Radiomics and radiogenomics in gliomas: a contemporary update

British Journal of Cancer, Journal Year: 2021, Volume and Issue: 125(5), P. 641 - 657

Published: May 6, 2021

Abstract The natural history and treatment landscape of primary brain tumours are complicated by the varied tumour behaviour or secondary gliomas (high-grade transformation low-grade lesions), as well dilemmas with identification radiation necrosis, progression, pseudoprogression on MRI. Radiomics radiogenomics promise to offer precise diagnosis, predict prognosis, assess response modern chemotherapy/immunotherapy therapy. This is achieved a triumvirate morphological, textural, functional signatures, derived from high-throughput extraction quantitative voxel-level MR image metrics. However, lack standardisation acquisition parameters inconsistent methodology between working groups have made validations unreliable, hence multi-centre studies involving heterogenous study populations warranted. We elucidate novel radiomic radiogenomic workflow concepts state-of-the-art descriptors in sub-visual processing, relevant literature applications such machine learning techniques glioma management.

Language: Английский

---

Combined molecular subtyping, grading, and segmentation of glioma using multi-task deep learning

Neuro-Oncology, Journal Year: 2022, Volume and Issue: 25(2), P. 279 - 289

Published: July 5, 2022

Abstract Background Accurate characterization of glioma is crucial for clinical decision making. A delineation the tumor also desirable in initial stages but time-consuming. Previously, deep learning methods have been developed that can either non-invasively predict genetic or histological features glioma, automatically delineate tumor, not both tasks at same time. Here, we present our method molecular subtype and grade, while simultaneously providing a tumor. Methods We single multi-task convolutional neural network uses full 3D, structural, preoperative MRI scans to IDH mutation status, 1p/19q co-deletion grade segmenting trained using patient cohort containing 1508 patients from 16 institutes. tested on an independent dataset 240 13 different Results In test set, achieved IDH-AUC 0.90, AUC 0.85, 0.81 (grade II/III/IV). For delineation, mean whole Dice score 0.84. Conclusions predicts multiple, clinically relevant glioma. Evaluation shows achieves high performance it generalizes well broader population. This first-of-its-kind opens door more generalizable, instead hyper-specialized, AI methods.

Language: Английский

---

Clinical applications of artificial intelligence and radiomics in neuro-oncology imaging

Insights into Imaging, Journal Year: 2021, Volume and Issue: 12(1)

Published: Oct. 21, 2021

Abstract This article is a comprehensive review of the basic background, technique, and clinical applications artificial intelligence (AI) radiomics in field neuro-oncology. A variety AI utilized conventional advanced techniques to differentiate brain tumors from non-neoplastic lesions such as inflammatory demyelinating lesions. It used diagnosis gliomas discrimination lymphomas metastasis. Also, semiautomated automated tumor segmentation has been developed for radiotherapy planning follow-up. role grading, prediction treatment response, prognosis gliomas. Radiogenomics allowed connection imaging phenotype its molecular environment. In addition, applied assessment extra-axial pediatric with high performance detection, classification, stratification patient’s prognoses.

Language: Английский

---

Advanced imaging techniques for neuro-oncologic tumor diagnosis, with an emphasis on PET-MRI imaging of malignant brain tumors

Current Oncology Reports, Journal Year: 2021, Volume and Issue: 23(3)

Published: Feb. 18, 2021

Abstract Purpose of Review This review will explore the latest in advanced imaging techniques, with a focus on complementary nature multiparametric, multimodality using magnetic resonance (MRI) and positron emission tomography (PET). Recent Findings Advanced MRI techniques including perfusion-weighted (PWI), MR spectroscopy (MRS), diffusion-weighted (DWI), chemical exchange saturation transfer (CEST) offer significant advantages over conventional when evaluating tumor extent, predicting grade, assessing treatment response. PET performed addition to provides information regarding metabolic properties, particularly simultaneously. 18 F-fluoroethyltyrosine (FET) improves specificity diagnosis evaluation post-treatment changes. Incorporation radiogenomics machine learning methods further improve imaging. Summary The combining across modalities for brain incorporating technologies such as has potential reshape landscape neuro-oncology.

Language: Английский

---

Quantitative MRI-based radiomics for noninvasively predicting molecular subtypes and survival in glioma patients

npj Precision Oncology, Journal Year: 2021, Volume and Issue: 5(1)

Published: July 26, 2021

Abstract Gliomas can be classified into five molecular groups based on the status of IDH mutation, 1p/19q codeletion, and TERT promoter whereas they need to obtained by biopsy or surgery. Thus, we aimed use MRI-based radiomics noninvasively predict assess their prognostic value. We retrospectively identified 357 patients with gliomas extracted radiomic features from preoperative MRI images. Single-layered signatures were generated using a single MR sequence Bayesian-regularization neural networks. Image fusion models built combing significant signatures. By separately predicting markers, predictive obtained. Prognostic nomograms developed clinicopathologic data progression-free survival (PFS) overall (OS). The results showed that image model incorporating contrast-enhanced T1-weighted imaging (cT1WI) apparent diffusion coefficient (ADC) achieved an AUC 0.884 0.669 for status, respectively. cT1WI-based signature alone yielded favorable performance in (AUC = 0.815). comparable actual ones PFS (C-index: 0.709 vs. 0.722, P 0.241) OS 0.703 0.751, 0.359). Subgroup analyses grades similar findings. C-index 0.736 0.735 OS, Accordingly, may useful detecting regardless grades.

Language: Английский

---

A deep learning framework integrating MRI image preprocessing methods for brain tumor segmentation and classification

IBRO Neuroscience Reports, Journal Year: 2022, Volume and Issue: 13, P. 523 - 532

Published: Nov. 7, 2022

Glioma grading is critical in treatment planning and prognosis. This study aims to address this issue through MRI-based classification develop an accurate model for glioma diagnosis. Here, we employed a deep learning pipeline with three essential steps: (1) MRI images were segmented using preprocessing approaches UNet architecture, (2) brain tumor regions extracted segmentation, then (3) high-grade gliomas low-grade classified the VGG GoogleNet implementations. Among additional techniques used conjunction segmentation task, combination of data augmentation Window Setting Optimization was found be most effective tool, resulting Dice coefficient 0.82, 0.91, 0.72 enhancing tumor, whole core, respectively. While proposed models achieve comparable accuracies about 93 % on testing dataset, combined obtains highest accuracy 97.44 %. In conclusion, presented architecture illustrates realistic detecting gliomas; moreover, it emphasizes significance improving performance.

Language: Английский

---

Predictive Modeling in Medicine

Encyclopedia, Journal Year: 2023, Volume and Issue: 3(2), P. 590 - 601

Published: May 11, 2023

Predictive modeling is a complex methodology that involves leveraging advanced mathematical and computational techniques to forecast future occurrences or outcomes. This tool has numerous applications in medicine, yet its full potential remains untapped within this field. Therefore, it imperative delve deeper into the benefits drawbacks associated with utilizing predictive medicine for more comprehensive understanding of how approach may be effectively leveraged improved patient care. When implemented successfully, yielded impressive results across various medical specialities. From predicting disease progression identifying high-risk patients who require early intervention, there are countless examples successful implementations healthcare settings worldwide. However, despite these successes, significant challenges remain practitioners when applying models real-world scenarios. These issues include concerns about data quality availability as well navigating regulatory requirements surrounding use sensitive information—all factors can impede progress toward realizing true impact on improving health

Language: Английский

---

Recapitulating the Key Advances in the Diagnosis and Prognosis of High-Grade Gliomas: Second Half of 2021 Update

International Journal of Molecular Sciences, Journal Year: 2023, Volume and Issue: 24(7), P. 6375 - 6375

Published: March 28, 2023

High-grade gliomas (World Health Organization grades III and IV) are the most frequent fatal brain tumors, with median overall survivals of 24–72 14–16 months, respectively. We reviewed progress in diagnosis prognosis high-grade published second half 2021. A literature search was performed PubMed using general terms “radio* gliom*” a time limit from 1 July 2021 to 31 December Important advances were provided both imaging non-imaging diagnoses these hard-to-treat cancers. Our prognostic capacity also increased during This review article demonstrates slow, but steady improvements, scientifically technically, which express an chance that patients may be correctly diagnosed without invasive procedures. The those strictly depends on final results complex diagnostic process, widely varying survival rates.

Language: Английский

---

A Pipeline for the Implementation and Visualization of Explainable Machine Learning for Medical Imaging Using Radiomics Features

Sensors, Journal Year: 2022, Volume and Issue: 22(14), P. 5205 - 5205

Published: July 12, 2022

Machine learning (ML) models have been shown to predict the presence of clinical factors from medical imaging with remarkable accuracy. However, these complex can be difficult interpret and are often criticized as "black boxes". Prediction that provide no insight into how their predictions obtained trust for making important decisions, such diagnoses or treatment. Explainable machine (XML) methods, Shapley values, made it possible explain behavior ML algorithms identify which predictors contribute most a prediction. Incorporating XML methods software tools has potential increase in ML-powered aid physicians decisions. Specifically, field analysis used explaining deep learning-based model saliency maps highlight areas an image. they do not straightforward interpretation qualities image area important. Here, we describe novel pipeline uses radiomics data values outcome prediction built well-defined predictors. We present visualization results clinician-focused dashboard generalized various settings. demonstrate use this workflow developing using MRI glioma patients genetic mutation.

Language: Английский

---

Neuroprotective Potential of Aromatic Herbs: Rosemary, Sage, and Lavender

Frontiers in Neuroscience, Journal Year: 2022, Volume and Issue: 16

Published: June 28, 2022

Hundreds of millions people around the world suffer from neurological disorders or have experienced them intermittently, which has significantly reduced their quality life. The common treatments for are relatively expensive and may lead to a wide variety side effects including sleep attacks, gastrointestinal effects, blood pressure changes, etc. On other hand, several herbal medications attracted colossal popularity worldwide in recent years due availability, affordable prices, few effects. Aromatic plants, sage ( Salvia officinalis ), lavender Lavandula angustifolia rosemary Rosmarinus ) already shown anxiolytics, anti-inflammatory, antioxidant, neuroprotective They also potential treating disorders, Alzheimer's disease, Parkinson's migraine, cognitive disorders. This review summarizes data on aromatic herbs, sage, lavender, rosemary.

Language: Английский

---