Biologically-informed deep neural networks provide quantitative assessment of intratumoral heterogeneity in post-treatment glioblastoma DOI Creative Commons
Hairong Wang, Michael Argenziano, Hyunsoo Yoon

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2022, Volume and Issue: unknown

Published: Dec. 20, 2022

Abstract Intratumoral heterogeneity poses a significant challenge to the diagnosis and treatment of glioblastoma (GBM). This is further exacerbated during GBM recurrence, as treatment-induced reactive changes produce additional intratumoral that ambiguous differentiate on clinical imaging. There an urgent need develop non-invasive approaches map heterogeneous landscape histopathological alterations throughout entire lesion for each patient. We propose predictively fuse Magnetic Resonance Imaging (MRI) with underlying in recurrent using machine learning (ML) by leveraging image-localized biopsies their associated locoregional MRI features. To this end, we BioNet, biologically-informed neural network model, predict regional distributions three tissue-specific gene modules: proliferating tumor, reactive/inflammatory cells, infiltrated brain tissue. BioNet offers valuable insights into integration multiple implicit qualitative biological domain knowledge, which are challenging describe mathematical formulations. performs significantly better than range existing methods cross-validation blind test datasets. Voxel-level prediction maps modules help reveal heterogeneity, can improve surgical targeting confirmatory evaluation neuro-oncological effectiveness. The nature approach potentially facilitate regular monitoring over time, making timely therapeutic adjustment. These results also highlight emerging role ML precision medicine.

Language: Английский

A vision of 14 T MR for fundamental and clinical science DOI Creative Commons
Steven Bates, Serge O. Dumoulin,

Paul J. M. Folkers

et al.

Magnetic Resonance Materials in Physics Biology and Medicine, Journal Year: 2023, Volume and Issue: 36(2), P. 211 - 225

Published: April 10, 2023

Abstract Objective We outline our vision for a 14 Tesla MR system. This comprises novel whole-body magnet design utilizing high temperature superconductor; console and associated electronic equipment; an optimized radiofrequency coil setup proton measurement in the brain, which also has local shim capability; high-performance gradient set. Research fields The system can be considered ‘mesocope’: device capable of measuring on biologically relevant scales. In neuroscience increased spatial resolution will anatomically resolve all layers cortex, cerebellum, subcortical structures, inner nuclei. Spectroscopic imaging simultaneously measure excitatory inhibitory activity, characterizing excitation/inhibition balance neural circuits. medical research (including brain disorders) we visualize fine-grained patterns structural abnormalities relate these changes to functional molecular changes. significantly spectral make it possible detect (dynamic in) individual metabolites with pathological pathways including interactions dynamic disease processes. Conclusions offer new perspectives fundamental research. anticipate that this initiative usher era ultra-high-field MR.

Language: Английский

Citations

16
Synthetic Post-Contrast Imaging through Artificial Intelligence: Clinical Applications of Virtual and Augmented Contrast Media DOI Creative Commons
Luca Pasquini, Antonio Napolitano,

Matteo Pignatelli

et al.

Pharmaceutics, Journal Year: 2022, Volume and Issue: 14(11), P. 2378 - 2378

Published: Nov. 4, 2022

Contrast media are widely diffused in biomedical imaging, due to their relevance the diagnosis of numerous disorders. However, risk adverse reactions, concern potential damage sensitive organs, and recently described brain deposition gadolinium salts, limit use contrast clinical practice. In recent years, application artificial intelligence (AI) techniques imaging has led development ‘virtual’ ‘augmented’ contrasts. The idea behind these applications is generate synthetic post-contrast images through AI computational modeling starting from information available on other acquired during same scan. models, non-contrast (virtual contrast) or low-dose (augmented used as input data images, which often undistinguishable native ones. this review, we discuss most advances relative media.

Language: Английский

Citations

18
Advancements in Imaging and Neurosurgical Techniques for Brain Tumor Resection: A Comprehensive Review DOI Open Access

Nidhi H Vadhavekar,

Tara Sabzvari,

Simone Laguardia

et al.

Cureus, Journal Year: 2024, Volume and Issue: unknown

Published: Oct. 31, 2024

Brain tumor surgery has witnessed significant advancements over the past few decades, resulting in improved patient outcomes. Despite these advancements, brain tumors remain a formidable public health challenge due to their high morbidity and mortality rates. This review explores evolution of neurosurgical techniques for resection, emphasizing balance between minimizing invasiveness maximizing precision. Traditional approaches like craniotomy keyhole crucial, but rise minimally invasive such as endoscopic endonasal laser interstitial thermal therapy (LITT) revolutionized field. Awake been substantial stepping stone towards preservation neurological function among patients. Additionally, integration mapping technologies including intraoperative MRI, ultrasound fluorescence-guided enhanced precision resections, particularly eloquent areas. These innovations, while promising, also come with challenges, steep learning curves limited access advanced technology certain regions. As field progresses, ongoing research is essential refine improve accessibility, ultimately aiming increase survival rates preserve patients tumors. The imaging refined surgical tools, artificial intelligence (AI) planning expected further safety effectiveness procedures future. provides comprehensive analysis current strategies potential future directions surgery.

Language: Английский

Citations

1
An Update on CNS tumors in Germline Replication-Repair Deficiency (RRD) Syndromes DOI Creative Commons
Anirban Das, Ayse B. Ercan, Uri Tabori

et al.

Neuro-Oncology Advances, Journal Year: 2024, Volume and Issue: 6(1)

Published: Jan. 1, 2024

Abstract DNA replication-repair deficiency (RRD) arises from pathogenic variants in the mismatch repair and/or polymerase-proofreading genes. Multiple germline cancer predisposition syndromes children and young adults, including constitutional (CMMRD), Lynch, deficiency, rare digenic can lead to RRD cancers. The most frequent brain tumors these are high-grade gliomas. Embryonal like medulloblastoma have also been described. Lower-grade reported surveillance initiatives. latter has an extremely high rate of malignant transformation. Novel functional assays quantifying genomic microsatellite indel load demonstrated be highly sensitive specific for diagnosis cancers with CMMRD. Importantly, uniformly harbor mutation burden. High T-cell infiltration makes aggressive amenable immune checkpoint inhibition, irrespective their genetic background. Synergistic combinations successful patients failing inhibitor monotherapy. Future directions include development innovative approaches improve Additionally, use novel tools circulating tumor over time useful monitor disease burden treatment responses patients.

Language: Английский

Citations

1
A Radiologist’s Guide to IDH-Wildtype Glioblastoma for Efficient Communication With Clinicians: Part II–Essential Information on Post-Treatment Imaging DOI
Philipp Kickingereder, Philipp Karschnia, Felix Sahm

et al.

Korean Journal of Radiology, Journal Year: 2024, Volume and Issue: 26

Published: Jan. 1, 2024

Owing to recent advancements in various postoperative treatment modalities, such as radiation, chemotherapy, antiangiogenic treatment, and immunotherapy, the radiological clinical assessment of patients with isocitrate dehydrogenase-wildtype glioblastoma using post-treatment imaging has become increasingly challenging. This review highlights challenges differentiating treatment-related changes pseudoprogression, radiation necrosis, pseudoresponse from true tumor progression aims serve a guideline for efficient communication clinicians optimal management imaging.

Language: Английский

Citations

1
Introducing the American Society of Neuroradiology PET-Guided Diagnosis and Management in Neuro-Oncology Study Group DOI
Alireza Nabavizadeh, Norbert Galldiks, M. Veronesi

et al.

American Journal of Neuroradiology, Journal Year: 2024, Volume and Issue: 45(5), P. 535 - 536

Published: March 28, 2024

The American Society of Neuroradiology (ASNR) is pleased to introduce its latest initiative: the PET-Guided Diagnosis and Management in Neuro-Oncology Study Group. This group will focus on leveraging molecular imaging theranostics enhance care adult pediatric patients with primary

Language: Английский

Citations

0
Spatial immunosampling of MRI-defined glioblastoma regions reveals immunologic fingerprint of non-contrast enhancing, infiltrative tumor margins DOI Creative Commons
Matthew M. Grabowski, Dionysios C. Watson, Kunho Chung

et al.

medRxiv (Cold Spring Harbor Laboratory), Journal Year: 2023, Volume and Issue: unknown

Published: March 9, 2023

Glioblastoma (GBM) treatment includes maximal safe resection of the core and MRI contrast-enhancing (CE) tumor. Complete infiltrative non-contrast-enhancing (NCE) tumor rim is rarely achieved. We established a safe, semi-automated workflow for spatially-registered sampling MRI-defined GBM regions in 19 patients with downstream analysis biobanking, enabling studies NCE, wherefrom recurrence/progression typically occurs. Immunophenotyping revealed underrepresentation myeloid cell subsets CD8+ T cells NCE. While NCE phenotypically functionally resembled those matching CE tumor, activated (CD69hi) effector memory were overrepresented. Contrarily, CD25hi Tregs other underrepresented. Overall, our study demonstrated that MRI-guided, spatially-registered, intraoperative immunosampling feasible as part routine surgery. Further elucidation shared spatially distinct microenvironmental biology will enable development therapeutic approaches targeting to decrease recurrence.

Language: Английский

Citations

1
Biologically-informed deep neural networks provide quantitative assessment of intratumoral heterogeneity in post-treatment glioblastoma DOI Creative Commons
Jing Li, Hairong Wang, Michael Argenziano

et al.

Research Square (Research Square), Journal Year: 2024, Volume and Issue: unknown

Published: March 27, 2024

Abstract Intratumoral heterogeneity poses a significant challenge to the diagnosis and treatment of glioblastoma (GBM). This is further exacerbated during GBM recurrence, as treatment-induced reactive changes produce additional intratumoral that ambiguous differentiate on clinical imaging. There an urgent need develop non-invasive approaches map heterogeneous landscape histopathological alterations throughout entire lesion for each patient. We propose predictively fuse Magnetic Resonance Imaging (MRI) with underlying in recurrent using machine learning (ML) by leveraging image-localized biopsies their associated locoregional MRI features. To this end, we BioNet, biologically-informed neural network model, predict regional distributions three tissue-specific gene modules: proliferating tumor, reactive/inflammatory cells, infiltrated brain tissue. BioNet offers valuable insights into integration multiple implicit qualitative biological domain knowledge, which are challenging describe mathematical formulations. performs significantly better than range existing methods cross-validation blind test datasets. Voxel-level prediction maps modules help reveal heterogeneity, can improve surgical targeting confirmatory evaluation neuro-oncological effectiveness. The nature approach potentially facilitate regular monitoring over time, making timely therapeutic adjustment. These results also highlight emerging role ML precision medicine.

Language: Английский

Citations

0
The FKBP51s Splice Isoform Predicts Unfavorable Prognosis in Patients with Glioblastoma DOI Creative Commons
Carolina Giordano, Laura Marrone, Simona Romano

et al.

Cancer Research Communications, Journal Year: 2024, Volume and Issue: 4(5), P. 1296 - 1306

Published: April 23, 2024

The primary treatment for glioblastoma (GBM) is removing the tumor mass as defined by MRI. However, MRI has limited diagnostic and predictive value. Tumor-associated macrophages (TAM) are abundant in GBM microenvironment (TME) found peripheral blood (PB). FKBP51 expression, with its canonical spliced isoforms, constitutive immune cells aberrant GBM. Spliced FKBP51s supports M2 polarization. To find an immunologic signature that combined could advance diagnosis, we immunophenotyped of TME PB from 37 patients using classical M1-M2 markers. We also determined levels FKBP51s, PD-L1, HLA-DR. Tumors expressing showed increase various phenotypes regulatory T PB, indicating immunosuppression. activated STAT3 were associated reduced survival. Correlative studies tumor/macrophages cocultures allowed to interpret TAMs. Tumor volume correlated M1 infiltration TME. Cocultures spheroids produced polarization, suggesting may infiltrate alongside cancer stem cells. adherent developed phenotype CD163/FKBP51s pSTAT6, a transcription factor enabling migration invasion. In recurrences, increased counts monocyte/macrophages callosal accompanied concomitant decrease TME-infiltrating macrophages. PD-L1/FKBP51s connoted necrotic tumors. conclusion, identifies subtype significantly impairs system. Moreover, marks features glioma malignancy can aid patient monitoring.

Language: Английский

Citations

0
Biologically informed deep neural networks provide quantitative assessment of intratumoral heterogeneity in post treatment glioblastoma DOI Creative Commons
Hairong Wang, Michael Argenziano, Hyunsoo Yoon

et al.

npj Digital Medicine, Journal Year: 2024, Volume and Issue: 7(1)

Published: Oct. 19, 2024

Intratumoral heterogeneity poses a significant challenge to the diagnosis and treatment of recurrent glioblastoma. This study addresses need for non-invasive approaches map heterogeneous landscape histopathological alterations throughout entire lesion each patient. We developed BioNet, biologically-informed neural network, predict regional distributions two primary tissue-specific gene modules: proliferating tumor (Pro) reactive/inflammatory cells (Inf). BioNet significantly outperforms existing methods (p < 2e-26). In cross-validation, achieved AUCs 0.80 0.81 (Inf), with accuracies 80% 75%, respectively. blind tests, 0.76 81% 74%. Competing had lower or around 0.6 70%. BioNet's voxel-level prediction maps reveal intratumoral heterogeneity, potentially improving biopsy targeting evaluation. approach facilitates regular monitoring timely therapeutic adjustments, highlighting role ML in precision medicine.

Language: Английский

Citations

0