Multiparametric Magnetic Resonance Imaging Scans Research Articles

Impact of non-contrast-enhanced imaging input sequences on the generation of virtual contrast-enhanced breast MRI scans using neural network.

To investigate how different combinations of T1-weighted (T1w), T2-weighted (T2w), and diffusion-weighted imaging (DWI) impact the performance of virtual contrast-enhanced (vCE) breast MRI. The IRB-approved, retrospective study included 1064 multiparametric breast MRI scans (age: 52 ± 12 years) obtained from 2017 to 2020 (single site, two 3-T MRI). Eleven independent neural networks were trained to derive vCE images from varying input combinations of T1w, T2w, and multi-b-value DWI sequences (b-value = 50-1500 s/mm2). Three readers evaluated the vCE images with regard to qualitative scores of diagnostic image quality, image sharpness, satisfaction with contrast/signal-to-noise ratio, and lesion/non-mass enhancement conspicuity. Quantitative metrics (SSIM, PSNR, NRMSE, and median symmetrical accuracy) were analyzed and statistically compared between the input combinations for the full breast volume and both enhancing and non-enhancing target findings. The independent test set consisted of 187 cases. The quantitative metrics significantly improved in target findings when multi-b-value DWI sequences were included during vCE training (p < 0.05). Non-significant effects (p > 0.05) were observed for the quantitative metrics on the full breast volume when comparing input combinations including T1w. Using T1w and DWI acquisitions during vCE training is necessary to achieve high satisfaction with contrast/SNR and good conspicuity of the enhancing findings. The input combination of T1w, T2w, and DWI sequences with three b-values showed the best qualitative performance. vCE breast MRI performance is significantly influenced by input sequences. Quantitative metrics and visual quality of vCE images significantly benefit when multi b-value DWI is added to morphologic T1w-/T2w sequences as input for model training. Question How do different MRI sequences impact the performance of virtual contrast-enhanced (vCE) breast MRI? Findings The input combination of T1-weighted, T2-weighted, and diffusion-weighted imaging sequences with three b-values showed the best qualitative performance. Clinical relevance While in the future neural networks providing virtual contrast-enhanced images might further improve accessibility to breast MRI, the significant influence of input data needs to be considered during translational research.

Sorafenib plus memory-like natural killer cell immunochemotherapy boosts treatment response in liver cancer

BackgroundHeterogeneity of hepatocellular carcinoma (HCC) presents significant challenges for therapeutic strategies and necessitates combinatorial treatment approaches to counteract suppressive behavior of tumor microenvironment and achieve improved outcomes. Here, we employed cytokines to induce memory-like behavior in natural killer (NK) cells, thereby enhancing their cytotoxicity against HCC. Additionally, we evaluated the potential benefits of combining sorafenib with this newly developed memory-like NK cell (pNK) immunochemotherapy in a preclinical model.MethodsHCC tumors were grown in SD rats using subcapsular implantation. Interleukin 12/18 cytokines were supplemented to NK cells to enhance cytotoxicity through memory activation. Tumors were diagnosed using MRI, and animals were randomly assigned to control, pNK immunotherapy, sorafenib chemotherapy, or combination therapy groups. NK cells were delivered locally via the gastrointestinal tract, while sorafenib was administered systemically. Therapeutic responses were monitored with weekly multi-parametric MRI scans over three weeks. Afterward, tumor tissues were harvested for histopathological analysis. Structural and functional changes in tumors were evaluated by analyzing MRI and histopathology data using ANOVA and pairwise T-test analyses.ResultsThe tumors were allowed to grow for six days post-cell implantation before treatment commenced. At baseline, tumor diameter averaged 5.27 mm without significant difference between groups (p = 0.16). Both sorafenib and combination therapy imposed greater burden on tumor dimensions compared to immunotherapy alone in the first week. By the second week of treatment, combination therapy had markedly expanded its therapeutic efficacy, resulting in the most significant tumor regression observed (6.05 ± 1.99 vs. 13.99 ± 8.01 mm). Histological analysis demonstrated significantly improved cell destruction in the tumor microenvironment associated with combination treatment (63.79%). Interestingly, we observed fewer viable tumor regions in the sorafenib group (38.9%) compared to the immunotherapy group (45.6%). Notably, there was a significantly higher presence of NK cells in the tumor microenvironment with combination therapy (34.79%) compared to other groups (ranging from 2.21 to 26.50%). Although the tumor sizes in the monotherapy groups were similar, histological analysis revealed a stronger response in pNK cell immunotherapy group compared to the sorafenib group.ConclusionsExperimental results indicated that combination therapy significantly enhanced treatment response, resulting in substantial tumor growth reduction in alignment with histological analysis.

Long-term Results of Hypofractionated Radiotherapy With Intra-prostatic Boosts in Men With Intermediate- and High-risk Prostate Cancer: A Phase II Trial

AimsIn the conventionally fractionated phase III FLAME prostate trial, focal boosts improved local control and biochemical disease-free survival (bDFS). We explored the toxicity and effectiveness of a moderately hypofractionated schedule with focal boosts. Material and methodsBIOPROP20 is a phase II single-arm non-randomised trial for intermediate- to very high-risk localised prostate cancer patients with bulky tumour volumes. Multi-parametric magnetic resonance imaging (MRI) and 18F-choline positron emission tomography-computed tomography (PET-CT) scans were used for staging and boost volume definition. Patients were treated with 60Gy in 20 fractions with a boost dose up to 68Gy. Five patients with positive lymph nodes on the PET-CT scan received radiotherapy to pelvic lymph nodes (45Gy to elective nodes, boosted up to 50Gy to involved nodes). Primary outcomes were acute (≤18 weeks) and late urinary and gastrointestinal toxicity, prospectively recorded up to 5 years with Common Terminology Criteria for Adverse Events v4 (CTCAE). Secondary outcomes were biochemical or clinical progression, metastasis-free survival (MFS), and overall survival (OS). Results61 patients completed radiotherapy with hormone therapy (range: 6–36 months). Cumulative acute and late gastrointestinal toxicity was low at 6.6% and 5.0%, respectively. Cumulative acute and late urinary toxicity was 49.2% and 30.1%, respectively; the prevalence reduced to 5.9% at 5 years. At 5 years: 6 patients had biochemical progression (bDFS: 88.5%; 95% CI: 80.2–97.6%), the MFS was 82.4% (95% CI: 73.0–92.9%), 5 patients died (OS: 91.2%; 95% CI: 84.1–98.9%), one with prostate cancer. The prostate, boost, nodal planning volumes, and the organs at risk (rectum, bowel, urethra, and bladder) met the optimal protocol dose constraints. There was a trend to increased urinary toxicity with increasing urethral (RR: 1.95, 95% CI: 0.73–5.22, p = 0.18), but not bladder dose. ConclusionFocal boosts with a 20 fraction hypofractionated prostate radiotherapy schedule are associated with an acceptable risk of gastrointestinal and urinary toxicity and achieve good cancer control. ClinicalTrials.gov identifierNCT02125175.

Optimizing Glioblastoma, IDH-wildtype Treatment Outcomes : A Radiomics and Support Vector Machine -Based Approach to Overall Survival Estimation.

Glioblastoma multiforme (GBM), particularly the IDH-wildtype type, represents a significant clinical challenge due to its aggressive nature and poor prognosis. Despite advancements in medical imaging and its modalities, survival rates have not improved significantly, demanding innovative treatment planning and outcome prediction approaches. This study utilizes a Support Vector Machine (SVM) classifier using radiomics features to predict the overall survival (OS) of GBM, IDH-wildtype patients to short (< 12 Months) and long (>=12 Months) survivors. A dataset comprising multi-parametric MRI (mpMRI) scans from 574 patients was analyzed. Radiomic features were extracted from T1, T2, FLAIR, and T1-Gd sequences. Low variance features were removed, and Recursive Feature Elimination (RFE) was used to select the most informative features. The SVM model was trained using a k-fold cross-validation approach. Furthermore, clinical parameters such as age, gender, and MGMT promoter methylation status were integrated to enhance prediction accuracy. The model showed reasonable results in terms of cross-validated AUC of 0.84 (95% CI: 0.80-0.90) with (p-value < 0.001) effectively categorizing patients into short and long survivors. Log-rank test (Chi-square statistics) analysis for the developed model was 0.00029 along with the 1.20 Cohen's d effect size. Most importantly, clinical data integration further refined the survival estimates, providing a more fitted prediction that considers individual patient characteristics by Kaplan-Meier curve with p-value<0.0001. The proposed method significantly enhances the predictive accuracy of OS outcomes in GBM, IDH-wildtype patients. By integrating detailed imaging features with key clinical indicators, this model offers a robust tool for personalized treatment planning, potentially improving OS.

Multi-parametric MRI-based machine learning model for prediction of pathological grade of renal injury in a rat kidney cold ischemia-reperfusion injury model

BackgroundRenal cold ischemia-reperfusion injury (CIRI), a pathological process during kidney transplantation, may result in delayed graft function and negatively impact graft survival and function. There is a lack of an accurate and non-invasive tool for evaluating the degree of CIRI. Multi-parametric MRI has been widely used to detect and evaluate kidney injury. The machine learning algorithms introduced the opportunity to combine biomarkers from different MRI metrics into a single classifier.ObjectiveTo evaluate the performance of multi-parametric magnetic resonance imaging for grading renal injury in a rat model of renal cold ischemia-reperfusion injury using a machine learning approach.MethodsEighty male SD rats were selected to establish a renal cold ischemia -reperfusion model, and all performed multiparametric MRI scans (DWI, IVIM, DKI, BOLD, T1mapping and ASL), followed by pathological analysis. A total of 25 parameters of renal cortex and medulla were analyzed as features. The pathology scores were divided into 3 groups using K-means clustering method. Lasso regression was applied for the initial selecting of features. The optimal features and the best techniques for pathological grading were obtained. Multiple classifiers were used to construct models to evaluate the predictive value for pathology grading.ResultsAll rats were categorized into mild, moderate, and severe injury group according the pathologic scores. The 8 features that correlated better with the pathologic classification were medullary and cortical Dp, cortical T2*, cortical Fp, medullary T2*, ∆T1, cortical RBF, medullary T1. The accuracy(0.83, 0.850, 0.81, respectively) and AUC (0.95, 0.93, 0.90, respectively) for pathologic classification of the logistic regression, SVM, and RF are significantly higher than other classifiers. For the logistic model and combining logistic, RF and SVM model of different techniques for pathology grading, the stable and perform are both well. Based on logistic regression, IVIM has the highest AUC (0.93) for pathological grading, followed by BOLD(0.90).ConclusionThe multi-parametric MRI-based machine learning model could be valuable for noninvasive assessment of the degree of renal injury.

BIOM-14. UTILITY OF MIRNA-BASED LIQUID BIOPSY AND RADIOGRAPHIC IMAGING IN PEDIATRIC BRAIN TUMOR PATIENTS

Abstract BACKGROUND MRI is the current gold standard imaging technique for diagnostic evaluation and monitoring of pediatric CNS tumors; however, MRI measures can only provide inference into tumor biology and molecular stratification. miRNAs are an abundant and stable nucleic acid found circulating in blood and cerebrospinal fluid (CSF) and can be utilized as a tumor biomarker. METHODS We hypothesized that correlating miRNA biomarkers with radiological tumor measurements may provide improved diagnostic and monitoring for patients with pediatric brain tumors. Using a cohort of 54 pediatric brain tumors including low grade glioma, ependymoma, germ cell tumor, medulloblastoma, atypical teratoid rhabdoid tumor and high-grade glioma we attempted to combine MRI findings and circulating miRNA data. The miRNA expression was profiled in 33 CSF and 52 plasma samples using the HTG EdgeSeq platform. Clinically acquired, multi-parametric MRI scans at time-points close to liquid biopsy collection were collected retrospectively and used to generate volumetric tumor segmentations. Machine learning methods were implemented to evaluate combinatory features. RESULTS We identified unique miRNA targets that significantly correlated with MRI features, clinical findings, and patient outcomes. In both CSF and plasma, miRNA expression was found to correlate with diagnosis and clinical features including tumor grade and disease burden (p &amp;lt; 0.05). In CSF, miRNA expression was correlated with MRI features including cystic core presence, edema, leptomeningeal disease and tumor location (p &amp;lt;0.05). In plasma, differences in miRNA expression were related to MRI measurements including cystic core volume, location, and leptomeningeal disease (p &amp;lt; 0.05). Combination of miRNA targets for plasma or CSF with imaging-based features significantly improved histology specific multimodal diagnosis predictions. CONCLUSIONS These results demonstrate the potential utility of miRNAs as a pediatric brain tumor biomarker which when combined with imaging features can improve minimally and non-invasive diagnostics and the subsequent management of pediatric brain tumor patients.

IMG-12. CBTN BRAIN TUMOR SEGMENTATION INITIATIVE: UPDATES ON MODEL RELEASE, HGG SEGMENTATION, AND SURVIVAL ANALYSIS

Abstract BACKGROUND Assessment of treatment responses in pediatric brain tumors requires accurate tumor segmentation, particularly important for nonresectable tumors like diffuse midline glioma (DMG), including diffuse intrinsic pontine glioma (DIPGs). Evaluating tumor progression in these tumors relies on monitoring tumor size changes in longitudinal MRI exams which is challenged by their infiltrative growth patterns. Our team developed a comprehensive multi-institutional, multi-histology autosegmentation deep learning (DL) model leveraging data from the Children’s Brain Tumor Network (CBTN), involving continuous enhancements and enriched by an expanding dataset including post-surgical/post-treatment studies. METHODS Recently, we trained and validated this DL model (publicly available) on a multi-institutional dataset of 644 PBTs, including 273 pediatric-type diffuse high-grade gliomas (HGGs, including DMG/DIPGs), using pre-operative and post-operative/post-treatment multiparametric MRI scans. We tested the model in 23 longitudinal MRI exams of 8 DMGs from the PNOC003 and PNOC007 clinical trials and applied Cox regression for predicting overall survival (OS) using tumor volume, computed with our autosegmentation model, as a time-varying variable in 18 patients. RESULTS For HGGs in treatment-naïve sessions, our model demonstrated strong performance with median Dice scores of 0.89 for whole tumor and tumor core (encompassing all regions, except for edema), and 0.73 for enhancing tumor segmentation. In longitudinal MRI exams of DMGs from the aforementioned clinical trials, the model achieved median Dice scores of 0.79 for whole tumor, and 0.78 for enhancing tumor and tumor core segmentation. Cox regression showed the significance of longitudinal volumetric measurements (p=0.08) for predicting OS in DMGs. CONCLUSIONS Our approach provides reliable tumor segmentation in HGGs across diverse datasets, crucial for treatment response assessment. Looking ahead, we plan to validate our model on a larger set of longitudinal exams and use volumetric measurements to predict tumor progression. This ongoing effort aims to improve model performance, extend its generalizability, and mitigate potential model decay.

Comparing Magnetic Resonance Imaging and Prostate-Specific Membrane Antigen-Positron Emission Tomography for Prediction of Extraprostatic Extension of Prostate Cancer and Surgical Guidance: A Prospective Nonrandomized Clinical Trial.

Survivors of surgically managed prostate cancer may experience urinary incontinence and erectile dysfunction. Our aim was to determine if 68Ga-prostate-specific membrane antigen-11 positron emission tomography CT (PSMA-PET) in addition to multiparametric (mp) MRI scans improved surgical decision-making for nonnerve-sparing or nerve-sparing approach. We prospectively enrolled 50 patients at risk for extraprostatic extension (EPE) who were scheduled for prostatectomy. After mpMRI and PSMA-PET images were read for EPE prediction, surgeons prospectively answered questionnaires based on mpMRI and PSMA-PET scans on the decision for nerve-sparing or nonnerve-sparing approach. Final whole-mount pathology was the reference standard. Sensitivity, specificity, positive predictive value, negative predictive value, and receiver operating characteristic curves were calculated and McNemar's test was used to compare imaging modalities. The median age and PSA were 61.5 years and 7.0 ng/dL. The sensitivity for EPE along the posterior neurovascular bundle was higher for PSMA-PET than mpMRI (86% vs 57%, P = .03). For MRI, the specificity, positive predictive value, negative predictive value, and area under the curve for the receiver operating characteristic curves were 77%, 40%, 87%, and 0.67, and for PSMA-PET were 73%, 46%, 95%, and 0.80. PSMA-PET and mpMRI reads differed on 27 nerve bundles, with PSMA-PET being correct in 20 cases and MRI being correct in 7 cases. Surgeons predicted correct nerve-sparing approach 74% of the time with PSMA-PET scan in addition to mpMRI compared to 65% with mpMRI alone (P = .01). PSMA-PET scan was more sensitive than mpMRI for EPE along the neurovascular bundles and improved surgical decisions for nerve-sparing approach. Further study of PSMA-PET for surgical guidance is warranted in the unfavorable intermediate-risk or worse populations. NCT04936334.

#2974 Serial multiparametric magnetic resonance imaging in a prospective cohort of patients with acute kidney injury

Abstract Background and Aims Recovery from acute kidney injury (AKI) is traditionally measured using serum creatinine but this can often over-estimate the degree of renal recovery. It is known, however, that failure of recovery of serum creatinine 90 days after AKI strongly associates with subsequent long-term reductions in renal function making it an important clinical timepoint. Magnetic resonance imaging (MRI) is an imaging modality with promise to improve the understanding and characterisation of renal pathophysiology. In a single multiparametric MRI (mpMRI) scan, multiple measures can assess renal morphology, tissue microstructure (T1 relaxation time and diffusion-weighted-imaging (DWI)), oxygenation, perfusion and blood flow. We previously published the first study of renal mpMRI at the time of AKI and during follow up, this showed increased total kidney volume (TKV) and T1 as a measure of inflammation/fibrosis at the time of AKI which remained elevated in some participants at 3 and 12 months despite normalisation of serum creatinine. This study aimed to assess the degree of change between 30 and 90 days post AKI, as the first 90 days after AKI appear to be the important in terms of outcomes. Method Prospective observational study of 10 participants with AKI of all stages recruited at the time of AKI and followed up with monthly bloods and urine. MRI scans were collected on a Philips Ingenia 3T at day 30-60 and day 90. The 1-hour protocol included: T2-weighted scans for automated segmentation to compute total kidney volume (TKV), as well as T1 &amp; T2 mapping, DWI, ASL and phase contrast MRI for kidney perfusion and blood flow, and TRUST-MRI and BOLD R2* for kidney oxygenation, and MRE for stiffness. Results 10 participants were scanned at 2 timepoints, (mean day 45 for 1st scan, 95 for 2nd scan). Baseline characteristics are summarised in Table 1 along with key clinical measures in Table 2. 2 participants had CKD at baseline. Despite a range of peak creatinine values and AKI stages the majority of participants serum creatinine normalised to within 20% of baseline by 30 days (v3) (Fig. 1). Fig. 2 shows the TKV for AKI patients across the 2 visits as compared with healthy values. Despite normalisation of biochemistry, 5 participants had enlarged TKV. The predominant AKI aetiology for these patients was sepsis. Across all participants there is little dynamic change between visits. The participant with the small TKV across both visits had pre-existing CKD, as well as the highest peak creatine. The participant did not recover renal function by day 90 (Figs 1(j) and 2). Conclusion Despite recovery of biochemistry, at least half of this AKI cohort had abnormal MR measures at that persisted through one and three months after the AKI episode. There is a trend towards larger TKV in patients with AKI due to sepsis, which needs to be explored in a larger cohort. Analysis of more parameters as well as the addition of liquid biomarkers will enrich these preliminary findings.

Deformable registration of magnetic resonance images using unsupervised deep learning in neuro-/radiation oncology

PurposeAccurate deformable registration of magnetic resonance imaging (MRI) scans containing pathologies is challenging due to changes in tissue appearance. In this paper, we developed a novel automated three-dimensional (3D) convolutional U-Net based deformable image registration (ConvUNet-DIR) method using unsupervised learning to establish correspondence between baseline pre-operative and follow-up MRI scans of patients with brain glioma.MethodsThis study involved multi-parametric brain MRI scans (T1, T1-contrast enhanced, T2, FLAIR) acquired at pre-operative and follow-up time for 160 patients diagnosed with glioma, representing the BraTS-Reg 2022 challenge dataset. ConvUNet-DIR, a deep learning-based deformable registration workflow using 3D U-Net style architecture as a core, was developed to establish correspondence between the MRI scans. The workflow consists of three components: (1) the U-Net learns features from pairs of MRI scans and estimates a mapping between them, (2) the grid generator computes the sampling grid based on the derived transformation parameters, and (3) the spatial transformation layer generates a warped image by applying the sampling operation using interpolation. A similarity measure was used as a loss function for the network with a regularization parameter limiting the deformation. The model was trained via unsupervised learning using pairs of MRI scans on a training data set (n = 102) and validated on a validation data set (n = 26) to assess its generalizability. Its performance was evaluated on a test set (n = 32) by computing the Dice score and structural similarity index (SSIM) quantitative metrics. The model’s performance also was compared with the baseline state-of-the-art VoxelMorph (VM1 and VM2) learning-based algorithms.ResultsThe ConvUNet-DIR model showed promising competency in performing accurate 3D deformable registration. It achieved a mean Dice score of 0.975 ± 0.003 and SSIM of 0.908 ± 0.011 on the test set (n = 32). Experimental results also demonstrated that ConvUNet-DIR outperformed the VoxelMorph algorithms concerning Dice (VM1: 0.969 ± 0.006 and VM2: 0.957 ± 0.008) and SSIM (VM1: 0.893 ± 0.012 and VM2: 0.857 ± 0.017) metrics. The time required to perform a registration for a pair of MRI scans is about 1 s on the CPU.ConclusionsThe developed deep learning-based model can perform an end-to-end deformable registration of a pair of 3D MRI scans for glioma patients without human intervention. The model could provide accurate, efficient, and robust deformable registration without needing pre-alignment and labeling. It outperformed the state-of-the-art VoxelMorph learning-based deformable registration algorithms and other supervised/unsupervised deep learning-based methods reported in the literature.

Fully automatic mpMRI analysis using deep learning predicts peritumoral glioblastoma infiltration and subsequent recurrence.

Glioblastoma (GBM) is most aggressive and common adult brain tumor. The standard treatments typically include maximal surgical resection, followed adjuvant radiotherapy and chemotherapy. However, the efficacy of these treatment is often limited, as tumor often infiltrate into the surrounding brain tissue, often extending beyond the radiologically defined margins. This infiltration contributes to the high recurrence rate and poor prognosis associated with GBM patients, necessitating advanced methods for early and accurate detection of tumor infiltration. Despite the great promise traditional supervised machine learning shows in predicting tumor infiltration beyond resectable margins, these methods are heavily reliant on expert-drawn Regions of Interest (ROIs), which are used to construct multi-variate models of different Magnetic Resonance (MR) signal characteristics associated with tumor infiltration. This process is both time consuming and resource intensive. Addressing this limitation, our study proposes a novel integration of fully automatic methods for generating ROIs with deep learning algorithms to create predictive maps of tumor infiltration. This approach uses pre-operative multi-parametric MRI (mpMRI) scans, encompassing T1, T1Gd, T2, T2-FLAIR, and ADC sequences, to fully leverage the knowledge from previously drawn ROIs. Subsequently, a patch based Convolutional Neural Network (CNN) model is trained on these automatically generated ROIs to predict areas of potential tumor infiltration. The performance of this model was evaluated using a leave-one-out cross-validation approach. Generated predictive maps binarized for comparison against post-recurrence mpMRI scans. The model demonstrates robust predictive capability, evidenced by the average cross-validated accuracy of 0.87, specificity of 0.88, and sensitivity of 0.90. Notably, the odds ratio of 8.62 indicates that regions identified as high-risk on the predictive map were significantly more likely to exhibit tumor recurrence than low-risk regions. The proposed method demonstrates that a fully automatic mpMRI analysis using deep learning can successfully predict tumor infiltration in peritumoral region for GBM patients while bypassing the intensive requirement for expert-drawn ROIs.

CDPNet: a radiomic feature learning method with epigenetic application to estimating MGMT promoter methylation status in glioblastoma.

Radiomics has been widely recognized for its effectiveness in decoding tumor phenotypes through the extraction of quantitative imaging features. However, the robustness of radiomic methods to estimate clinically relevant biomarkers non-invasively remains largely untested. In this study, we propose Cascaded Data Processing Network (CDPNet), a radiomic feature learning method to predict tumor molecular status from medical images. We apply CDPNet to an epigenetic case, specifically targeting the estimation of O6-methylguanine-DNA-methyltransferase (MGMT) promoter methylation from Magnetic Resonance Imaging (MRI) scans of glioblastoma patients. CDPNet has three components: 1) Principal Component Analysis (PCA), 2) Fisher Linear Discriminant (FLD), and 3) a combination of hashing and blockwise histograms. The outlined architectural framework capitalizes on PCA to reconstruct input image patches, followed by FLD to extract discriminative filter banks, and finally using binary hashing and blockwise histogram module for indexing, pooling, and feature generation. To validate the effectiveness of CDPNet, we conducted an exhaustive evaluation on a comprehensive retrospective cohort comprising 484 IDH-wildtype glioblastoma patients with pre-operative multi-parametric MRI scans (T1, T1-Gd, T2, and T2-FLAIR). The prediction of MGMT promoter methylation status was cast as a binary classification problem. The developed model underwent rigorous training via 10-fold cross-validation on a discovery cohort of 446 patients. Subsequently, the model's performance was evaluated on a distinct and previously unseen replication cohort of 38 patients. Our method achieved an accuracy of 70.11% and an area under the curve of 0.71 (95% CI: 0.65 - 0.74).

Abstract 1175: Differential patterns of immune infiltration in the tumor immune microenvironment associate with therapeutic response in primary prostate cancer following chemohormonal therapy

Abstract There is a critical need to develop novel therapeutic strategies and diagnostic tools to precisely deliver treatments to improve survival for men with prostate cancer. To support this development, improved strategies are needed to better understand heterogenous tumor microenvironments and tumor biology that associate with variable treatment responses. We hypothesized that the tumor immune microenvironment (TIME) plays a critical role in treatment resistance. In this study we aimed to evaluate TIME signatures of treatment response and resistance utilizing a novel, integrated technological tool to identify response patterns and enable precision sampling for comparative cellular and molecular analysis. 30 patients with newly diagnosed, locally advanced, high-risk, primary prostate cancer underwent 18F-DCFPyL PSMA PET/MRI with multiparametric MRI scans followed by 3 cycles of chemohormonal therapy (NCT03358563). Repeat PSMA PET/MRI was performed prior to prostatectomy and scans were used to categorize lesions as complete response (CR), partial response (PR), no response (NR) or normal tissue. MRI scans were used to print a 3D mold of the prostate to allow microdissection of regions of interest from the resected prostate. Cellular infiltrates were analyzed by flow cytometry in 3 to 5 tissue specimens per patient. Statistical analysis was performed with One-way ANOVA with Tukey-correction. The frequency of CD8+ T cells in the total CD45+ infiltrate was highest in normal and CR areas and was significantly reduced in PR vs CR (p&amp;lt;0.01). CXCR3+CD8+ and CD103+CD8+ T cell frequencies were also reduced in PR vs CR foci (p&amp;lt;0.01, p&amp;lt;0.05, respectively). Meanwhile, the frequency of CXCR3+CD8+ T cells was highest and significantly elevated in CR vs normal tissue. A tendency of reduced CCR6+, CXCR5+, and CCR4+CD8+ T cells was observed in PR vs CR foci, while those frequencies remained higher in normal and CR areas. An increase in total CD8+ and CD103+CD8+ T cells associated with longer progression-free survival. Additionally, the analysis of EpCAM+ cells showed a significant increase in B7H3 expression in PR vs CR lesions (p&amp;lt;0.05) and a tendency of reduced HLA I expression in foci that associated with treatment resistance. We are currently integrating analysis of myeloid cells and transcriptomic analysis of sorted CD4+, CD8+ and CD11b/CD14+ cells to further dissect patterns of therapeutic response in our study cohort. In conclusion, PSMA/PET MRI based precision sampling of tumor tissue associated with differential therapeutic response patterns captured differences in the TIME infiltrates and these observations may provide hypothesis to test biological mechanisms to expedite discovery of targetable mechanisms to improve tumor stratification and targeting in high-risk prostate cancer. Citation Format: Erika Heninger, Jamie M. Sperger, Kristin Weinstein, Brian P. Johnson, Peter G. Geiger, Shane Wells, Steve Y. Cho, Wei Huang, Philippos Tsourkas, Sean McIlwain, Irene M. Ong, Sheena C. Kerr, David F. Jarrard, David J. Beebe, Joshua M. Lang. Differential patterns of immune infiltration in the tumor immune microenvironment associate with therapeutic response in primary prostate cancer following chemohormonal therapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1175.

Integrating imaging and genomic data for the discovery of distinct glioblastoma subtypes: a joint learning approach

Glioblastoma is a highly heterogeneous disease, with variations observed at both phenotypical and molecular levels. Personalized therapies would be facilitated by non-invasive in vivo approaches for characterizing this heterogeneity. In this study, we developed unsupervised joint machine learning between radiomic and genomic data, thereby identifying distinct glioblastoma subtypes. A retrospective cohort of 571 IDH-wildtype glioblastoma patients were included in the study, and pre-operative multi-parametric MRI scans and targeted next-generation sequencing (NGS) data were collected. L21-norm minimization was used to select a subset of 12 radiomic features from the MRI scans, and 13 key driver genes from the five main signal pathways most affected in glioblastoma were selected from the genomic data. Subtypes were identified using a joint learning approach called Anchor-based Partial Multi-modal Clustering on both radiomic and genomic modalities. Kaplan–Meier analysis identified three distinct glioblastoma subtypes: high-risk, medium-risk, and low-risk, based on overall survival outcome (p < 0.05, log-rank test; Hazard Ratio = 1.64, 95% CI 1.17–2.31, Cox proportional hazard model on high-risk and low-risk subtypes). The three subtypes displayed different phenotypical and molecular characteristics in terms of imaging histogram, co-occurrence of genes, and correlation between the two modalities. Our findings demonstrate the synergistic value of integrated radiomic signatures and molecular characteristics for glioblastoma subtyping. Joint learning on both modalities can aid in better understanding the molecular basis of phenotypical signatures of glioblastoma, and provide insights into the biological underpinnings of tumor formation and progression.

Emerging Trends in AI and Radiomics for Bladder, Kidney, and Prostate Cancer: A Critical Review.

This comprehensive review critically examines the transformative impact of artificial intelligence (AI) and radiomics in the diagnosis, prognosis, and management of bladder, kidney, and prostate cancers. These cutting-edge technologies are revolutionizing the landscape of cancer care, enhancing both precision and personalization in medical treatments. Our review provides an in-depth analysis of the latest advancements in AI and radiomics, with a specific focus on their roles in urological oncology. We discuss how AI and radiomics have notably improved the accuracy of diagnosis and staging in bladder cancer, especially through advanced imaging techniques like multiparametric MRI (mpMRI) and CT scans. These tools are pivotal in assessing muscle invasiveness and pathological grades, critical elements in formulating treatment plans. In the realm of kidney cancer, AI and radiomics aid in distinguishing between renal cell carcinoma (RCC) subtypes and grades. The integration of radiogenomics offers a comprehensive view of disease biology, leading to tailored therapeutic approaches. Prostate cancer diagnosis and management have also seen substantial benefits from these technologies. AI-enhanced MRI has significantly improved tumor detection and localization, thereby aiding in more effective treatment planning. The review also addresses the challenges in integrating AI and radiomics into clinical practice, such as the need for standardization, ensuring data quality, and overcoming the "black box" nature of AI. We emphasize the importance of multicentric collaborations and extensive studies to enhance the applicability and generalizability of these technologies in diverse clinical settings. In conclusion, AI and radiomics represent a major paradigm shift in oncology, offering more precise, personalized, and patient-centric approaches to cancer care. While their potential to improve diagnostic accuracy, patient outcomes, and our understanding of cancer biology is profound, challenges in clinical integration and application persist. We advocate for continued research and development in AI and radiomics, underscoring the need to address existing limitations to fully leverage their capabilities in the field of oncology.

Predicting response to neoadjuvant chemotherapy with liquid biopsies and multiparametric MRI in patients with breast cancer

Accurate prediction of response to neoadjuvant chemotherapy (NAC) can help tailor treatment to individual patients’ needs. Little is known about the combination of liquid biopsies and computer extracted features from multiparametric magnetic resonance imaging (MRI) for the prediction of NAC response in breast cancer. Here, we report on a prospective study with the aim to explore the predictive potential of this combination in adjunct to standard clinical and pathological information before, during and after NAC. The study was performed in four Dutch hospitals. Patients without metastases treated with NAC underwent 3 T multiparametric MRI scans before, during and after NAC. Liquid biopsies were obtained before every chemotherapy cycle and before surgery. Prediction models were developed using penalized linear regression to forecast residual cancer burden after NAC and evaluated for pathologic complete response (pCR) using leave-one-out-cross-validation (LOOCV). Sixty-one patients were included. Twenty-three patients (38%) achieved pCR. Most prediction models yielded the highest estimated LOOCV area under the curve (AUC) at the post-treatment timepoint. A clinical-only model including tumor grade, nodal status and receptor subtype yielded an estimated LOOCV AUC for pCR of 0.76, which increased to 0.82 by incorporating post-treatment radiological MRI assessment (i.e., the “clinical-radiological” model). The estimated LOOCV AUC was 0.84 after incorporation of computer-extracted MRI features, and 0.85 when liquid biopsy information was added instead of the radiological MRI assessment. Adding liquid biopsy information to the clinical-radiological resulted in an estimated LOOCV AUC of 0.86. In conclusion, inclusion of liquid biopsy-derived markers in clinical-radiological prediction models may have potential to improve prediction of pCR after NAC in breast cancer.

Prevalence of Bladder Cancers Incidentally Detected During Multiparametric MRI Scans of the Prostate Gland and the Clinical Significance of Scoring Them According to VI-RADS: A Pictorial Single-Centre Study.

To determine the prevalence of incidentally detected bladder cancers (BCs) on multiparametric magnetic resonance imaging (mpMRI) of the prostate and to highlight the clinical importance of scoring them according to the Vesical Imaging-Reporting and Data System (VI-RADS). VI-RADS scores for incidental bladder lesions on mpMRI of the prostate were collected in 1693 patients with elevated prostate-specific antigen but no hematuria. The study included 19 patients with 28 incidental bladder lesions. During this period, 39 incidental bladder lesions were found in 30 patients, representing 1.7% of cases. Of the 28 lesions, 11 were categorized by VI-RADS as VI-RADS 1, 14 as VI-RADS 2, 1 as VI-RADS 3, 1 as VI-RADS 4, and 1 as VI-RADS 5. Histopathological examination revealed 1 benign lesion, 24 non-muscle invasive BCs, and 3 muscle-invasive BCs in the 19 patients. Impressively, 97% of the incidental lesions detected by prostate mpMRI and categorized by VI-RADS were BCs without apparent prostate cancer invasion. Notably, 93% of these lesions were consistent with histopathological findings of muscle invasion and extravesical spread. Our study concludes the prevalence 1% incidental BC in prostate mpMRI. The research underscores a thorough bladder examination during prostate MRI scans. Utilizing mpMRI assists in distinguishing varying BC stages, aiding treatment decisions, and patient outcomes. VI-RADS categorization aligns with histopathological results, enhancing diagnosis, and healthcare communication. Early detection significantly influences patient care by enabling timely interventions and suitable treatment strategies, particularly for low-stage BCs linked to reduced progression and recurrence rates.

Standardized evaluation of the extent of resection in glioblastoma with automated early post-operative segmentation.

Standard treatment of patients with glioblastoma includes surgical resection of the tumor. The extent of resection (EOR) achieved during surgery significantly impacts prognosis and is used to stratify patients in clinical trials. In this study, we developed a U-Net-based deep-learning model to segment contrast-enhancing tumor on post-operative MRI exams taken within 72 h of resection surgery and used these segmentations to classify the EOR as either maximal or submaximal. The model was trained on 122 multiparametric MRI scans from our institution and achieved a mean Dice score of 0.52 ± 0.03 on an external dataset (n = 248), a performance -on par with the interrater agreement between expert annotators as reported in literature. We obtained an EOR classification precision/recall of 0.72/0.78 on the internal test dataset (n = 462) and 0.90/0.87 on the external dataset. Furthermore, Kaplan-Meier curves were used to compare the overall survival between patients with maximal and submaximal resection in the internal test dataset, as determined by either clinicians or the model. There was no significant difference between the survival predictions using the model's and clinical EOR classification. We find that the proposed segmentation model is capable of reliably classifying the EOR of glioblastoma tumors on early post-operative MRI scans. Moreover, we show that stratification of patients based on the model's predictions offers at least the same prognostic value as when done by clinicians.

NIMG-66. RADIOGENOMICS-BASED HETEROGENEITY ANALYSIS OF GLIOBLASTOMA REVEALS MR IMAGING PHENOTYPICAL CHARACTERIZATION OF EGFR MUTATIONS

Abstract PURPOSE EGFR is one of the most frequently altered genes in glioblastoma, while also being an attractive therapeutic target for treatment. Having in vivo, imaging-based markers of EGFR mutations, and as well as characterizing molecular heterogeneity, are important for patient management and stratification into trials. Toward this end, we present deep learning models of imaging signatures of EGFR mutations built from MRI. METHODS Our cohort consists of 286 glioblastoma patients with multi-parametric MRI (mpMRI) scans (T1, T1-Gd, T2, T2-FLAIR, DSC, DTI). Radiomics features, including histograms, morphologic and textural descriptors, were first derived. Genomic information was determined from a targeted next generation sequencing (NGS) panel. We jointly trained a variational auto-encoder and a multi-label classifier for predicting 4 key driver genes EGFR, NF1, PTEN, TP53 to learn a latent representation reflecting the molecular heterogeneity in the dataset. The latent representation was analyzed using principal component analysis (PCA). We calculated the Euclidean distance between imaging features of tumors with EGFR plus 1 of the other 3 mutations, from those of tumors having only EGFR mutations. We also analyzed the imaging signatures and variant allele frequencies (VAFs) of individuals with multiple (typically 3-4) resected tissue samples (n=49). RESULTS Co-occurrence of mutations in any of NF1, PTEN, TP53 genes with EGFR mutations resulted into dominance of imaging signatures of these other mutations over that of EGFR mutations. Interestingly, tumors with homogeneously high VAFs in EGFR mutations (n=3) exhibit strong phenotypical difference from the exclusive EGFR mutated group, while individuals with heterogeneous EGFR mutations (n=4) do not show such differences. CONCLUSION Our preliminary findings suggest EGFR imaging characteristics can be dominated by those of other co-occurring mutations. Tumors with homogeneously high VAFs, potentially indicating presence of germline EGFR mutations, lead to relatively distinct imaging phenotypes.

NIMG-45. EXTERNAL EVALUATION OF A MACHINE LEARNING MODEL EMPLOYING RADIOMICS TO IDENTIFY REGIONS OF LOCAL RECURRENCE IN GLIOBLASTOMA FROM POSTOPERATIVE MRI

Abstract The standard surgical goal for operable glioblastomas worldwide is the complete excision of the enhancing tumor. Nevertheless, the peritumoral region harbors infiltrating cells leading to recurrence. Our study aims to evaluate a predictive model designed to predict potential regions of recurrence using voxel-based radiomics analysis on postoperative magnetic resonance imaging (MRI). This retrospective, multi-institutional study involved glioblastoma patients who had undergone complete resection. The training cohort comprised 49 patients from two Spanish institutions, with model evaluation using an external dataset of 27 patients from a Norwegian institution and the public Ivy-Gap dataset. We used follow-up MRI scans for ground truth definition of recurrence, while postoperative multiparametric MRI scans provided the data for extracting voxel-based radiomic features. Our method employed an unsupervised, hybrid approach, which combined a deep neural network and instance optimization, to align the MRI sequences for each patient, registering the follow-up with the postoperative scans. To generate ground truth labels, we segmented the peritumoral region in the postoperative scans, identified as the area showing T2/FLAIR signal alterations, in addition to the enhancing tumor visible in the registered follow-up scan. Overlapping voxels between the peritumoral and enhancing tumor regions were classified as recurrence, while non-overlapping voxels were labeled as nonrecurrence. Radiomic features were subsequently extracted from the postoperative peritumoral region. We trained four machine learning classifiers to predict recurrence. Among these, the Categorical Boosting (CatBoost) classifier demonstrated excellent performance on the test dataset, achieving an average area under the curve (AUC) of 0.83 ± 0.07, and an accuracy of 0.80 ± 0.12, based on voxel-wise comparison with the ground truth. Our study resulted in a method that accurately predicts the area of future tumor recurrence in glioblastoma patients' MRI scans. This innovation could potentially tailor surgical and radiotherapy treatment strategies to these specific areas, possibly extending patient survival.