Imaging biomarkers and radiomics in pediatric oncology: a view from the PRIMAGE (PRedictive In silico Multiscale Analytics to support cancer personalized diaGnosis and prognosis, Empowered by imaging biomarkers) project.

This review paper presents the practical development of imaging biomarkers in the scope of the PRIMAGE (PRedictive In silico Multiscale Analytics to support cancer personalized diaGnosis and prognosis, Empowered by imaging biomarkers) project, as a noninvasive and reliable way to improve the diagnosis and prognosis in pediatric oncology. The PRIMAGE project is a European multi-center research initiative that focuses on developing medical imaging-derived artificial intelligence (AI) solutions designed to enhance overall management and decision-making for two types of pediatric cancer: neuroblastoma and diffuse intrinsic pontine glioma. To allow this, the PRIMAGE project has created an open-cloud platform that combines imaging, clinical, and molecular data together with AI models developed from this data, creating a comprehensive decision support environment for clinicians managing patients with these two cancers. In order to achieve this, a standardized data processing and analysis workflow was implemented to generate robust and reliable predictions for different clinical endpoints. Magnetic resonance (MR) image harmonization and registration was performed as part of the workflow. Subsequently, an automated tool for the detection and segmentation of tumors was trained and internally validated. The Dice similarity coefficient obtained for the independent validation dataset was 0.997, indicating compatibility with the manual segmentation variability. Following this, radiomics and deep features were extracted and correlated with clinical endpoints. Finally, reproducible and relevant imaging quantitative features were integrated with clinical and molecular data to enrich both the predictive models and a set of visual analytics tools, making the PRIMAGE platform a complete clinical decision aid system. In order to ensure the advancement of research in this field and to foster engagement with the wider research community, the PRIMAGE data repository and platform are currently being integrated into the European Federation for Cancer Images (EUCAIM), which is the largest European cancer imaging research infrastructure created to date.

Development and validation of an image biomarker to identify metabolic dysfunction associated steatohepatitis: MR-MASH score.

BACKGROUND AND AIMS: Diagnosis of metabolic dysfunction-associated steatohepatitis (MASH) requires histology. In this study, a magnetic resonance imaging (MRI) score was developed and validated to identify MASH in patients with metabolic dysfunction-associated steatotic liver disease (MASLD). Secondarily, a screening strategy for MASH diagnosis was investigated. METHODS: This prospective multicentre study included 317 patients with biopsy-proven MASLD and contemporaneous MRI. The discovery cohort (Spain, Portugal) included 194 patients. NAFLD activity score (NAS) and fibrosis were assessed with the NASH-CRN histologic system. MASH was defined by the presence of steatosis, lobular inflammation, and ballooning, with NAS ≥4 with or without fibrosis. An MRI-based composite biomarker of Proton Density Fat Fraction and waist circumference (MR-MASH score) was developed. Findings were afterwards validated in an independent cohort (United States, Spain) with different MRI protocols. RESULTS: In the derivation cohort, 51% (n = 99) had MASH. The MR-MASH score identified MASH with an AUC = .88 (95% CI .83-.93) and strongly correlated with NAS (r = .69). The MRI score lower cut-off corresponded to 88% sensitivity with 86% NPV, while the upper cut-off corresponded to 92% specificity with 87% PPV. MR-MASH was validated with an AUC = .86 (95% CI .77-.92), 91% sensitivity (lower cut-off) and 87% specificity (upper cut-off). A two-step screening strategy with sequential MR-MASH examination performed in patients with indeterminate-high FIB-4 or transient elastography showed an 83-84% PPV to identify MASH. The AUC of MR-MASH was significantly higher than that of the FAST score (p < .001). CONCLUSIONS: The MR-MASH score has clinical utility in the identification and management of patients with MASH at risk of progression.

Pancreatic steatosis and iron overload increases cardiovascular risk in non-alcoholic fatty liver disease.

Objective: To assess the prevalence of pancreatic steatosis and iron overload in non-alcoholic fatty liver disease (NAFLD) and their correlation with liver histology severity and the risk of cardiometabolic diseases. Method: A prospective, multicenter study including NAFLD patients with biopsy and paired Magnetic Resonance Imaging (MRI) was performed. Liver biopsies were evaluated according to NASH Clinical Research Network, hepatic iron storages were scored, and digital pathology quantified the tissue proportionate areas of fat and iron. MRI-biomarkers of fat fraction (PDFF) and iron accumulation (R2*) were obtained from the liver and pancreas. Different metabolic traits were evaluated, cardiovascular disease (CVD) risk was estimated with the atherosclerotic CVD score, and the severity of iron metabolism alteration was determined by grading metabolic hiperferritinemia (MHF). Associations between CVD, histology and MRI were investigated. Results: In total, 324 patients were included. MRI-determined pancreatic iron overload and moderate-to severe steatosis were present in 45% and 25%, respectively. Liver and pancreatic MRI-biomarkers showed a weak correlation (r=0.32 for PDFF, r=0.17 for R2*). Pancreatic PDFF increased with hepatic histologic steatosis grades and NASH diagnosis (p<0.001). Prevalence of pancreatic steatosis and iron overload increased with the number of metabolic traits (p<0.001). Liver R2* significantly correlated with MHF (AUC=0.77 [0.72-0.82]). MRI-determined pancreatic steatosis (OR=3.15 [1.63-6.09]), and iron overload (OR=2.39 [1.32-4.37]) were independently associated with high-risk CVD. Histologic diagnosis of NASH and advanced fibrosis were also associated with high-risk CVD. Conclusion: Pancreatic steatosis and iron overload could be of utility in clinical decision-making and prognostication of NAFLD.

Data infrastructures for AI in medical imaging: a report on the experiences of five EU projects.

Artificial intelligence (AI) is transforming the field of medical imaging and has the potential to bring medicine from the era of 'sick-care' to the era of healthcare and prevention. The development of AI requires access to large, complete, and harmonized real-world datasets, representative of the population, and disease diversity. However, to date, efforts are fragmented, based on single-institution, size-limited, and annotation-limited datasets. Available public datasets (e.g., The Cancer Imaging Archive, TCIA, USA) are limited in scope, making model generalizability really difficult. In this direction, five European Union projects are currently working on the development of big data infrastructures that will enable European, ethically and General Data Protection Regulation-compliant, quality-controlled, cancer-related, medical imaging platforms, in which both large-scale data and AI algorithms will coexist. The vision is to create sustainable AI cloud-based platforms for the development, implementation, verification, and validation of trustable, usable, and reliable AI models for addressing specific unmet needs regarding cancer care provision. In this paper, we present an overview of the development efforts highlighting challenges and approaches selected providing valuable feedback to future attempts in the area.Key points• Artificial intelligence models for health imaging require access to large amounts of harmonized imaging data and metadata.• Main infrastructures adopted either collect centrally anonymized data or enable access to pseudonymized distributed data.• Developing a common data model for storing all relevant information is a challenge.• Trust of data providers in data sharing initiatives is essential.• An online European Union meta-tool-repository is a necessity minimizing effort duplication for the various projects in the area.

Independent Validation of a Deep Learning nnU-Net Tool for Neuroblastoma Detection and Segmentation in MR Images.

OBJECTIVES: To externally validate and assess the accuracy of a previously trained fully automatic nnU-Net CNN algorithm to identify and segment primary neuroblastoma tumors in MR images in a large children cohort. METHODS: An international multicenter, multivendor imaging repository of patients with neuroblastic tumors was used to validate the performance of a trained Machine Learning (ML) tool to identify and delineate primary neuroblastoma tumors. The dataset was heterogeneous and completely independent from the one used to train and tune the model, consisting of 300 children with neuroblastic tumors having 535 MR T2-weighted sequences (486 sequences at diagnosis and 49 after finalization of the first phase of chemotherapy). The automatic segmentation algorithm was based on a nnU-Net architecture developed within the PRIMAGE project. For comparison, the segmentation masks were manually edited by an expert radiologist, and the time for the manual editing was recorded. Different overlaps and spatial metrics were calculated to compare both masks. RESULTS: The median Dice Similarity Coefficient (DSC) was high 0.997; 0.944-1.000 (median; Q1-Q3). In 18 MR sequences (6%), the net was not able neither to identify nor segment the tumor. No differences were found regarding the MR magnetic field, type of T2 sequence, or tumor location. No significant differences in the performance of the net were found in patients with an MR performed after chemotherapy. The time for visual inspection of the generated masks was 7.9 ± 7.5 (mean ± Standard Deviation (SD)) seconds. Those cases where manual editing was needed (136 masks) required 124 ± 120 s. CONCLUSIONS: The automatic CNN was able to locate and segment the primary tumor on the T2-weighted images in 94% of cases. There was an extremely high agreement between the automatic tool and the manually edited masks. This is the first study to validate an automatic segmentation model for neuroblastic tumor identification and segmentation with body MR images. The semi-automatic approach with minor manual editing of the deep learning segmentation increases the radiologist's confidence in the solution with a minor workload for the radiologist.

Automated prostate multi-regional segmentation in magnetic resonance using fully convolutional neural networks.

OBJECTIVE: Automatic MR imaging segmentation of the prostate provides relevant clinical benefits for prostate cancer evaluation such as calculation of automated PSA density and other critical imaging biomarkers. Further, automated T2-weighted image segmentation of central-transition zone (CZ-TZ), peripheral zone (PZ), and seminal vesicle (SV) can help to evaluate clinically significant cancer following the PI-RADS v2.1 guidelines. Therefore, the main objective of this work was to develop a robust and reproducible CNN-based automatic prostate multi-regional segmentation model using an intercontinental cohort of prostate MRI. METHODS: A heterogeneous database of 243 T2-weighted prostate studies from 7 countries and 10 machines of 3 different vendors, with the CZ-TZ, PZ, and SV regions manually delineated by two experienced radiologists (ground truth), was used to train (n = 123) and test (n = 120) a U-Net-based model with deep supervision using a cyclical learning rate. The performance of the model was evaluated by means of dice similarity coefficient (DSC), among others. Segmentation results with a DSC above 0.7 were considered accurate. RESULTS: The proposed method obtained a DSC of 0.88 ± 0.01, 0.85 ± 0.02, 0.72 ± 0.02, and 0.72 ± 0.02 for the prostate gland, CZ-TZ, PZ, and SV respectively in the 120 studies of the test set when comparing the predicted segmentations with the ground truth. No statistically significant differences were found in the results obtained between manufacturers or continents. CONCLUSION: Prostate multi-regional T2-weighted MR images automatic segmentation can be accurately achieved by U-Net like CNN, generalizable in a highly variable clinical environment with different equipment, acquisition configurations, and population. KEY POINTS: • Deep learning techniques allows the accurate segmentation of the prostate in three different regions on MR T2w images. • Multi-centric database proved the generalization of the CNN model on different institutions across different continents. • CNN models can be used to aid on the diagnosis and follow-up of patients with prostate cancer.

Machine Learning Models and Multiparametric Magnetic Resonance Imaging for the Prediction of Pathologic Response to Neoadjuvant Chemotherapy in Breast Cancer.

BACKGROUND: Most breast cancer (BC) patients fail to achieve pathological complete response (pCR) after neoadjuvant chemotherapy (NAC). The aim of this study was to evaluate whether imaging features (perfusion/diffusion imaging biomarkers + radiomic features) extracted from pre-treatment multiparametric (mp)MRIs were able to predict, alone or in combination with clinical data, pCR to NAC. METHODS: Patients with stage II-III BC receiving NAC and undergoing breast mpMRI were retrospectively evaluated. Imaging features were extracted from mpMRIs performed before NAC. Three different machine learning models based on imaging features, clinical data or imaging features + clinical data were trained to predict pCR. Confusion matrices and performance metrics were obtained to assess model performance. Statistical analyses were conducted to evaluate differences between responders and non-responders. RESULTS: Fifty-eight patients (median [range] age, 52 [45-58] years) were included, of whom 12 showed pCR. The combined model improved pCR prediction compared to clinical and imaging models, yielding 91.5% of accuracy with no false positive cases and only 17% false negative results. Changes in different parameters between responders and non-responders suggested a possible increase in vascularity and reduced tumour heterogeneity in patients with pCR, with the percentile 25th of time-to-peak (TTP), a classical perfusion parameter, being able to discriminate both groups in a 75% of the cases. CONCLUSIONS: A combination of mpMRI-derived imaging features and clinical variables was able to successfully predict pCR to NAC. Specific patient profiles according to tumour vascularity and heterogeneity might explain pCR differences, where TTP could emerge as a putative surrogate marker for pCR.

Automated Whole-Liver MRI Segmentation to Assess Steatosis and Iron Quantification in Chronic Liver Disease.

Background Standardized manual region of interest (ROI) sampling strategies for hepatic MRI steatosis and iron quantification are time consuming, with variable results. Purpose To evaluate the performance of automatic MRI whole-liver segmentation (WLS) for proton density fat fraction (PDFF) and iron estimation (transverse relaxometry [R2*]) versus manual ROI, with liver biopsy as the reference standard. Materials and Methods This prospective, cross-sectional, multicenter study recruited participants with chronic liver disease who underwent liver biopsy and chemical shift-encoded 3.0-T MRI between January 2017 and January 2021. Biopsy evaluation included histologic grading and digital pathology. MRI liver sampling strategies included manual ROI (two observers) and automatic whole-liver (deep learning algorithm) segmentation for PDFF- and R2*-derived measurements. Agreements between segmentation methods were measured using intraclass correlation coefficients (ICCs), and biases were evaluated using Bland-Altman analyses. Linear regression analyses were performed to determine the correlation between measurements and digital pathology. Results A total of 165 participants were included (mean age ± standard deviation, 55 years ± 12; 96 women; 101 of 165 participants [61%] with nonalcoholic fatty liver disease). Agreements between mean measurements were excellent, with ICCs of 0.98 for both PDFF and R2*. The median bias was 0.5% (interquartile range, -0.4% to 1.2%) for PDFF and 2.7 sec-1 (interquartile range, 0.2-5.3 sec-1) for R2* (P < .001 for both). Margins of error were lower for WLS than ROI-derived parameters (-0.03% for PDFF and -0.3 sec-1 for R2*). ROI and WLS showed similar performance for steatosis (ROI AUC, 0.96; WLS AUC, 0.97; P = .53) and iron overload (ROI AUC, 0.85; WLS AUC, 0.83; P = .09). Correlations with digital pathology were high (P < .001) between the fat ratio and PDFF (ROI r = 0.89; WLS r = 0.90) and moderate (P < .001) between the iron ratio and R2* (ROI r = 0.65; WLS r = 0.64). Conclusion Proton density fat fraction and transverse relaxometry measurements derived from MRI automatic whole-liver segmentation (WLS) were accurate for steatosis and iron grading in chronic liver disease and correlated with digital pathology. Automated WLS estimations were higher, with a lower margin of error than manual region of interest estimations. © RSNA, 2021 Online supplemental material is available for this article. See also the editorial by Moura Cunha and Fowler in this issue.

Radiomics and radiogenomics in head and neck squamous cell carcinoma: Potential contribution to patient management and challenges.

The application of imaging biomarkers in oncology is still in its infancy, but with the expansion of radiomics and radiogenomics a revolution is expected in this field. This may be of special interest in head and neck cancer, since it can promote precision medicine and personalization of treatment by overcoming several intrinsic obstacles in this pathology. Our goal is to provide the medical oncologist with the basis to approach these disciplines and appreciate their main uses in clinical research and clinical practice in the medium term. Aligned with this objective we analyzed the most relevant studies in the field, also highlighting novel opportunities and current challenges.

Precise whole liver automatic segmentation and quantification of PDFF and R2* on MR images.

OBJECTIVE: To automate the segmentation of whole liver parenchyma on multi-echo chemical shift encoded (MECSE) MR examinations using convolutional neural networks (CNNs) to seamlessly quantify precise organ-related imaging biomarkers such as the fat fraction and iron load. METHODS: A retrospective multicenter collection of 183 MECSE liver MR examinations was conducted. An encoder-decoder CNN was trained (107 studies) following a 5-fold cross-validation strategy to improve the model performance and ensure lack of overfitting. Proton density fat fraction (PDFF) and R2* were quantified on both manual and CNN segmentation masks. Different metrics were used to evaluate the CNN performance over both unseen internal (46 studies) and external (29 studies) validation datasets to analyze reproducibility. RESULTS: The internal test showed excellent results for the automatic segmentation with a dice coefficient (DC) of 0.93 ± 0.03 and high correlation between the quantification done with the predicted mask and the manual segmentation (rPDFF = 1 and rR2* = 1; p values < 0.001). The external validation was also excellent with a different vendor but the same magnetic field strength, proving the generalization of the model to other manufacturers with DC of 0.94 ± 0.02. Results were lower for the 1.5-T MR same vendor scanner with DC of 0.87 ± 0.06. Both external validations showed high correlation in the quantification (rPDFF = 1 and rR2* = 1; p values < 0.001). In both internal and external validation datasets, the relative error for the PDFF and R2* quantification was below 4% and 1% respectively. CONCLUSION: Liver parenchyma can be accurately segmented with CNN in a vendor-neutral virtual approach, allowing to obtain reproducible automatic whole organ virtual biopsies. KEY POINTS: • Whole liver parenchyma can be automatically segmented using convolutional neural networks. • Deep learning allows the creation of automatic pipelines for the precise quantification of liver-related imaging biomarkers such as PDFF and R2*. • MR "virtual biopsy" can become a fast and automatic procedure for the assessment of chronic diffuse liver diseases in clinical practice.

The Aging Imageomics Study: rationale, design and baseline characteristics of the study population.

Biomarkers of aging are urgently needed to identify individuals at high risk of developing age-associated disease or disability. Growing evidence from population-based studies points to whole-body magnetic resonance imaging's (MRI) enormous potential for quantifying subclinical disease burden and for assessing changes that occur with aging in all organ systems. The Aging Imageomics Study aims to identify biomarkers of human aging by analyzing imaging, biopsychosocial, cardiovascular, metabolomic, lipidomic, and microbiome variables. This study recruited 1030 participants aged ≥50 years (mean 67, range 50-96 years) that underwent structural and functional MRI to evaluate the brain, large blood vessels, heart, abdominal organs, fat, spine, musculoskeletal system and ultrasonography to assess carotid intima-media thickness and plaques. Patients were notified of incidental findings detected by a certified radiologist when necessary. Extensive data were also collected on anthropometrics, demographics, health history, neuropsychology, employment, income, family status, exposure to air pollution and cardiovascular status. In addition, several types of samples were gathered to allow for microbiome, metabolomic and lipidomic profiling. Using big data techniques to analyze all the data points from biological phenotyping together with health records and lifestyle measures, we aim to cultivate a deeper understanding about various biological factors (and combinations thereof) that underlie healthy and unhealthy aging.

PRIMAGE project: predictive in silico multiscale analytics to support childhood cancer personalised evaluation empowered by imaging biomarkers.

PRIMAGE is one of the largest and more ambitious research projects dealing with medical imaging, artificial intelligence and cancer treatment in children. It is a 4-year European Commission-financed project that has 16 European partners in the consortium, including the European Society for Paediatric Oncology, two imaging biobanks, and three prominent European paediatric oncology units. The project is constructed as an observational in silico study involving high-quality anonymised datasets (imaging, clinical, molecular, and genetics) for the training and validation of machine learning and multiscale algorithms. The open cloud-based platform will offer precise clinical assistance for phenotyping (diagnosis), treatment allocation (prediction), and patient endpoints (prognosis), based on the use of imaging biomarkers, tumour growth simulation, advanced visualisation of confidence scores, and machine-learning approaches. The decision support prototype will be constructed and validated on two paediatric cancers: neuroblastoma and diffuse intrinsic pontine glioma. External validation will be performed on data recruited from independent collaborative centres. Final results will be available for the scientific community at the end of the project, and ready for translation to other malignant solid tumours.

Automated vertebrae localization and identification by decision forests and image-based refinement on real-world CT data.

PURPOSE: Development of a fully automatic algorithm for the automatic localization and identification of vertebral bodies in computed tomography (CT). MATERIALS AND METHODS: This algorithm was developed using a dataset based on real-world data of 232 thoraco-abdominopelvic CT scans retrospectively collected. In order to achieve an accurate solution, a two-stage automated method was developed: decision forests for a rough prediction of vertebral bodies position, and morphological image processing techniques to refine the previous detection by locating the position of the spinal canal. RESULTS: The mean distance error between the predicted vertebrae centroid position and truth was 13.7 mm. The identification rate was 79.6% on the thoracic region and of 74.8% on the lumbar segment. CONCLUSION: The algorithm provides a new method to detect and identify vertebral bodies from arbitrary field-of-view body CT scans.

Macrovascular Networks on Contrast-Enhanced Magnetic Resonance Imaging Improves Survival Prediction in Newly Diagnosed Glioblastoma.

A higher degree of angiogenesis is associated with shortened survival in glioblastoma. Feasible morphometric parameters for analyzing vascular networks in brain tumors in clinical practice are lacking. We investigated whether the macrovascular network classified by the number of vessel-like structures (nVS) visible on three-dimensional T1-weighted contrast⁻enhanced (3D-T1CE) magnetic resonance imaging (MRI) could improve survival prediction models for newly diagnosed glioblastoma based on clinical and other imaging features. Ninety-seven consecutive patients (62 men; mean age, 58 ± 15 years) with histologically proven glioblastoma underwent 1.5T-MRI, including anatomical, diffusion-weighted, dynamic susceptibility contrast perfusion, and 3D-T1CE sequences after 0.1 mmol/kg gadobutrol. We assessed nVS related to the tumor on 1-mm isovoxel 3D-T1CE images, and relative cerebral blood volume, relative cerebral flow volume (rCBF), delay mean time, and apparent diffusion coefficient in volumes of interest for contrast-enhancing lesion (CEL), non-CEL, and contralateral normal-appearing white matter. We also assessed Visually Accessible Rembrandt Images scoring system features. We used ROC curves to determine the cutoff for nVS and univariate and multivariate cox proportional hazards regression for overall survival. Prognostic factors were evaluated by Kaplan-Meier survival and ROC analyses. Lesions with nVS > 5 were classified as having highly developed macrovascular network; 58 (60.4%) tumors had highly developed macrovascular network. Patients with highly developed macrovascular network were older, had higher volumeCEL, increased rCBFCEL, and poor survival; nVS correlated negatively with survival (r = -0.286; p = 0.008). On multivariate analysis, standard treatment, age at diagnosis, and macrovascular network best predicted survival at 1 year (AUC 0.901, 83.3% sensitivity, 93.3% specificity, 96.2% PPV, 73.7% NPV). Contrast-enhanced MRI macrovascular network improves survival prediction in newly diagnosed glioblastoma.