Artificial intelligence, machine learning, computer-aided diagnosis, and radiomics: advances in imaging towards to precision medicine

Radiologia Brasileira, Journal Year: 2019, Volume and Issue: 52(6), P. 387 - 396

Published: Sept. 25, 2019

Abstract The discipline of radiology and diagnostic imaging has evolved greatly in recent years. We have observed an exponential increase the number exams performed, subspecialization medical fields, increases accuracy various methods, making it a challenge for radiologist to “know everything about all regions”. In addition, are no longer only qualitative diagnostic, providing now quantitative information on disease severity, as well identifying biomarkers prognosis treatment response. view this, computer-aided diagnosis systems been developed with objective complementing helping therapeutic decision-making process. With advent artificial intelligence, “big data”, machine learning, we moving toward rapid expansion use these tools daily life physicians, each patient unique, leading concept multidisciplinary approach precision medicine. this article, will present main aspects computational currently available analysis images principles such analysis, together terms concepts involved, examining impact that development intelligence had imaging.

Language: Английский

---

Radiomics and Artificial Intelligence for Biomarker and Prediction Model Development in Oncology

Computational and Structural Biotechnology Journal, Journal Year: 2019, Volume and Issue: 17, P. 995 - 1008

Published: Jan. 1, 2019

Language: Английский

---

Performance comparison of modified ComBat for harmonization of radiomic features for multicenter studies

Scientific Reports, Journal Year: 2020, Volume and Issue: 10(1)

Published: June 24, 2020

Abstract Multicenter studies are needed to demonstrate the clinical potential value of radiomics as a prognostic tool. However, variability in scanner models, acquisition protocols and reconstruction settings unavoidable radiomic features notoriously sensitive these factors, which hinders pooling them statistical analysis. A harmonization method called ComBat was developed deal with “batch effect” gene expression microarray data used “center-effect”. Our goal evaluate modifications allowing for more flexibility choosing reference improving robustness estimation. Two modified versions were evaluated: M-ComBat allows transform all distributions chosen reference, instead overall mean, providing flexibility. B-ComBat adds bootstrap Monte Carlo improved BM-ComBat combines both modifications. The four compared regarding their ability harmonize multicenter context two different datasets. first contains 119 locally advanced cervical cancer patients from 3 centers, magnetic resonance imaging positron emission tomography imaging. In that case applied labels corresponding each center. second one 98 laryngeal 5 centers contrast-enhanced computed tomography. specific case, because highly heterogeneous even within five unsupervised clustering determine applying ComBat. impact evaluated through three machine learning pipelines modelling step predicting outcomes, across performance metrics (balanced accuracy Matthews correlation coefficient). Before harmonization, almost had significantly between labels. These differences successfully removed versions. predictive models always provided best results. This observed consistently datasets, metrics. proposed allow They also slightly but improve power resulting models.

Language: Английский

---

Histologic subtype classification of non-small cell lung cancer using PET/CT images

European Journal of Nuclear Medicine and Molecular Imaging, Journal Year: 2020, Volume and Issue: 48(2), P. 350 - 360

Published: Aug. 10, 2020

Language: Английский

---

Fusion-based tensor radiomics using reproducible features: Application to survival prediction in head and neck cancer

Computer Methods and Programs in Biomedicine, Journal Year: 2023, Volume and Issue: 240, P. 107714 - 107714

Published: July 8, 2023

Language: Английский

---

Prediction of disease-free survival by the PET/CT radiomic signature in non-small cell lung cancer patients undergoing surgery

European Journal of Nuclear Medicine and Molecular Imaging, Journal Year: 2017, Volume and Issue: 45(2), P. 207 - 217

Published: Sept. 23, 2017

Language: Английский

---

A machine learning based delta-radiomics process for early prediction of treatment response of pancreatic cancer

npj Precision Oncology, Journal Year: 2019, Volume and Issue: 3(1)

Published: Oct. 4, 2019

Changes of radiomic features over time in longitudinal images, delta radiomics, can potentially be used as a biomarker to predict treatment response. This study aims develop delta-radiomic process based on machine learning by (1) acquiring and registering (2) segmenting populating regions interest (ROIs), (3) extracting calculating their changes (delta-radiomic features, DRFs), (4) reducing feature space determining candidate DRFs showing treatment-induced changes, (5) creating outcome prediction models using learning. was demonstrated retrospectively analyzing daily non-contrast CTs acquired during routine CT-guided-chemoradiation therapy for 90 pancreatic cancer patients. A total 2520 CT sets (28-daily-fractions-per-patient) along with pathological response were analyzed. Over 1300 extracted from the segmented ROIs. Highly correlated ruled out Spearman correlations. Correlation between selected established linear-regression-models. T test linear-mixed-effects-models determine which changed significantly compared first fraction. Bayesian-regularization-neural-network build model. The model trained 50 patients leave-one-out-cross-validation. Performance judged area-under-ROC-curve. External independent validation done data remaining 40 results show that 13 passed tests significant following 2-4 weeks treatment. best performing combination differentiating good versus bad responders (CV-AUC = 0.94) obtained normalized-entropy-to-standard-deviation-difference-(NESTD), kurtosis, coarseness. With further studies larger sets, radiomics may into early

Language: Английский

---

Genomic and immune heterogeneity are associated with differential responses to therapy in melanoma

npj Genomic Medicine, Journal Year: 2017, Volume and Issue: 2(1)

Published: April 3, 2017

Appreciation for genomic and immune heterogeneity in cancer has grown though the relationship of these factors to treatment response not been thoroughly elucidated. To better understand this, we studied a large cohort melanoma patients treated with targeted therapy or checkpoint blockade (n = 60). Heterogeneity therapeutic responses via radiologic assessment was observed majority patients. Synchronous metastases were analyzed deep profiling, revealed substantial all studied, considerable diversity T cell frequency, few shared clones (<8% on average) across cohort. Variables related identified approaches through novel radiomic assessment. These data yield insight into differential melanoma, have key translational implications age precision medicine.

Language: Английский

---

Radiographic prediction of meningioma grade by semantic and radiomic features

PLoS ONE, Journal Year: 2017, Volume and Issue: 12(11), P. e0187908 - e0187908

Published: Nov. 16, 2017

Objectives The clinical management of meningioma is guided by tumor grade and biological behavior. Currently, the assessment follows surgical resection histopathologic review. Reliable techniques for pre-operative determination may enhance decision-making. Methods A total 175 patients (103 low-grade 72 high-grade) with contrast-enhanced T1-MRI were included. Fifteen radiomic (quantitative) 10 semantic (qualitative) features applied to quantify imaging phenotype. Area under curve (AUC) odd ratios (OR) computed multiple-hypothesis correction. Random-forest classifiers developed validated on an independent dataset (n = 44). Results Twelve radiographic (eight four semantic) significantly associated grade. High-grade tumors exhibited necrosis/hemorrhage (ORsem 6.6, AUCrad 0.62–0.68), intratumoral heterogeneity 7.9, 0.65), non-spherical shape (AUCrad 0.61), larger volumes 0.69) compared tumors. Radiomic sematic could predict (AUCsem 0.76 0.78). Furthermore, combining them increased classification power (AUCradio 0.86). Clinical variables alone did not effectively (AUCclin 0.65) or show complementary value data (AUCcomb 0.84). Conclusions We found a strong association between grade, ready application management. Combining qualitative quantitative improved power.

Language: Английский

---

Harmonization strategies for multicenter radiomics investigations

Physics in Medicine and Biology, Journal Year: 2020, Volume and Issue: 65(24), P. 24TR02 - 24TR02

Published: July 20, 2020

Abstract Carrying out large multicenter studies is one of the key goals to be achieved towards a faster transfer radiomics approach in clinical setting. This requires large-scale data analysis, hence need for integrating radiomic features extracted from images acquired different centers. challenging as exhibit variable sensitivity differences scanner model, acquisition protocols and reconstruction settings, which similar so-called ‘batch-effects’ genomics studies. In this review we discuss existing methods perform integration with aid reducing unwanted variation associated batch effects. We also future potential role deep learning providing solutions addressing multicentre

Language: Английский

---