Training and external validation of pre-treatment FDG PET-CT-based models for outcome prediction in anal squamous cell carcinoma.

OBJECTIVES: The incidence of anal squamous cell carcinoma (ASCC) is increasing worldwide, with a significant proportion of patients treated with curative intent having recurrence. The ability to accurately predict progression-free survival (PFS) and overall survival (OS) would allow for development of personalised treatment strategies. The aim of the study was to train and external test radiomic/clinical feature derived time-to-event prediction models. METHODS: Consecutive patients with ASCC treated with curative intent at two large tertiary referral centres with baseline FDG PET-CT were included. Radiomic feature extraction was performed using LIFEx software on the pre-treatment PET-CT. Two distinct predictive models for PFS and OS were trained and tuned at each of the centres, with the best performing models externally tested on the other centres' patient cohort. RESULTS: A total of 187 patients were included from centre 1 (mean age 61.6 ± 11.5 years, median follow up 30 months, PFS events = 57/187, OS events = 46/187) and 257 patients were included from centre 2 (mean age 62.6 ± 12.3 years, median follow up 35 months, PFS events = 70/257, OS events = 54/257). The best performing model for PFS and OS was achieved using a Cox regression model based on age and metabolic tumour volume (MTV) with a training c-index of 0.7 and an external testing c-index of 0.7 (standard error = 0.4). CONCLUSIONS: A combination of patient age and MTV has been demonstrated using external validation to have the potential to predict OS and PFS in ASCC patients. CLINICAL RELEVANCE STATEMENT: A Cox regression model using patients' age and metabolic tumour volume showed good predictive potential for progression-free survival in external testing. The benefits of a previous radiomics model published by our group could not be confirmed on external testing. KEY POINTS: • A predictive model based on patient age and metabolic tumour volume showed potential to predict overall survival and progression-free survival and was validated on an external test cohort. • The methodology used to create a predictive model from age and metabolic tumour volume was repeatable using external cohort data. • The predictive ability of positron emission tomography-computed tomography-derived radiomic features diminished when the influence of metabolic tumour volume was accounted for.

Prediction of prostate tumour hypoxia using pre-treatment MRI-derived radiomics: preliminary findings.

PURPOSE: To develop a machine learning (ML) model based on radiomic features (RF) extracted from whole prostate gland magnetic resonance imaging (MRI) for prediction of tumour hypoxia pre-radiotherapy. MATERIAL AND METHODS: Consecutive patients with high-grade prostate cancer and pre-treatment MRI treated with radiotherapy between 01/12/2007 and 1/08/2013 at two cancer centres were included. Cancers were dichotomised as normoxic or hypoxic using a biopsy-based 32-gene hypoxia signature (Ragnum signature). Prostate segmentation was performed on axial T2-weighted (T2w) sequences using RayStation (v9.1). Histogram standardisation was applied prior to RF extraction. PyRadiomics (v3.0.1) was used to extract RFs for analysis. The cohort was split 80:20 into training and test sets. Six different ML classifiers for distinguishing hypoxia were trained and tuned using five different feature selection models and fivefold cross-validation with 20 repeats. The model with the highest mean validation area under the curve (AUC) receiver operating characteristic (ROC) curve was tested on the unseen set, and AUCs were compared via DeLong test with 95% confidence interval (CI). RESULTS: 195 patients were included with 97 (49.7%) having hypoxic tumours. The hypoxia prediction model with best performance was derived using ridge regression and had a test AUC of 0.69 (95% CI: 0.14). The test AUC for the clinical-only model was lower (0.57), but this was not statistically significant (p = 0.35). The five selected RFs included textural and wavelet-transformed features. CONCLUSION: Whole prostate MRI-radiomics has the potential to non-invasively predict tumour hypoxia prior to radiotherapy which may be helpful for individualised treatment optimisation.

Utility of pre-treatment FDG PET/CT-derived machine learning models for outcome prediction in classical Hodgkin lymphoma.

OBJECTIVES: Relapse occurs in ~20% of patients with classical Hodgkin lymphoma (cHL) despite treatment adaption based on 2-deoxy-2-[18F]fluoro-D-glucose positron emission tomography/computed tomography response. The objective was to evaluate pre-treatment FDG PET/CT-derived machine learning (ML) models for predicting outcome in patients with cHL. METHODS: All cHL patients undergoing pre-treatment PET/CT at our institution between 2008 and 2018 were retrospectively identified. A 1.5 × mean liver standardised uptake value (SUV) and a fixed 4.0 SUV threshold were used to segment PET/CT data. Feature extraction was performed using PyRadiomics with ComBat harmonisation. Training (80%) and test (20%) cohorts stratified around 2-year event-free survival (EFS), age, sex, ethnicity and disease stage were defined. Seven ML models were trained and hyperparameters tuned using stratified 5-fold cross-validation. Area under the curve (AUC) from receiver operator characteristic analysis was used to assess performance. RESULTS: A total of 289 patients (153 males), median age 36 (range 16-88 years), were included. There was no significant difference between training (n = 231) and test cohorts (n = 58) (p value > 0.05). A ridge regression model using a 1.5 × mean liver SUV segmentation had the highest performance, with mean training, validation and test AUCs of 0.82 ± 0.002, 0.79 ± 0.01 and 0.81 ± 0.12. However, there was no significant difference between a logistic model derived from metabolic tumour volume and clinical features or the highest performing radiomic model. CONCLUSIONS: Outcome prediction using pre-treatment FDG PET/CT-derived ML models is feasible in cHL patients. Further work is needed to determine optimum predictive thresholds for clinical use. KEY POINTS: • A fixed threshold segmentation method led to more robust radiomic features. • A radiomic-based model for predicting 2-year event-free survival in classical Hodgkin lymphoma patients is feasible. • A predictive model based on ridge regression was the best performing model on our dataset.

A methodological framework for AI-assisted diagnosis of active aortitis using radiomic analysis of FDG PET-CT images: Initial analysis.

BACKGROUND: The aim of this study was to explore the feasibility of assisted diagnosis of active (peri-)aortitis using radiomic imaging biomarkers derived from [18F]-Fluorodeoxyglucose Positron Emission Tomography-Computed Tomography (FDG PET-CT) images. METHODS: The aorta was manually segmented on FDG PET-CT in 50 patients with aortitis and 25 controls. Radiomic features (RF) (n = 107), including SUV (Standardized Uptake Value) metrics, were extracted from the segmented data and harmonized using the ComBat technique. Individual RFs and groups of RFs (i.e., signatures) were used as input in Machine Learning classifiers. The diagnostic utility of these classifiers was evaluated with area under the receiver operating characteristic curve (AUC) and accuracy using the clinical diagnosis as the ground truth. RESULTS: Several RFs had high accuracy, 84% to 86%, and AUC scores 0.83 to 0.97 when used individually. Radiomic signatures performed similarly, AUC 0.80 to 1.00. CONCLUSION: A methodological framework for a radiomic-based approach to support diagnosis of aortitis was outlined. Selected RFs, individually or in combination, showed similar performance to the current standard of qualitative assessment in terms of AUC for identifying active aortitis. This framework could support development of a clinical decision-making tool for a more objective and standardized assessment of aortitis.

Can MR textural analysis improve the prediction of extracapsular nodal spread in patients with oral cavity cancer?

OBJECTIVE: To explore the utility of MR texture analysis (MRTA) for detection of nodal extracapsular spread (ECS) in oral cavity squamous cell carcinoma (SCC). METHODS: 115 patients with oral cavity SCC treated with surgery and adjuvant (chemo)radiotherapy were identified retrospectively. First-order texture parameters (entropy, skewness and kurtosis) were extracted from tumour and nodal regions of interest (ROIs) using proprietary software (TexRAD). Nodal MR features associated with ECS (flare sign, irregular capsular contour; local infiltration; nodal necrosis) were reviewed and agreed in consensus by two experienced radiologists. Diagnostic performance characteristics of MR features of ECS were compared with primary tumour and nodal MRTA prediction using histology as the gold standard. Receiver operating characteristic (ROC) and regression analyses were also performed. RESULTS: Nodal entropy derived from contrast-enhanced T1-weighted images was significant in predicting ECS (p = 0.018). MR features had varying accuracy: flare sign (70%); irregular contour (71%); local infiltration (66%); and nodal necrosis (64%). Nodal entropy combined with irregular contour was the best predictor of ECS (p = 0.004, accuracy 79%). CONCLUSION: First-order nodal MRTA combined with imaging features may improve ECS prediction in oral cavity SCC. KEY POINTS: • Nodal MR textural analysis can aid in predicting extracapsular spread (ECS). • Medium filter contrast-enhanced T1 nodal entropy was strongly significant in predicting ECS. • Combining nodal entropy with irregular nodal contour improves predictive accuracy.

A Scoping Review of Machine-Learning Derived Radiomic Analysis of CT and PET Imaging to Investigate Atherosclerotic Cardiovascular Disease.

BACKGROUND: Cardiovascular disease affects the carotid arteries, coronary arteries, aorta and the peripheral arteries. Radiomics involves the extraction of quantitative data from imaging features that are imperceptible to the eye. Radiomics analysis in cardiovascular disease has largely focused on CT and MRI modalities. This scoping review aims to summarise the existing literature on radiomic analysis techniques in cardiovascular disease. METHODS: MEDLINE and Embase databases were searched for eligible studies evaluating radiomic techniques in living human subjects derived from CT, MRI or PET imaging investigating atherosclerotic disease. Data on study population, imaging characteristics and radiomics methodology were extracted. RESULTS: Twenty-nine studies consisting of 5753 patients (3752 males) were identified, and 78.7% of patients were from coronary artery studies. Twenty-seven studies employed CT imaging (19 CT carotid angiography and 6 CT coronary angiography (CTCA)), and two studies studied PET/CT. Manual segmentation was most frequently undertaken. Processing techniques included voxel discretisation, voxel resampling and filtration. Various shape, first-order, second-order and higher-order radiomic features were extracted. Logistic regression was most commonly used for machine learning. CONCLUSION: Most published evidence was feasibility/proof of concept work. There was significant heterogeneity in image acquisition, segmentation techniques, processing and analysis between studies. There is a need for the implementation of standardised imaging acquisition protocols, adherence to published reporting guidelines and economic evaluation.

Tumour Size and Overall Survival in a Cohort of Patients with Unifocal Glioblastoma: A Uni- and Multivariable Prognostic Modelling and Resampling Study.

Published models inconsistently associate glioblastoma size with overall survival (OS). This study aimed to investigate the prognostic effect of tumour size in a large cohort of patients diagnosed with GBM and interrogate how sample size and non-linear transformations may impact on the likelihood of finding a prognostic effect. In total, 279 patients with a IDH-wildtype unifocal WHO grade 4 GBM between 2014 and 2020 from a retrospective cohort were included. Uni-/multivariable association between core volume, whole volume (CV and WV), and diameter with OS was assessed with (1) Cox proportional hazard models +/- log transformation and (2) resampling with 1,000,000 repetitions and varying sample size to identify the percentage of models, which showed a significant effect of tumour size. Models adjusted for operation type and a diameter model adjusted for all clinical variables remained significant (p = 0.03). Multivariable resampling increased the significant effects (p < 0.05) of all size variables as sample size increased. Log transformation also had a large effect on the chances of a prognostic effect of WV. For models adjusted for operation type, 19.5% of WV vs. 26.3% log-WV (n = 50) and 69.9% WV and 89.9% log-WV (n = 279) were significant. In this large well-curated cohort, multivariable modelling and resampling suggest tumour volume is prognostic at larger sample sizes and with log transformation for WV.

Automated extraction of body composition metrics from abdominal CT or MR imaging: A scoping review.

PURPOSE: To review methodological approaches for automated segmentation of subcutaneous adipose tissue, visceral adipose tissue, and skeletal muscle from abdominal cross-sectional imaging for body composition analysis. METHOD: Four databases were searched for publications describing automated segmentation of subcutaneous adipose tissue, visceral adipose tissue, and/or skeletal muscle from abdominal CT or MR imaging between 2019 and 2023. Included reports were evaluated to assess how imaging modality, cohort size, vertebral level, model dimensionality, and use of a volume or single slice affected segmentation accuracy and/or clinical utility. Exclusion criteria included reports not in English language, manual or semi-automated segmentation methods, reports prior to 2019 or solely of paediatric patients, and those not describing the use of abdominal CT or MR. RESULTS: After exclusions, 172 reports were included in the review. CT imaging was utilised approximately four times as often as MRI, and segmentation accuracy did not significantly differ between the two modalities. Cohort size had no significant effect on segmentation accuracy. There was little evidence to refute the current practice of extracting body composition metrics from the third lumbar vertebral level. There was no clear benefit of using a 3D model to perform segmentation over a 2D approach. CONCLUSION: Automated segmentation of intra-abdominal soft tissues for body composition analysis is an intense area of research activity. Segmentation accuracy is not affected by cross-sectional imaging modality. Extracting metrics from a single slice at the third lumbar vertebral level is a common approach, however, extracting metrics from a volumetric slab surrounding this level may increase the resilience of the technique, which is important for clinical translation. A paucity of publicly available datasets led to most reports using different data sources, preventing direct comparison of segmentation techniques. Future efforts should prioritise creating a standardised dataset to facilitate benchmarking of different algorithms and subsequent clinical adoption.

Prediction of Patient Outcomes in Locally Advanced Cervical Carcinoma Following Chemoradiotherapy-Comparative Effectiveness of Magnetic Resonance Imaging and 2-Deoxy-2-[18F]fluoro-D-glucose Imaging.

PURPOSE: To evaluate the utility and comparative effectiveness of three five-point qualitative scoring systems for assessing response on PET-CT and MRI imaging individually and in combination, following curative-intent chemoradiotherapy (CRT) in locally advanced cervical cancer (LACC). Their performance in the prediction of subsequent patient outcomes was also assessed; Methods: Ninety-seven patients with histologically confirmed LACC treated with CRT using standard institutional protocols at a single centre who underwent PET-CT and MRI at staging and post treatment were identified retrospectively from an institutional database. The post-CRT imaging studies were independently reviewed, and response assessed using five-point scoring tools for T2WI, DWI, and FDG PET-CT. Patient characteristics, staging, treatment, and follow-up details including progression-free survival (PFS) and overall survival (OS) outcomes were collected. To compare diagnostic performance metrics, a two-proportion z-test was employed. A Kaplan-Meier analysis (Mantel-Cox log-rank) was performed. RESULTS: The T2WI (p < 0.00001, p < 0.00001) and DWI response scores (p < 0.00001, p = 0.0002) had higher specificity and accuracy than the PET-CT. The T2WI score had the highest positive predictive value (PPV), while the negative predictive value (NPV) was consistent across modalities. The combined MR scores maintained high NPV, PPV, specificity, and sensitivity, and the PET/MR consensus scores showed superior diagnostic accuracy and specificity compared to the PET-CT score alone (p = 0.02926, p = 0.0083). The Kaplan-Meier analysis revealed significant differences in the PFS based on the T2WI (p < 0.001), DWI (p < 0.001), combined MR (p = 0.003), and PET-CT/MR consensus scores (p < 0.001) and in the OS for the T2WI (p < 0.001), DWI (p < 0.001), and combined MR scores (p = 0.031) between responders and non-responders. CONCLUSION: Post-CRT response assessment using qualitative MR scoring and/or consensus PET-CT and MRI scoring was a better predictor of outcome compared to PET-CT assessment alone. This requires validation in a larger prospective study but offers the potential to help stratify patient follow-up in the future.

Can Patient Characteristics and Pre-Treatment MRI Features Predict Survival After Stereotactic Ablative Radiotherapy (SABR) Treatment in Hepatocellular Carcinoma (HCC): Preliminary Assessment.

BACKGROUND: The study purpose was to develop a machine learning (ML)-based predictive model for event-free survival (EFS) in patients with hepatocellular carcinoma (HCC) undergoing stereotactic ablative radiotherapy (SABR). METHODS: Patients receiving SABR for HCC at a single institution, between 2017 and 2020, were included in the study. They were split into training and test (85%:15%) cohorts. Events of interest were HCC recurrence or death. Three ML models were trained, the features were selected, and the hyperparameters were tuned. The performance was measured using Harrell's C index with the best-performing model being tested on the unseen cohort. RESULTS: Overall, 41 patients were included (training = 34, test = 7) and 64 lesions were analysed (training = 50, test = 14), resulting in 30 events (60% rate) in the training set (death = 6, recurrence = 24) and 8 events (57% rate) in the test set (death = 5, recurrence = 3). A Cox regression model, using age at treatment, albumin, and intra-lesional fat identified through MRI as variables, had the best performance with a mean training score of 0.78 (standard deviation (SD) 0.02), a mean validation of 0.78 (SD 0.18), and a test score of 0.94. CONCLUSIONS: Predicting the outcomes in patients with HCC, following SABR, using a novel model is feasible and warrants further evaluation.

Glioblastoma and radiotherapy: A multicenter AI study for Survival Predictions from MRI (GRASP study).

BACKGROUND: The aim was to predict survival of glioblastoma at 8 months after radiotherapy (a period allowing for completing a typical course of adjuvant temozolomide), by applying deep learning to the first brain MRI after radiotherapy completion. METHODS: Retrospective and prospective data were collected from 206 consecutive glioblastoma, isocitrate dehydrogenase -wildtype patients diagnosed between March 2014 and February 2022 across 11 UK centers. Models were trained on 158 retrospective patients from 3 centers. Holdout test sets were retrospective (n = 19; internal validation), and prospective (n = 29; external validation from 8 distinct centers). Neural network branches for T2-weighted and contrast-enhanced T1-weighted inputs were concatenated to predict survival. A nonimaging branch (demographics/MGMT/treatment data) was also combined with the imaging model. We investigated the influence of individual MR sequences; nonimaging features; and weighted dense blocks pretrained for abnormality detection. RESULTS: The imaging model outperformed the nonimaging model in all test sets (area under the receiver-operating characteristic curve, AUC P = .038) and performed similarly to a combined imaging/nonimaging model (P > .05). Imaging, nonimaging, and combined models applied to amalgamated test sets gave AUCs of 0.93, 0.79, and 0.91. Initializing the imaging model with pretrained weights from 10 000s of brain MRIs improved performance considerably (amalgamated test sets without pretraining 0.64; P = .003). CONCLUSIONS: A deep learning model using MRI images after radiotherapy reliably and accurately determined survival of glioblastoma. The model serves as a prognostic biomarker identifying patients who will not survive beyond a typical course of adjuvant temozolomide, thereby stratifying patients into those who might require early second-line or clinical trial treatment.

Clinical Utility of Second-Look FDG PET-CT to Stratify Incomplete Metabolic Response Post (Chemo) Radiotherapy in Oropharyngeal Squamous Cell Carcinoma.

BACKGROUND: Incomplete response on FDG PET-CT following (chemo)radiotherapy (CRT) for head and neck squamous cell carcinoma (HNSCC) hinders optimal management. The study assessed the utility of an interval (second look) PET-CT. METHODS: Patients with oropharyngeal squamous cell carcinoma cancer (OPSCC) treated with CRT at a single centre between 2013 and 2020 who underwent baseline, response, and second-look PET-CT were included. Endpoints were conversion rate to complete metabolic response (CMR) and test characteristics of second-look PET-CT. RESULTS: In total, 714 patients with OPSCC underwent PET-CT post-radiotherapy. In total, 88 patients with incomplete response underwent second-look PET-CT a median of 13 weeks (interquartile range 10-15 weeks) after the initial response assessment. In total, 27/88 (31%) second-look PET-CTs showed conversion to CMR, primary tumour CMR in 20/60 (30%), and nodal CMR in 13/37 (35%). In total, 1/34 (3%) with stable tumour/nodal uptake at the second-look PET-CT relapsed. Sensitivity, specificity, positive (PPV), and negative predictive value (NPV) of second-look PET-CT were 95%, 49%, 50%, and 95% for tumour and 92%, 50%, 50%, and 92% for nodes, respectively. Primary tumour progression following CMR occurred in one patient, two patients with residual nodal uptake at second-look PET-CT progressed locoregionally, and one patient developed metastatic disease following CMR in residual nodes. CONCLUSION: Most patients undergoing second-look PET-CT converted to CMR or demonstrated stable PET signal. NPV was high, suggesting the potential to avoid unnecessary surgical intervention.

Imaging Spectrum of the Developing Glioblastoma: A Cross-Sectional Observation Study.

Glioblastoma (GBM) has the typical radiological appearance (TRA) of a centrally necrotic, peripherally enhancing tumor with surrounding edema. The objective of this study was to determine whether the developing GBM displays a spectrum of imaging changes detectable on routine clinical imaging prior to TRA GBM. Patients with pre-operative imaging diagnosed with GBM (1 January 2014-31 March 2022) were identified from a neuroscience center. The imaging was reviewed by an experienced neuroradiologist. Imaging patterns preceding TRA GBM were analyzed. A total of 76 out of 555 (14%) patients had imaging preceding TRA GBM, 57 had solitary lesions, and 19 had multiple lesions (total = 84 lesions). Here, 83% of the lesions had cortical or cortical/subcortical locations. The earliest imaging features for 84 lesions were T2 hyperintensity/CT low density (n = 18), CT hyperdensity (n = 51), and T2 iso-intensity (n = 15). Lesions initially showing T2 hyperintensity/CT low density later showed T2 iso-intensity. When CT and MRI were available, all CT hyperdense lesions showed T2 iso-intensity, reduced diffusivity, and the following enhancement patterns: nodular 35%, solid 29%, none 26%, and patchy peripheral 10%. The mean time to develop TRA GBM from T2 hyperintensity was 140 days and from CT hyperdensity was 69 days. This research suggests that the developing GBM shows a spectrum of imaging features, progressing through T2 hyperintensity to CT hyperdensity, T2 iso-intensity, reduced diffusivity, and variable enhancement to TRA GBM. Red flags for non-TRA GBM lesions are cortical/subcortical CT hyperdense/T2 iso-intense/low ADC. Future research correlating this imaging spectrum with pathophysiology may provide insight into GBM growth patterns.

Comparative effectiveness of standard vs. AI-assisted PET/CT reading workflow for pre-treatment lymphoma staging: a multi-institutional reader study evaluation.

Background: Fluorine-18 fluorodeoxyglucose (FDG)-positron emission tomography/computed tomography (PET/CT) is widely used for staging high-grade lymphoma, with the time to evaluate such studies varying depending on the complexity of the case. Integrating artificial intelligence (AI) within the reporting workflow has the potential to improve quality and efficiency. The aims of the present study were to evaluate the influence of an integrated research prototype segmentation tool implemented within diagnostic PET/CT reading software on the speed and quality of reporting with variable levels of experience, and to assess the effect of the AI-assisted workflow on reader confidence and whether this tool influenced reporting behaviour. Methods: Nine blinded reporters (three trainees, three junior consultants and three senior consultants) from three UK centres participated in a two-part reader study. A total of 15 lymphoma staging PET/CT scans were evaluated twice: first, using a standard PET/CT reporting workflow; then, after a 6-week gap, with AI assistance incorporating pre-segmentation of disease sites within the reading software. An even split of PET/CT segmentations with gold standard (GS), false-positive (FP) over-contour or false-negative (FN) under-contour were provided. The read duration was calculated using file logs, while the report quality was independently assessed by two radiologists with >15 years of experience. Confidence in AI assistance and identification of disease was assessed via online questionnaires for each case. Results: There was a significant decrease in time between non-AI and AI-assisted reads (median 15.0 vs. 13.3 min, p < 0.001). Sub-analysis confirmed this was true for both junior (14.5 vs. 12.7 min, p = 0.03) and senior consultants (15.1 vs. 12.2 min, p = 0.03) but not for trainees (18.1 vs. 18.0 min, p = 0.2). There was no significant difference between report quality between reads. AI assistance provided a significant increase in confidence of disease identification (p < 0.001). This held true when splitting the data into FN, GS and FP. In 19/88 cases, participants did not identify either FP (31.8%) or FN (11.4%) segmentations. This was significantly greater for trainees (13/30, 43.3%) than for junior (3/28, 10.7%, p = 0.05) and senior consultants (3/30, 10.0%, p = 0.05). Conclusions: The study findings indicate that an AI-assisted workflow achieves comparable performance to humans, demonstrating a marginal enhancement in reporting speed. Less experienced readers were more influenced by segmentation errors. An AI-assisted PET/CT reading workflow has the potential to increase reporting efficiency without adversely affecting quality, which could reduce costs and report turnaround times. These preliminary findings need to be confirmed in larger studies.

An Automated Method for Artifical Intelligence Assisted Diagnosis of Active Aortitis Using Radiomic Analysis of FDG PET-CT Images.

The aim of this study was to develop and validate an automated pipeline that could assist the diagnosis of active aortitis using radiomic imaging biomarkers derived from [18F]-Fluorodeoxyglucose Positron Emission Tomography-Computed Tomography (FDG PET-CT) images. The aorta was automatically segmented by convolutional neural network (CNN) on FDG PET-CT of aortitis and control patients. The FDG PET-CT dataset was split into training (43 aortitis:21 control), test (12 aortitis:5 control) and validation (24 aortitis:14 control) cohorts. Radiomic features (RF), including SUV metrics, were extracted from the segmented data and harmonized. Three radiomic fingerprints were constructed: A-RFs with high diagnostic utility removing highly correlated RFs; B used principal component analysis (PCA); C-Random Forest intrinsic feature selection. The diagnostic utility was evaluated with accuracy and area under the receiver operating characteristic curve (AUC). Several RFs and Fingerprints had high AUC values (AUC > 0.8), confirmed by balanced accuracy, across training, test and external validation datasets. Good diagnostic performance achieved across several multi-centre datasets suggests that a radiomic pipeline can be generalizable. These findings could be used to build an automated clinical decision tool to facilitate objective and standardized assessment regardless of observer experience.

Harmonisation of scanner-dependent contrast variations in magnetic resonance imaging for radiation oncology, using style-blind auto-encoders.

Background and purpose: Magnetic Resonance Imaging (MRI) exhibits scanner dependent contrast, which limits generalisability of radiomics and machine-learning for radiation oncology. Current deep-learning harmonisation requires paired data, retraining for new scanners and often suffers from geometry-shift which alters anatomical information. The aim of this study was to investigate style-blind auto-encoders for MRI harmonisation to accommodate unpaired training data, avoid geometry-shift and harmonise data from previously unseen scanners. Materials and methods: A style-blind auto-encoder, using adversarial classification on the latent-space, was designed for MRI harmonisation. The public CC359 T1-w MRI brain dataset includes six scanners (three manufacturers, two field strengths), of which five were used for training. MRI from all six (including one unseen) scanner were harmonised to common contrast. Harmonisation extent was quantified via Kolmogorov-Smirnov testing of residual scanner dependence of 3D radiomic features, and compared to WhiteStripe normalisation. Anatomical content preservation was measured through change in structural similarity index on contrast-cycling (δSSIM). Results: The percentage of radiomics features showing statistically significant scanner-dependence was reduced from 41% (WhiteStripe) to 16% for white matter and from 39% to 27% for grey matter. δSSIM < 0.0025 on harmonisation and de-harmonisation indicated excellent anatomical content preservation. Conclusions: Our method harmonised MRI contrast effectively, preserved critical anatomical details at high fidelity, trained on unpaired data and allowed zero-shot harmonisation. Robust and clinically translatable harmonisation of MRI will enable generalisable radiomic and deep-learning models for a range of applications, including radiation oncology treatment stratification, planning and response monitoring.

Accuracy of Response Assessment FDG PET-CT Post (Chemo)Radiotherapy in HPV Negative Oropharynx Squamous Cell Carcinoma.

Background: Data on the accuracy of response assessment 2-[fluorine-18]-fluoro-2-deoxy-D-glucose (FDG) positron emission tomography-computed tomography (PET-CT) following (chemo)radiotherapy in patients with oropharynx squamous cell carcinoma (OPSCC) is predominantly based on HPV-positive disease. There is a paucity of data for HPV-negative disease, which has a less favourable prognosis. Methods: 96 patients treated with (chemo)radiotherapy for HPV-negative OPSCC with baseline and response assessment FDG PET-CT between 2013−2020, were analysed. PET-CT response was classified as negative, equivocal, or positive based on qualitative reporting. PET-CT response categories were analysed with reference to clinicopathological outcomes. Test characteristics were evaluated, comparing negative results to equivocal and positive results together. Post-test probabilities were calculated separately for positive and equivocal or negative results. Results: Median follow-up was 26 months. The negative predictive value of a negative scan was 93.7 and 93.2%, respectively, for primary tumour and nodal disease. For a negative scan, the post-test probability was 0.06 for primary and 0.07 for nodal disease. The post-test probability of an equivocal scan was 0.51 and 0.72 for primary and lymph node, respectively. The post-test probability of a positive scan approached 1. For patients with/without a negative scan, two-year overall survival and progression-free survival were 83% versus 30% and 79% versus 17% (p < 0.001), respectively. Conclusion: The NPV of a negative response assessment PET-CT in HPV-negative OPSCC is high, supporting a strategy of clinical monitoring. Contrasting with the published literature for HPV-positive OPSCC, an equivocal response scan was associated with a moderate rate of residual disease.

Discovery of Pre-Treatment FDG PET/CT-Derived Radiomics-Based Models for Predicting Outcome in Diffuse Large B-Cell Lymphoma.

BACKGROUND: Approximately 30% of patients with diffuse large B-cell lymphoma (DLBCL) will have recurrence. The aim of this study was to develop a radiomic based model derived from baseline PET/CT to predict 2-year event free survival (2-EFS). METHODS: Patients with DLBCL treated with R-CHOP chemotherapy undergoing pre-treatment PET/CT between January 2008 and January 2018 were included. The dataset was split into training and internal unseen test sets (ratio 80:20). A logistic regression model using metabolic tumour volume (MTV) and six different machine learning classifiers created from clinical and radiomic features derived from the baseline PET/CT were trained and tuned using four-fold cross validation. The model with the highest mean validation receiver operator characteristic (ROC) curve area under the curve (AUC) was tested on the unseen test set. RESULTS: 229 DLBCL patients met the inclusion criteria with 62 (27%) having 2-EFS events. The training cohort had 183 patients with 46 patients in the unseen test cohort. The model with the highest mean validation AUC combined clinical and radiomic features in a ridge regression model with a mean validation AUC of 0.75 ± 0.06 and a test AUC of 0.73. CONCLUSIONS: Radiomics based models demonstrate promise in predicting outcomes in DLBCL patients.

Exploratory Analysis of Serial 18F-fluciclovine PET-CT and Multiparametric MRI during Chemoradiation for Glioblastoma.

Anti-1-amino-3-18fluorine-fluorocyclobutane-1-carboxylic acid (18F-fluciclovine) positron emission tomography (PET) shows preferential glioma uptake but there is little data on how uptake correlates with post-contrast T1-weighted (Gd-T1) and dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) activity during adjuvant treatment. This pilot study aimed to compare 18F-fluciclovine PET, DCE-MRI and Gd-T1 in patients undergoing chemoradiotherapy for glioblastoma (GBM), and in a parallel pre-clinical GBM model, to investigate correlation between 18F-fluciclovine uptake, MRI findings, and tumour biology. 18F-fluciclovine-PET-computed tomography (PET-CT) and MRI including DCE-MRI were acquired before, during and after adjuvant chemoradiotherapy (60 Gy in 30 fractions with temozolomide) in GBM patients. MRI volumes were manually contoured; PET volumes were defined using semi-automatic thresholding. The similarity of the PET and DCE-MRI volumes outside the Gd-T1 volume boundary was measured using the Dice similarity coefficient (DSC). CT-2A tumour-bearing mice underwent MRI and 18F-fluciclovine PET-CT. Post-mortem mice brains underwent immunohistochemistry staining for ASCT2 (amino acid transporter), nestin (stemness) and Ki-67 (proliferation) to assess for biologically active tumour. 6 patients were recruited (GBM 1-6) and grouped according to overall survival (OS)-short survival (GBM-SS, median OS 249 days) and long survival (GBM-LS, median 903 days). For GBM-SS, PET tumour volumes were greater than DCE-MRI, in turn greater than Gd-T1. For GBM-LS, Gd-T1 and DCE-MRI were greater than PET. Tumour-specific 18F-fluciclovine uptake on pre-clinical PET-CT corresponded to immunostaining for Ki-67, nestin and ASCT2. Results suggest volumes of 18F-fluciclovine-PET activity beyond that depicted by DCE-MRI and Gd-T1 are associated with poorer prognosis in patients undergoing chemoradiotherapy for GBM. The pre-clinical model confirmed 18F-fluciclovine uptake reflected biologically active tumour.

Federated learning enables big data for rare cancer boundary detection.

Although machine learning (ML) has shown promise across disciplines, out-of-sample generalizability is concerning. This is currently addressed by sharing multi-site data, but such centralization is challenging/infeasible to scale due to various limitations. Federated ML (FL) provides an alternative paradigm for accurate and generalizable ML, by only sharing numerical model updates. Here we present the largest FL study to-date, involving data from 71 sites across 6 continents, to generate an automatic tumor boundary detector for the rare disease of glioblastoma, reporting the largest such dataset in the literature (n = 6, 314). We demonstrate a 33% delineation improvement for the surgically targetable tumor, and 23% for the complete tumor extent, over a publicly trained model. We anticipate our study to: 1) enable more healthcare studies informed by large diverse data, ensuring meaningful results for rare diseases and underrepresented populations, 2) facilitate further analyses for glioblastoma by releasing our consensus model, and 3) demonstrate the FL effectiveness at such scale and task-complexity as a paradigm shift for multi-site collaborations, alleviating the need for data-sharing.