarkers, is of ultimate importance. Advancements in the field of immunodetection and nanomaterials have enabled the application of new transduction approaches for the sensitive detection of single or multiple cancer biomarkers in biological fluids. Immunosensors based on surface-enhanced Raman spectroscopy (SERS) are examples where the special properties of nanostructured materials and immunoreagents are combined to develop analytical tools that hold promise for point-of-care applications. In this frame, the subject of this review article is to present the advancements made so far regarding the immunochemical determination of cancer biomarkers by SERS. Thus, after a short introduction about the principles of both immunoassays and SERS, an extended presentation of up-to-date works regarding both single and multi-analyte determination of cancer biomarkers is presented. Finally, future perspectives on the field of SERS immunosensors for cancer markers detection are briefly discussed.

SERS Determination of Oxidative Stress Markers in Saliva Using Substrates with Silver Nanoparticle-Decorated Silicon Nanowires.

Glutathione and malondialdehyde are two compounds commonly used to evaluate the oxidative stress status of an organism. Although their determination is usually performed in blood serum, saliva is gaining ground as the biological fluid of choice for oxidative stress determination at the point of need. For this purpose, surface-enhanced Raman spectroscopy (SERS), which is a highly sensitive method for the detection of biomolecules, could offer additional advantages regarding the analysis of biological fluids at the point of need. In this work, silicon nanowires decorated with silver nanoparticles made by metal-assisted chemical etching were evaluated as substrates for the SERS determination of glutathione and malondialdehyde in water and saliva. In particular, glutathione was determined by monitoring the reduction in the Raman signal obtained from substrates modified with crystal violet upon incubation with aqueous glutathione solutions. On the other hand, malondialdehyde was detected after a reaction with thiobarbituric acid to produce a derivative with a strong Raman signal. The detection limits achieved after optimization of several assay parameters were 50 and 3.2 nM for aqueous solutions of glutathione and malondialdehyde, respectively. In artificial saliva, however, the detection limits were 2.0 and 0.32 µM for glutathione and malondialdehyde, respectively, which are, nonetheless, adequate for the determination of these two markers in saliva.

Enhanced Photoluminescence of R6G Dyes from Metal Decorated Silicon Nanowires Fabricated through Metal Assisted Chemical Etching.

In this study, we developed active substrates consisting of Ag-decorated silicon nanowires on a Si substrate using a single-step Metal Assisted Chemical Etching (MACE) process, and evaluated their performance in the identification of low concentrations of Rhodamine 6G using surface-enhanced photoluminescence spectroscopy. Different structures with Ag-aggregates as well as Ag-dendrites were fabricated and studied depending on the etching parameters. Moreover, the addition of Au nanoparticles by simple drop-casting on the MACE-treated surfaces can enhance the photoluminescence significantly, and the structures have shown a Limit of Detection of Rhodamine 6G down to 10-12 M for the case of the Ag-dendrites enriched with Au nanoparticles.

APOBEC3B Is Co-Expressed with PKCα/NF-κB in Oral and Oropharyngeal Squamous Cell Carcinomas.

The enzymatic activity of APOBEC3B (A3B) has been implicated as a prime source of mutagenesis in head and neck squamous cell carcinoma (HNSCC). The expression of Protein Kinase C α (PKCα) and Nuclear Factor-κΒ p65 (NF-κΒ p65) has been linked to the activation of the classical and the non-canonical NF-κB signaling pathways, respectively, both of which have been shown to lead to the upregulation of A3B. Accordingly, the aim of the present study was to evaluate the expression of PKCα, NF-κΒ p65 and A3B in non-HPV related oral and oropharyngeal squamous cell carcinomas (SCC), by means of immunohistochemistry and in silico methods. PKCα was expressed in 29/36 (80%) cases of oral and oropharyngeal SCCs, with 25 (69%) cases showing a PKCα+/A3B+ phenotype and only 6/36 (17%) cases showing a PKCα-/A3B+ phenotype. Εxpression of NF-κB p65 was seen in 33/35 (94%) cases of oral and oropharyngeal SCCs, with 30/35 (86%) cases showing an NF-κB p65+/A3B+ phenotype and only 2/35 (6%) cases showing an NF-κB p65-/A3B+ phenotype. In addition, mRNA expression analysis, using the UALCAN database, revealed strong expression of all three genes. These findings indicate that the expression of A3B is associated with PKCα/NF-κB p65 expression and suggest a potential role for the PKC/NF-κB signaling pathway in the development of oral and oropharyngeal cancer.

The Dynamic and Crucial Role of the Arginine Methylproteome in Myoblast Cell Differentiation.

Protein arginine methylation is an extensive and functionally significant post-translational modification. However, little is known about its role in differentiation at the systems level. Using stable isotope labeling by amino acids in cell culture (SILAC) proteomics of whole proteome analysis in proliferating or five-day differentiated mouse C2C12 myoblasts, followed by high-resolution mass spectrometry, biochemical assays, and specific immunoprecipitation of mono- or dimethylated arginine peptides, we identified several protein families that were differentially methylated on arginine. Our study is the first to reveal global changes in the arginine mono- or dimethylation of proteins in proliferating myoblasts and differentiated myocytes and to identify enriched protein domains and novel short linear motifs (SLiMs). Our data may be crucial for dissecting the links between differentiation and cancer growth.

Enhancing cancer differentiation with synthetic MRI examinations via generative models: a systematic review.

Contemporary deep learning-based decision systems are well-known for requiring high-volume datasets in order to produce generalized, reliable, and high-performing models. However, the collection of such datasets is challenging, requiring time-consuming processes involving also expert clinicians with limited time. In addition, data collection often raises ethical and legal issues and depends on costly and invasive procedures. Deep generative models such as generative adversarial networks and variational autoencoders can capture the underlying distribution of the examined data, allowing them to create new and unique instances of samples. This study aims to shed light on generative data augmentation techniques and corresponding best practices. Through in-depth investigation, we underline the limitations and potential methodology pitfalls from critical standpoint and aim to promote open science research by identifying publicly available open-source repositories and datasets.

COVID-19 Disease and Vaccination: Knowledge, Fears, Perceptions and Feelings of Regret for Not Having Been Vaccinated among Hospitalized Greek Patients Suffering SARS-CoV-2 Infection.

BACKGROUND: The development of vaccines against COVID-19 has greatly altered the natural course of this infection, reducing the disease's severity and patients' hospitalization. However, hesitancy against vaccination remains an obstacle in the attempt to achieve appropriate herd immunity that could reduce the spread of SARS-CoV-2. The aim of this study was to investigate the perceptions and attitudes of COVID-19 patients hospitalized during the fourth pandemic wave in two Greek hospitals and assess whether their experience had changed their intentions regarding vaccination against COVID-19. METHODS: This is a cross-sectional, questionnaire-based survey, conducted from 31 August 2021 to 18 February 2022 in the COVID-19 departments of two tertiary care hospitals. The questionnaire included questions regarding the patients' educational level, knowledge and beliefs regarding SARS-CoV-2, personal protection measures, beliefs regarding vaccination, vaccination status, reasons for not been vaccinated against SARS-CoV-2, feelings of regret for not been vaccinated, and willingness to be vaccinated in the future. All adult patients with COVID-19 were eligible, regardless of their vaccination status against SARS-CoV-2. RESULTS: In total, 162 patients agreed and participated in the study, with 97% of them suffering severe COVID-19. Their median age was 56 years, and 59.9% (97 patients) were male. Among them, 43.8% had been vaccinated against COVID-19. When unvaccinated patients were asked the reasons for not being vaccinated, the most frequent responses were that they were waiting for more scientific data, due to uncertainty about long-term consequences of the vaccine, and their fear of thrombosis. When at discharge, unvaccinated hospitalized COVID-19 patients were asked whether they would get vaccinated if they could turn time back, and 64.7% of them replied positively. CONCLUSIONS: The study reveals several patients' fears and misconceptions and suggests that there is room for implementing measures that could reduce knowledge gaps allowing for improvement of vaccination rates against COVID-19.

Induced differentiation and molecular characterization of monocytes-derived multipotential cells generated from commonly discarded leukapheresis filters.

Monocyte-derived multipotential cells (MOMCs) are a subpopulation of monocytes that appear to be capable of differentiating into many cell populations, thus MOMCs can be an ideal autologous transplantable cell source for regenerative medicine. In this study, we generated MOMCs from leukapheresis filters, evaluated their ability to differentiate to endothelium and osteocytes and performed their molecular characterization. For this purpose, leukapheresis filters were collected from a hospital blood donation department and used for leukocytes isolation. The isolated leukocytes were cultured in a medium supplemented with SDF-1a for MOMCs generation. We evaluated the expression of the multipotency genes ZNF217, ZNF878, ESRRB, SALL4, KLF4, SOX2, NANOG, OCT4, GAPDH, CD34 and c- MYC in MOMCs with real-time reverse transcription PCR (qRT-PCR) and the differentiation capacity of MOMCs to osteocytes and endothelium with qRT-PCR. The results suggest that MOMCs can be generated using leukocytes isolated from leukapheresis filters in the presence of SDF-1a. Furthermore, MOMCs expressed all the tested factors responsible to activate the networks of pluripotency of cells and can differentiate into endothelium and osteocytes. Therefore, the blood donors could benefit and be rewarded with the potential use of their own immune system cells for future treatment in the frame of personalized regenerative medicine.

Systems-Level Mapping of Cancer Testis Antigen 1b/a to Sarcoma Pathways Identifies Activated Ran Binding-2 E3 SUMO-Protein Ligase and Transducin-Like Enhancer Protein 1.

Here we describe the identification of genes and their encoded proteins that are expressed in advanced grade tumors by reconstruction of a sarcoma cancer testis gene 1b/a (catg1b/a) network. CTAG1B/A is an ortholog of the yeast/Drosophila transcription factor Pcc1p, and a member of the KEOPS transcription complex. It has been implicated in telomere maintenance and transcriptional regulation through association with chromatin remodeling factors and is only expressed during adult testis germ cell differentiation. Ctag1b/a is re-activated in synovial sarcomas and myxoid liposarcomas but not in differentiated liposarcomas. We mapped CTAG1B/A protein to sarcoma transcription pathways with gene set expression analysis (GSEA) and using independent samples, we immunohistochemically identified expression of at least two network neighbors, RANBP2, and TLE1, thus validating our approach. This work demonstrates that mapping unknown genes to functional pathways by network re-construction is a powerful tool that can be used to identify candidate oncoproteins.

Radiomics/Radiogenomics in Lung Cancer: Basic Principles and Initial Clinical Results.

Lung cancer is the leading cause of cancer-related deaths worldwide, and elucidation of its complicated pathobiology has been traditionally targeted by studies incorporating genomic as well other high-throughput approaches. Recently, a collection of methods used for cancer imaging, supplemented by quantitative aspects leading towards imaging biomarker assessment termed "radiomics", has introduced a novel dimension in cancer research. Integration of genomics and radiomics approaches, where identifying the biological basis of imaging phenotypes is feasible due to the establishment of associations between molecular features at the genomic-transcriptomic-proteomic level and radiological features, has recently emerged termed radiogenomics. This review article aims to briefly describe the main aspects of radiogenomics, while discussing its basic limitations related to lung cancer clinical applications for clinicians, researchers and patients.

Comparison of MRI Features of Fat Fraction and ADC for Early Treatment Response Assessment in Participants with Multiple Myeloma.

Background An imaging-based predictor of response could provide prognostic information early during treatment course in patients with multiple myeloma (MM). Purpose To investigate if very early changes in bone marrow relative fat fraction (rFF) and apparent diffusion coefficient (ADC) histogram metrics, occurring after one cycle of induction therapy in participants with newly diagnosed MM, could help predict overall best response status. Materials and Methods This prospective study included participants with MM who were enrolled between August 2014 and December 2017. Histogram metrics were extracted from ADC and rFF maps from MRI examinations performed before treatment and after the first treatment cycle. Participants were categorized into the very good partial response (VGPR) or better group and the less than VGPR group per the International Myeloma Working Group response criteria. ADC and rFF map metrics for predicting treatment response were compared using the Wilcoxon rank test, and the false discovery rate (FDR) was used to correct for multiple comparisons. Results A total of 23 participants (mean age, 65 years ± 11 [SD]; 13 men) were evaluated. There was no evidence of a difference in ADC metrics between the two responder groups after correcting for multiple comparisons. The rFF histogram changes between pretreatment MRI and MRI after the first treatment cycle (ΔrFF) that provided significant differences between the VGPR or better and less than VGPR groups were as follows: ΔrFF_10th Percentile (median, 0.5 [95% CI: 0, 1] vs -2.5 [95% CI: -5.1, 0.1], respectively), ΔrFF_90th Percentile (median, 2 [95% CI: 1, 6.8] vs -0.5 [95% CI: -1, 0]), ΔrFF_Mean (median, 3.4 [95% CI: 0.3, 7.6] vs -1.1 [95% CI: -1.8, -0.7]), and ΔrFF_Root Mean Squared (median, 3.2 [95% CI: 0.3, 6.1] vs -0.7 [95% CI: -1.3, -0.4]) (FDR-adjusted P = .03 for all), and the latter two also presented mean group increases in the VGPR or better group that were above the upper 95% CI limit for repeatability. Conclusion Very early changes in bone marrow relative fat fraction histogram metrics, calculated from MRI examination at baseline and after only one cycle of induction therapy, may help to predict very good partial response or better in participants with newly diagnosed multiple myeloma. © RSNA, 2022 Online supplemental material is available for this article.

Branch-duct intraductal papillary mucinous neoplasm (IPMN): Are cyst volumetry and other novel imaging features able to improve malignancy prediction compared to well-established resection criteria?

OBJECTIVES: Current guidelines base the management of intraductal papillary mucinous neoplasms (IPMN) on several well-established resection criteria (RC), including cyst size. However, malignancy may occur in small cysts. Since branch-duct (BD) IPMN are not perfect spheres, volumetric and morphologic analysis might better correlate with mucin production and grade of dysplasia. Nonetheless, their role in malignancy (high-grade dysplasia/invasive cancer) prediction has been poorly investigated. Previous studies evaluating RC also included patients with solid-mass-forming pancreatic cancer (PC), which may affect the RC yield. This study aimed to assess the role of volume, morphology, and other well-established RC in malignancy prediction in patients with BD- and mixed-type IPMN after excluding solid masses. METHODS: Retrospective ethical review-board-approved study of 106 patients (2008-2019) with histopathological diagnosis of BD- and mixed-type IPMN (without solid masses) and preoperative MRI available. Standard imaging and clinical features were collected, and the novel imaging features cyst-volume and elongation value [EV = 1 - (width/length)] calculated on T2-weighted images. Logistic regression analysis was performed. Statistical significance set at two-tails, p < 0.05. RESULTS: Neither volume (odds ratio (OR) = 1.01, 95% CI: 0.99-1.02, p = 0.12) nor EV (OR = 0.38, 95% CI: 0.02-5.93, p = 0.49) was associated with malignancy. Contrast-enhancing mural nodules (MN), main pancreatic duct (MPD) ≥ 5 mm, and elevated carbohydrate antigen (CA) 19-9 serum levels (> 37 µmol/L) were associated with malignancy (MN OR: 4.32, 95% CI: 1.18-15.76, p = 0.02; MPD ≥ 5 mm OR: 4.2, 95% CI: 1.34-13.1, p = 0.01; CA19-9 OR: 6.72; 95% CI: 1.89 - 23.89, p = 0.003). CONCLUSIONS: Volume and elongation value cannot predict malignancy in BD- and/or mixed-type IPMN. Mural nodules, MPD ≥ 5 mm and elevated CA19-9 serum levels are associated with higher malignancy risk even after the exclusion of solid masses. KEY POINTS: • Novel and well-established resection criteria for IPMN have been evaluated after excluding solid masses. • BD-IPMN volume and elongation value cannot predict malignancy. • Main pancreatic duct ≥ 5 mm, mural nodules, and elevated carbohydrate antigen 19-9 levels are associated with malignancy.

Radiomics in Oncology: A Practical Guide.

Radiomics refers to the extraction of mineable data from medical imaging and has been applied within oncology to improve diagnosis, prognostication, and clinical decision support, with the goal of delivering precision medicine. The authors provide a practical approach for successfully implementing a radiomic workflow from planning and conceptualization through manuscript writing. Applications in oncology typically are either classification tasks that involve computing the probability of a sample belonging to a category, such as benign versus malignant, or prediction of clinical events with a time-to-event analysis, such as overall survival. The radiomic workflow is multidisciplinary, involving radiologists and data and imaging scientists, and follows a stepwise process involving tumor segmentation, image preprocessing, feature extraction, model development, and validation. Images are curated and processed before segmentation, which can be performed on tumors, tumor subregions, or peritumoral zones. Extracted features typically describe the distribution of signal intensities and spatial relationship of pixels within a region of interest. To improve model performance and reduce overfitting, redundant and nonreproducible features are removed. Validation is essential to estimate model performance in new data and can be performed iteratively on samples of the dataset (cross-validation) or on a separate hold-out dataset by using internal or external data. A variety of noncommercial and commercial radiomic software applications can be used. Guidelines and artificial intelligence checklists are useful when planning and writing up radiomic studies. Although interest in the field continues to grow, radiologists should be familiar with potential pitfalls to ensure that meaningful conclusions can be drawn. Online supplemental material is available for this article. Published under a CC BY 4.0 license.

Improving performance and generalizability in radiogenomics: a pilot study for prediction of IDH1/2 mutation status in gliomas with multicentric data.

Purpose: Radiogenomics offers a potential virtual and noninvasive biopsy. However, radiogenomics models often suffer from generalizability issues, which cause a performance degradation on unseen data. In MRI, differences in the sequence parameters, manufacturers, and scanners make this generalizability issue worse. Such image acquisition information may be used to define different environments and select robust and invariant radiomic features associated with the clinical outcome that should be included in radiomics/radiogenomics models. Approach: We assessed 77 low-grade gliomas and glioblastomas multiform patients publicly available in TCGA and TCIA. Radiomics features were extracted from multiparametric MRI images (T1-weighted, contrast-enhanced T1-weighted, T2-weighted, and fluid-attenuated inversion recovery) and different regions-of-interest (enhancing tumor, nonenhancing tumor/necrosis, and edema). A method developed to find variables that are part of causal structures was used for feature selection and compared with an embedded feature selection approach commonly used in radiomics/radiogenomics studies, across two different scenarios: (1) leaving data from a center as an independent held-out test set and tuning the model with the data from the remaining centers and (2) use stratified partitioning to obtain the training and the held-out test sets. Results: In scenario (1), the performance of the proposed methodology and the traditional embedded method was AUC: 0.75 [0.25; 1.00] versus 0.83 [0.50; 1.00], Sens.: 0.67 [0.20; 0.93] versus 0.67 [0.20; 0.93], Spec.: 0.75 [0.30; 0.95] versus 0.75 [0.30; 0.95], and MCC: 0.42 [0.19; 0.68] versus 0.42 [0.19; 0.68] for center 1 as the held-out test set. The performance of both methods for center 2 as the held-out test set was AUC: 0.64 [0.36; 0.91] versus 0.55 [0.27; 0.82], Sens.: 0.00 [0.00; 0.73] versus 0.00 [0.00; 0.73], Spec.: 0.82 [0.52; 0.94] versus 0.91 [0.62; 0.98], and MCC: - 0.13 [ - 0.38 ; - 0.04 ] versus - 0.09 [ - 0.38 ; - 0.02 ] , whereas for center 3 was AUC: 0.80 [0.62; 0.95] versus 0.89 [0.56; 0.96], Sens.: 0.86 [0.48; 0.97] versus 0.86 [0.48; 0.97], Spec.: 0.72 [0.54; 0.85] versus 0.79 [0.61; 0.90], and MCC: 0.47 [0.41; 0.53] versus 0.55 [0.48; 0.60]. For center 4, the performance of both methods was AUC: 0.77 [0.51; 1.00] versus 0.75 [0.47; 0.97], Sens.: 0.53 [0.30; 0.75] versus 0.00 [0.00; 0.15], Spec.: 0.71 [0.35; 0.91] versus 0.86 [0.48; 0.97], and MCC: 0.23 [0.16; 0.31] versus. - 0.32 [ - 0.46 ; - 0.20 ] . In scenario (2), the performance of these methods was AUC: 0.89 [0.71; 1.00] versus 0.79 [0.58; 0.94], Sens.: 0.86 [0.80; 0.92] versus 0.43 [0.15; 0.74], Spec.: 0.87 [0.62; 0.96] versus 0.87 [0.62; 0.96], and MCC: 0.70 [0.60; 0.77] versus 0.33 [0.24; 0.42]. Conclusions: This proof-of-concept study demonstrated good performance by the proposed feature selection method in the majority of the studied scenarios, as it promotes robustness of features included in the models and the models' generalizability by making used imaging data of different scanners or with sequence parameters.