Dataset Information

Predicting pediatric optic pathway glioma progression using advanced magnetic resonance image analysis and machine learning.

ABSTRACT: Background:Optic pathway gliomas (OPGs) are low-grade tumors of the white matter of the visual system with a highly variable clinical course. The aim of the study was to generate a magnetic resonance imaging (MRI)-based predictive model of OPG tumor progression using advanced image analysis and machine learning techniques. Methods:We performed a retrospective case-control study of OPG patients managed between 2009 and 2015 at an academic children's hospital. Progression was defined as radiographic tumor growth or vision decline. To generate the model, optic nerves were manually highlighted and optic radiations (ORs) were segmented using diffusion tractography tools. For each patient, intensity distributions were obtained from within the segmented regions on all imaging sequences, including derivatives of diffusion tensor imaging (DTI). A machine learning algorithm determined the combination of features most predictive of progression. Results:Nineteen OPG patients with progression were matched to 19 OPG patients without progression. The mean time between most recent follow-up and most recently analyzed MRI was 3.5 ± 1.7 years. Eighty-three MRI studies and 532 extracted features were included. The predictive model achieved an accuracy of 86%, sensitivity of 89%, and specificity of 81%. Fractional anisotropy of the ORs was among the most predictive features (area under the curve 0.83, P < 0.05). Conclusions:Our findings show that image analysis and machine learning can be applied to OPGs to generate a MRI-based predictive model with high accuracy. As OPGs grow along the visual pathway, the most predictive features relate to white matter changes as detected by DTI, especially within ORs.

SUBMITTER: Pisapia JM

PROVIDER: S-EPMC7455885 | biostudies-literature | 2020 Jan-Dec

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Predicting pediatric optic pathway glioma progression using advanced magnetic resonance image analysis and machine learning.

Pisapia Jared M JM Akbari Hamed H Rozycki Martin M Thawani Jayesh P JP Storm Phillip B PB Avery Robert A RA Vossough Arastoo A Fisher Michael J MJ Heuer Gregory G GG Davatzikos Christos C

Neuro-oncology advances 20200101 1

<h4>Background</h4>Optic pathway gliomas (OPGs) are low-grade tumors of the white matter of the visual system with a highly variable clinical course. The aim of the study was to generate a magnetic resonance imaging (MRI)-based predictive model of OPG tumor progression using advanced image analysis and machine learning techniques.<h4>Methods</h4>We performed a retrospective case-control study of OPG patients managed between 2009 and 2015 at an academic children's hospital. Progression was define ...[more]

PMID: 32885166

Similar Datasets

Project description:BACKGROUND:A recent retrospective study confirmed that hepatic stiffness and splenic stiffness measured with magnetic resonance elastography (MRE) are strongly associated with the presence of esophageal varices. In addition, strong correlations have been reported between splenic stiffness values measured with MRE and hepatic venous pressure gradients in animal models. However, most studies have been conducted on adult populations, and previous pediatric MRE studies have only demonstrated the feasibility of MRE in pediatric populations, while the actual clinical application of spleen MRE has been limited. AIM:To assess the utility of splenic stiffness measurements by MRE to predict gastroesophageal varices in children. METHODS:We retrospectively reviewed abdominal MRE images taken on a 3T system in pediatric patients. Patients who had undergone Kasai operations for biliary atresia were selected for the Kasai group, and patients with normal livers and spleens were selected for the control group. Two-dimensional spin-echo echo-planar MRE acquisition centered on the liver, with a pneumatic driver at 60 Hz and a low amplitude, was performed to obtain hepatic and splenic stiffness values. Laboratory results for aspartate aminotransferase to platelet ratio index (APRI) were evaluated within six months of MRE, and the normalized spleen size ratio was determined with the upper normal size limit. All Kasai group patients underwent gastroesophageal endoscopy during routine follow-up. The Mann-Whitney U test, Kendall's tau b correlation and diagnostic performance analysis using the area under the curve (AUC) were performed for statistical analysis. RESULTS:The median spleen MRE value was 5.5 kPa in the control group (n = 9, age 9-18 years, range 4.7-6.4 kPa) and 8.6 kPa in the Kasai group (n = 22, age 4-18 years, range 5.0-17.8 kPa). In the Kasai group, the APRI, spleen size ratio and spleen MRE values were higher in patients with portal hypertension (n = 11) than in patients without (n = 11) (all P < 0.001) and in patients with gastroesophageal varices (n = 6) than in patients without (n = 16) (all P < 0.05), even though their liver MRE values were not different. The APRI (? = 0.477, P = 0.007), spleen size ratio (? = 0.401, P = 0.024) and spleen MRE values (? = 0.426, P = 0.016) also correlated with varices grades. The AUC in predicting gastroesophageal varices was 0.844 at a cut-off of 0.65 (100% sensitivity and 75% specificity) for the APRI, and 0.844 at a cut-off of 9.9 kPa (83.3% sensitivity and 81.3% specificity) for spleen MRE values. CONCLUSION:At a cut-off of 9.9 kPa, spleen MRE values predicted gastroesophageal varices as well as the APRI and spleen size ratio in biliary atresia patients after the Kasai operation. However, liver MRE values were not useful for this purpose.

Project description:This study was designed to evaluate proton magnetic resonance spectroscopy ((1)H-MRS) for monitoring the WHO grade II glioma (low-grade glioma (LGG)) treated with temozolomide (TMZ).This prospective study included adult patients with progressive LGG that was confirmed by magnetic resonance imaging (MRI). Temozolomide was administered at every 28 days. Response to TMZ was evaluated by monthly MRI examinations that included MRI with volumetric calculations and (1)H-MRS for assessing Cho/Cr and Cho/NAA ratios. Univariate, multivariate and receiver-operating characteristic statistical analyses were performed on the results.A total of 21 LGGs from 31 patients were included in the study, and followed for at least n=14 months during treatment. A total of 18 (86%) patients experienced a decrease in tumour volume with a greater decrease of metabolic ratios. Subsequently, five (28%) of these tumours resumed growth despite the continuation of TMZ administration with an earlier increase of metabolic ratios of 2 months. Three (14%) patients did not show any volume or metabolic change. The evolutions of the metabolic ratios, mean(Cho/Cr)(n) and mean(Cho/NAA)(n), were significantly correlated over time (Spearman ρ=+0.95) and followed a logarithmic regression (P>0.001). The evolutions over time of metabolic ratios, mean(Cho/Cr)(n) and mean(Cho/NAA)(n), were significantly correlated with the evolution of the mean relative decrease of tumour volume, mean(ΔV(n)/V(o)), according to a linear regression (P<0.001) in the 'response/no relapse' patient group, and with the evolution of the mean tumour volume (meanV(n)), according to an exponential regression (P<0.001) in the 'response/relapse' patient group. The mean relative decrease of metabolic ratio, mean(Δ(Cho/Cr)(n)/(Cho/Cr)(o)), at n=3 months was predictive of tumour response over the 14 months of follow-up. The mean relative change between metabolic ratios, mean((Cho/NAA)(n)-(Cho/Cr)(n))/(Cho/NAA)(n), at n=4 months was predictive of tumour relapse with a significant cutoff of 0.046, a sensitivity of 60% and a specificity of 100% (P=0.004).The (1)H-MRS profile changes more widely and rapidly than tumour volume during the response and relapse phases, and represents an early predictive factor of outcome over 14 months of follow-up. Thus, (1)H-MRS may be a promising, non-invasive tool for predicting and monitoring the clinical response to TMZ.

Dataset Information

Predicting pediatric optic pathway glioma progression using advanced magnetic resonance image analysis and machine learning.

Publications

Predicting pediatric optic pathway glioma progression using advanced magnetic resonance image analysis and machine learning.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets