Voxelwise Principal Component Analysis of Dynamic [S-Methyl-11C]Methionine PET Data in Glioma Patients

Applications of machine learning and deep learning in SPECT and PET imaging: General overview, challenges and future prospects

The integration of positron emission tomography (PET) and single-photon emission computed tomography (SPECT) imaging techniques with machine learning (ML) algorithms, including deep learning (DL) models, is a promising approach. This integration enhances the precision and efficiency of current diagnostic and treatment strategies while offering invaluable insights into disease mechanisms. In this comprehensive review, we delve into the transformative impact of ML and DL in this domain. Firstly, a brief analysis is provided of how these algorithms have evolved and which are the most widely applied in this domain. Their different potential applications in nuclear imaging are then discussed, such as optimization of image adquisition or reconstruction, biomarkers identification, multimodal fusion and the development of diagnostic, prognostic, and disease progression evaluation systems. This is because they are able to analyse complex patterns and relationships within imaging data, as well as extracting quantitative and objective measures. Furthermore, we discuss the challenges in implementation, such as data standardization and limited sample sizes, and explore the clinical opportunities and future horizons, including data augmentation and explainable AI. Together, these factors are propelling the continuous advancement of more robust, transparent, and reliable systems.

Evaluation of in vivo staging of amyloid deposition in cognitively unimpaired elderly aged 78-94

Amyloid-beta (Aβ) deposition is common in cognitively unimpaired (CU) elderly >85 years. This study investigated amyloid distribution and evaluated three published in vivo amyloid-PET staging schemes from a cognitively unimpaired (CU) cohort aged 84.9 ± 4.3 years (n = 75). SUV-based principal component analysis (PCA) was applied to 18F-flutemetamol PET data to determine an unbiased regional covariance pattern of tracer uptake across grey matter regions. PET staging schemes were applied to the data and compared to the PCA output. Concentration of p-tau181 was measured in blood plasma. The PCA revealed three distinct components accounting for 91.2% of total SUV variance. PC1 driven by the large common variance of uptake in neocortical and striatal regions was significantly positively correlated with global SUVRs, APOE4 status and p-tau181 concentration. PC2 represented mainly non-specific uptake in typical amyloid-PET reference regions, and PC3 the occipital lobe. Application of the staging schemes demonstrated that the majority of the CU cohort (up to 93%) were classified as having pathological amount and distribution of Aβ. Good correspondence existed between binary (+/-) classification and later amyloid stages, however, substantial differences existed between schemes for low stages with 8-17% of individuals being unstageable, i.e., not following the sequential progression of Aβ deposition. In spite of the difference in staging outcomes there was broad spatial overlap between earlier stages and PC1, most prominently in default mode network regions. This study critically evaluated the utility of in vivo amyloid staging from a single PET scan in CU elderly and found that early amyloid stages could not be consistently classified. The majority of the cohort had pathological Aβ, thus, it remains an open topic what constitutes abnormal brain Aβ in the oldest-old and what is the best method to determine that.

Cost function for low-dimensional manifold topology assessment

In reduced-order modeling, complex systems that exhibit high state-space dimensionality are described and evolved using a small number of parameters. These parameters can be obtained in a data-driven way, where a high-dimensional dataset is projected onto a lower-dimensional basis. A complex system is then restricted to states on a low-dimensional manifold where it can be efficiently modeled. While this approach brings computational benefits, obtaining a good quality of the manifold topology becomes a crucial aspect when models, such as nonlinear regression, are built on top of the manifold. Here, we present a quantitative metric for characterizing manifold topologies. Our metric pays attention to non-uniqueness and spatial gradients in physical quantities of interest, and can be applied to manifolds of arbitrary dimensionality. Using the metric as a cost function in optimization algorithms, we show that optimized low-dimensional projections can be found. We delineate a few applications of the cost function to datasets representing argon plasma, reacting flows and atmospheric pollutant dispersion. We demonstrate how the cost function can assess various dimensionality reduction and manifold learning techniques as well as data preprocessing strategies in their capacity to yield quality low-dimensional projections. We show that improved manifold topologies can facilitate building nonlinear regression models.

High Expression of CKS2 Predicts Adverse Outcomes: A Potential Therapeutic Target for Glioma

Cyclin-dependent kinase regulatory subunit 2 (CKS2) is a potential prognostic marker and is overexpressed in various cancers. This study analyzed sequencing and clinical data from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus, with external validation using the Chinese Glioma Genome Atlas (CGGA) data. CKS2 expression in the normal brain and tumor tissue was compared. cBioPortal and MethSurv were utilized to scrutinize the prognostic value of CKS2 methylation. Gene set enrichment examination and single-sample gene set enrichment analysis were employed to explore the potential biological functions of CKS2. Cell viability, colony formation, and transwell assays were conducted to evaluate the influence of CKS2 on glioma cell proliferation and invasion. Compared with normal brain tissue, the expression of CKS2 was upregulated in glioma samples (p < 0.001). Multivariate data analysis from TCGA and CGGA indicated that increased expression of CKS2 was an independent risk factor for the prognosis of overall survival in glioma patients. CKS2 methylation was negatively associated with CKS2 expression. Patients with CKS2 hypomethylation had worse overall survival compared with patients with CKS2 methylation, as suggested by the analysis of both TCGA and CGGA datasets. The expression level of CKS2 is closely related to tumor immunity, including the correlation of tumor immune cell infiltration, immune score, and co-expression of multiple immune-related genes. In addition, CKS2 is associated with several immune checkpoints and responses to the chemotherapy drug cisplatin. CKS2 knockdown impeded the expansion and aggression of glioma cell lines. The changes in CKS2 expression may provide a novel prognostic biomarker that can be used to improve patient overall survival rates.

PET Imaging in Neuro-Oncology: An Update and Overview of a Rapidly Growing Area

Simple Summary

Positron emission tomography (PET) is a functional imaging technique which plays an increasingly important role in the management of brain tumors. Owing different radiotracers, PET allows to image different metabolic aspects of the brain tumors. This review outlines currently available PET radiotracers and their respective indications in neuro-oncology. It specifically focuses on the investigation of gliomas, meningiomas, primary central nervous system lymphomas as well as brain metastases. Recent advances in the production of PET radiotracers, image analyses and translational applications to peptide radionuclide receptor therapy, which allow to treat brain tumors with radiotracers, are also discussed. The objective of this review is to provide a comprehensive overview of PET imaging’s potential in neuro-oncology as an adjunct to brain magnetic resonance imaging (MRI).

Abstract

PET plays an increasingly important role in the management of brain tumors. This review outlines currently available PET radiotracers and their respective indications. It specifically focuses on ¹⁸F-FDG, amino acid and somatostatin receptor radiotracers, for imaging gliomas, meningiomas, primary central nervous system lymphomas as well as brain metastases. Recent advances in radiopharmaceuticals, image analyses and translational applications to therapy are also discussed. The objective of this review is to provide a comprehensive overview of PET imaging’s potential in neuro-oncology as an adjunct to brain MRI for all medical professionals implicated in brain tumor diagnosis and care.

Relevance of Dynamic 18F-DOPA PET Radiomics for Differentiation of High-Grade Glioma Progression from Treatment-Related Changes

This study evaluates the relevance of ¹⁸F-DOPA PET static and dynamic radiomics for differentiation of high-grade glioma (HGG) progression from treatment-related changes (TRC) by comparing diagnostic performances to the current PET imaging standard of care. Eighty-five patients with histologically confirmed HGG and investigated by dynamic ¹⁸F-FDOPA PET in two institutions were retrospectively selected. ElasticNet logistic regression, Random Forest and XGBoost machine models were trained with different sets of features-radiomics extracted from static tumor-to-background-ratio (TBR) parametric images, radiomics extracted from time-to-peak (TTP) parametric images, as well as combination of both-in order to discriminate glioma progression from TRC at 6 months from the PET scan. Diagnostic performances of the models were compared to a logistic regression model with TBR_mean ± clinical features used as reference. Training was performed on data from the first center, while external validation was performed on data from the second center. Best radiomics models showed only slightly better performances than the reference model (respective AUCs of 0.834 vs. 0.792, p < 0.001). Our current results show similar findings at the multicentric level using different machine learning models and report a marginal additional value for TBR static and TTP dynamic radiomics over the classical analysis based on TBR values.

Effects of Carbidopa Premedication on 18F-FDOPA PET Imaging of Glioma: A Multiparametric Analysis

Simple Summary

¹⁸F-FDOPA PET imaging is routinely used and recommended to assess gliomas. Carbidopa is a peripheral enzyme inhibitor. Carbidopa premedication increases the radiotracer uptake on static images. None of the evidence-based data available to date recommends carbidopa premedication. Our study therefore determined the impact of carbidopa premedication on static, radiomics and dynamic parameters for ¹⁸F-FDOPA PET brain tumor imaging. We show that carbidopa premedication leads to higher SUV and TTP dynamic parameters and impacts SUV-dependent radiomics by the same magnitude in healthy brains and tumors. The carbidopa effect is therefore compensated for by correcting for the tumor-to-healthy-brain ratio, a significant advantage for harmonizing data for multicentric studies. Results were obtained from simulations of time-activity curves using compartmental modeling.

Abstract

Purpose: This study aimed to determine the impact of carbidopa premedication on static, dynamic and radiomics parameters of ¹⁸F-FDOPA PET in brain tumors. Methods: The study included 54 patients, 18 of whom received carbidopa, who underwent ¹⁸F-FDOPA PET for newly diagnosed gliomas. SUV-derived, 105 radiomics features and TTP dynamic parameters were extracted from volumes of interest in healthy brains and tumors. Simulation of the effects of carbidopa on time-activity curves were generated. Results: All static and TTP dynamic parameters were significantly higher in healthy brain regions of premedicated patients (ΔSUV_mean = +53%, ΔTTP = +48%, p < 0.001). Furthermore, carbidopa impacted 81% of radiomics features, of which 92% correlated with SUV_mean (absolute correlation coefficient ≥ 0.4). In tumors, premedication with carbidopa was an independent predictor of SUV_mean (ΔSUV_mean = +52%, p < 0.001) and TTP (ΔTTP = +24%, p = 0.025). All parameters were no longer significantly modified by carbidopa premedication when using ratios to healthy brain. Simulated data confirmed that carbidopa leads to higher tumor TTP values, corrected by the ratios. Conclusion: In ¹⁸F-FDOPA PET, carbidopa induces similarly higher SUV and TTP dynamic parameters and similarly impacts SUV-dependent radiomics in healthy brain and tumor regions, which is compensated for by correcting for the tumor-to-healthy-brain ratio. This is a significant advantage for multicentric study harmonization.

Dynamic 18F-FDopa PET Imaging for Newly Diagnosed Gliomas: Is a Semiquantitative Model Sufficient?

Purpose

Dynamic amino acid positron emission tomography (PET) has become essential in neuro-oncology, most notably for its prognostic value in the noninvasive prediction of isocitrate dehydrogenase (IDH) mutations in newly diagnosed gliomas. The 6-[¹⁸F]fluoro-l-DOPA (¹⁸F-FDOPA) kinetic model has an underlying complexity, while previous studies have predominantly used a semiquantitative dynamic analysis. Our study addresses whether a semiquantitative analysis can capture all the relevant information contained in time–activity curves for predicting the presence of IDH mutations compared to the more sophisticated graphical and compartmental models.

Methods

Thirty-seven tumour time–activity curves from ¹⁸F-FDOPA PET dynamic acquisitions of newly diagnosed gliomas (median age = 58.3 years, range = 20.3–79.9 years, 16 women, 16 IDH-wild type) were analyzed with a semiquantitative model based on classical parameters, with (SQ) or without (Ref SQ) a reference region, or on parameters of a fit function (SQ Fit), a graphical Logan model with input function (Logan) or reference region (Ref Logan), and a two-tissue compartmental model previously reported for ¹⁸F-FDOPA PET imaging of gliomas (2TCM). The overall predictive performance of each model was assessed with an area under the curve (AUC) comparison using multivariate analysis of all the parameters included in the model. Moreover, each extracted parameter was assessed in a univariate analysis by a receiver operating characteristic curve analysis.

Results

The SQ model with an AUC of 0.733 for predicting IDH mutations showed comparable performance to the other models with AUCs of 0.752, 0.814, 0.693, 0.786, and 0.863, respectively corresponding to SQ Fit, Ref SQ, Logan, Ref Logan, and 2TCM (p ≥ 0.10 for the pairwise comparisons with other models). In the univariate analysis, the SQ time-to-peak parameter had the best diagnostic performance (75.7% accuracy) compared to all other individual parameters considered.

Conclusions

The SQ model circumvents the complexities of the ¹⁸F-FDOPA kinetic model and yields similar performance in predicting IDH mutations when compared to the other models, most notably the compartmental model. Our study provides supportive evidence for the routine clinical application of the SQ model for the dynamic analysis of ¹⁸F-FDOPA PET images in newly diagnosed gliomas.

Dynamic 11C-Methionine PET-CT: Prognostic Factors for Disease Progression and Survival in Patients with Suspected Glioma Recurrence

Simple Summary

Recurrence after initial treatments is an expected event in glioma patients, particularly for high-grade glioma, with a median progression-free survival of 8–11 weeks. The prognostic evaluation of disease is a crucial step in the planning of therapeutic strategies, in both the primary and recurrence stages of disease. The aim of our retrospective study was to assess the prognostic value of ¹¹C-methionine PET-CT dynamic and semiquantitative parameters in patients with suspected glioma recurrence at MR, in terms of progression-free survival and overall survival. In a population of sixty-seven consecutive patients, both static and kinetic analyses provided parameters (i.e., tumour-to-background ratio and SUVmax associated with time-to-peak, respectively) able to predict both progression-free and overall survival in the whole population and in the high-grade glioma subgroup of patients. Dynamic ¹¹C-methionine PET-CT can be a useful diagnostic tool, in patients with suspicion of glioma recurrence, able to produce significant prognostic indices.

Abstract

Purpose: The prognostic evaluation of glioma recurrence patients is important in the therapeutic management. We investigated the prognostic value of ¹¹C-methionine PET-CT (MET-PET) dynamic and semiquantitative parameters in patients with suspected glioma recurrence. Methods: Sixty-seven consecutive patients who underwent MET-PET for suspected glioma recurrence at MR were retrospectively included. Twenty-one patients underwent static MET-PET; 46/67 underwent dynamic MET-PET. In all patients, SUVmax, SUVmean and tumour-to-background ratio (T/B) were calculated. From dynamic acquisition, the shape and slope of time-activity curves, time-to-peak and its SUVmax (SUVmax_TTP) were extrapolated. The prognostic value of PET parameters on progression-free (PFS) and overall survival (OS) was evaluated using Kaplan–Meier survival estimates and Cox regression. Results: The overall median follow-up was 19 months from MET-PET. Recurrence patients (38/67) had higher SUVmax (p = 0.001), SUVmean (p = 0.002) and T/B (p < 0.001); deceased patients (16/67) showed higher SUVmax (p = 0.03), SUVmean (p = 0.03) and T/B (p = 0.006). All static parameters were associated with PFS (all p < 0.001); T/B was associated with OS (p = 0.031). Regarding kinetic analyses, recurrence (27/46) and deceased (14/46) patients had higher SUVmax_TTP (p = 0.02, p = 0.01, respectively). SUVmax_TTP was the only dynamic parameter associated with PFS (p = 0.02) and OS (p = 0.006). At univariate analysis, SUVmax, SUVmean, T/B and SUVmax_TTP were predictive for PFS (all p < 0.05); SUVmax_TTP was predictive for OS (p = 0.02). At multivariate analysis, SUVmax_TTP remained significant for PFS (p = 0.03). Conclusion: Semiquantitative parameters and SUVmax_TTP were associated with clinical outcomes in patients with suspected glioma recurrence. Dynamic PET-CT acquisition, with static and kinetic parameters, can be a valuable non-invasive prognostic marker, identifying patients with worse prognosis who require personalised therapy.