scholarly journals Longitudinal structural and perfusion MRI enhanced by machine learning outperforms standalone modalities and radiological expertise in high-grade glioma surveillance

2021 ◽  
Author(s):  
Loizos Siakallis ◽  
Carole H. Sudre ◽  
Paul Mulholland ◽  
Naomi Fersht ◽  
Jeremy Rees ◽  
...  
Keyword(s):  
Machine Learning ◽  
Progressive Disease ◽  
High Grade Glioma ◽  
Perfusion Mri ◽  
Svm Classifier ◽  
High Grade ◽  
Time Points ◽  
Group I ◽  
Sensitivity Specificity ◽  
Time Point

Abstract Purpose Surveillance of patients with high-grade glioma (HGG) and identification of disease progression remain a major challenge in neurooncology. This study aimed to develop a support vector machine (SVM) classifier, employing combined longitudinal structural and perfusion MRI studies, to classify between stable disease, pseudoprogression and progressive disease (3-class problem). Methods Study participants were separated into two groups: group I (total cohort: 64 patients) with a single DSC time point and group II (19 patients) with longitudinal DSC time points (2-3). We retrospectively analysed 269 structural MRI and 92 dynamic susceptibility contrast perfusion (DSC) MRI scans. The SVM classifier was trained using all available MRI studies for each group. Classification accuracy was assessed for different feature dataset and time point combinations and compared to radiologists’ classifications. Results SVM classification based on combined perfusion and structural features outperformed radiologists’ classification across all groups. For the identification of progressive disease, use of combined features and longitudinal DSC time points improved classification performance (lowest error rate 1.6%). Optimal performance was observed in group II (multiple time points) with SVM sensitivity/specificity/accuracy of 100/91.67/94.7% (first time point analysis) and 85.71/100/94.7% (longitudinal analysis), compared to 60/78/68% and 70/90/84.2% for the respective radiologist classifications. In group I (single time point), the SVM classifier also outperformed radiologists’ classifications with sensitivity/specificity/accuracy of 86.49/75.00/81.53% (SVM) compared to 75.7/68.9/73.84% (radiologists). Conclusion Our results indicate that utilisation of a machine learning (SVM) classifier based on analysis of longitudinal perfusion time points and combined structural and perfusion features significantly enhances classification outcome (p value= 0.0001).

Combined structural and perfusion MRI enhanced by machine learning may outperform standalone modalities and radiological expertise in high-grade glioma surveillance: A proof-of-concept study.

2020 ◽  
Vol 38 (15_suppl) ◽  
pp. e14528-e14528
Author(s):  
Diana Roettger ◽  
Loizos Siakallis ◽  
Carole Sudre ◽  
Jasmina Panovska-Griffiths ◽  
Paul Mulholland ◽  
...  
Keyword(s):  
Machine Learning ◽  
High Grade Glioma ◽  
Treatment Monitoring ◽  
Perfusion Mri ◽  
Svm Classifier ◽  
High Grade ◽  
Time Points ◽  
Group I ◽  
Sensitivity Specificity ◽  
Time Point

e14528 Background: Treatment monitoring in patients with High-Grade Glioma (HGG) and identification of disease progression, remains a major challenge in clinical neurooncology. We aimed to develop a support vector machine (SVM) classifier utilising combined longitudinal conventional and Dynamic Susceptibility Contrast (DSC) perfusion MRI to classify between Stable Disease (SD), Pseudoprogression (PsP) and Progressive Disease (PD) in glioma patients under surveillance. Methods: Conventional (269) and perfusion (62) MRI studies of HGG patients acquired between 2012 and 2018 were prospectively analysed. Study participants were separated into two groups: Group I with a single DSC time point (64 participants) and Group II with multiple DSC time points (19 participants). The SVM classifier was trained using all available MRI for each group. Classification accuracy was assessed for the use of features extracted from different feature dataset and time point combinations and compared to the experienced radiologists’ predictions. Results: The study included 64 participants (mean age: 48.5 ± 12.8 yrs [standard deviation], 24 female). SVM classification based on combined perfusion and structural features outperformed standalone datasets across all groups. For the clinically relevant classification step (SD/PSP vs PD), both feature combination as well as the addition of multiple DSC time points, improved classification performance (lowest median error rate: 0.016). The SVM algorithm outperformed radiologists in predicting lesion destiny in both groups. Optimal performance was observed in Group II, in which SVM sensitivity/specificity/accuracy was 100/91.67/94.7% for analysis based on the first time point and 85.71/100/ 94.7% based on multiple time points compared to 60/78/68% and 70/90/84.2% for the respective radiologist classifications. In Group I, the SVM also exceeded radiologist predictions, albeit by a smaller margin and resulted in sensitivity/specificity of 86.49/75.00/81.53% (SVM) compared to 75.7/68.9/73.84% (radiologists). Conclusions: Our results indicate that the addition of multiple longitudinal perfusion time points as well as the combination of structural and perfusion features significantly enhance classification outcome in treatment monitoring of HGGs and machine-learning-assisted diagnosis has potentially superior accuracy than the radiologist's visual evaluation and expertise.

Comparison Between 18F-Dopa and 18F-Fet PET/CT in Patients with Suspicious Recurrent High Grade Glioma: A Literature Review and Our Experience

2019 ◽  
Vol 12 (3) ◽  
pp. 220-228 ◽  
Author(s):  
Laura Evangelista ◽  
Lea Cuppari ◽  
Luisa Bellu ◽  
Daniele Bertin ◽  
Mario Caccese ◽  
...  
Keyword(s):  
Case Series ◽  
High Grade Glioma ◽  
High Grade ◽  
True Negative ◽  
Fet Pet ◽  
Positron Emission ◽  
Pet Ct ◽  
Magnetic Resonance Imaging Mri ◽  
Definition Of ◽  
Sensitivity Specificity

Purpose: The aims of the present study were to: 1- critically assess the utility of L-3,4- dihydroxy-6-18Ffluoro-phenyl-alanine (18F-DOPA) and O-(2-18F-fluoroethyl)-L-tyrosine (18F-FET) Positron Emission Tomography (PET)/Computed Tomography (CT) in patients with high grade glioma (HGG) and 2- describe the results of 18F-DOPA and 18F-FET PET/CT in a case series of patients with recurrent HGG. Methods: We searched for studies using the following databases: PubMed, Web of Science and Scopus. The search terms were: glioma OR brain neoplasm and DOPA OR DOPA PET OR DOPA PET/CT and FET OR FET PET OR FET PET/CT. From a mono-institutional database, we retrospectively analyzed the 18F-DOPA and 18F-FET PET/CT of 29 patients (age: 56 ± 12 years) with suspicious for recurrent HGG. All patients underwent 18F-DOPA or 18F-FET PET/CT for a multidisciplinary decision. The final definition of recurrence was made by magnetic resonance imaging (MRI) and/or multidisciplinary decision, mainly based on the clinical data. Results: Fifty-one articles were found, of which 49 were discarded, therefore 2 studies were finally selected. In both the studies, 18F-DOPA and 18F-FET as exchangeable in clinical practice particularly for HGG patients. From our institutional experience, in 29 patients, we found that sensitivity, specificity and accuracy of 18F-DOPA PET/CT in HGG were 100% (95% confidence interval- 95%CI - 81-100%), 63% (95%CI: 39-82%) and 62% (95%CI: 39-81%), respectively. 18F-FET PET/CT was true positive in 4 and true negative in 4 patients. Sensitivity, specificity and accuracy for 18F-FET PET/CT in HGG were 100%. Conclusion: 18F-DOPA and 18F-FET PET/CT have a similar diagnostic accuracy in patients with recurrent HGG. However, 18F-DOPA PET/CT could be affected by inflammation conditions (false positive) that can alter the final results. Large comparative trials are warranted in order to better understand the utility of 18F-DOPA or 18F-FET PET/CT in patients with HGG.

Intraarterial delivery of bevacizumab and cetuximab utilizing blood-brain barrier disruption in children with high-grade glioma and diffuse intrinsic pontine glioma: results of a phase I trial

2021 ◽  
pp. 1-9
Author(s):  
Heather J. McCrea ◽  
Jana Ivanidze ◽  
Ashley O’Connor ◽  
Eliza H. Hersh ◽  
John A. Boockvar ◽  
...  
Keyword(s):  
Phase I ◽  
Blood Brain Barrier ◽  
Stable Disease ◽  
Phase I Trial ◽  
Progressive Disease ◽  
High Grade Glioma ◽  
Diffuse Intrinsic Pontine Glioma ◽  
Brain Barrier ◽  
High Grade ◽  
Pontine Glioma

OBJECTIVE Delivery of drugs intraarterially to brain tumors has been demonstrated in adults. In this study, the authors initiated a phase I trial of superselective intraarterial cerebral infusion (SIACI) of bevacizumab and cetuximab in pediatric patients with refractory high-grade glioma (diffuse intrinsic pontine glioma [DIPG] and glioblastoma) to determine the safety and efficacy in this population. METHODS SIACI was used to deliver mannitol (12.5 ml of 20% mannitol) to disrupt the blood-brain barrier (BBB), followed by bevacizumab (15 mg/kg) and cetuximab (200 mg/m2) to target VEGF and EGFR, respectively. Patients with brainstem tumors had a balloon inflated in the distal basilar artery during mannitol infusion. RESULTS Thirteen patients were treated (10 with DIPG and 3 with high-grade glioma). Toxicities included grade I epistaxis (2 patients) and grade I rash (2 patients). There were no dose-limiting toxicities. Of the 10 symptomatic patients, 6 exhibited subjective improvement; 92% showed decreased enhancement on day 1 posttreatment MRI. Of 10 patients who underwent MRI at 1 month, 5 had progressive disease and 5 had stable disease on FLAIR, whereas contrast-enhanced scans demonstrated progressive disease in 4 patients, stable disease in 2, partial response in 2, and complete response in 1. The mean overall survival for the 10 DIPG patients was 519 days (17.3 months), with a mean posttreatment survival of 214.8 days (7.2 months). CONCLUSIONS SIACI of bevacizumab and cetuximab was well tolerated in all 13 children. The authors’ results demonstrate safety of this method and warrant further study to determine efficacy. As molecular targets are clarified, novel means of bypassing the BBB, such as intraarterial therapy and convection-enhanced delivery, become more critical. Clinical trial registration no.: NCT01884740 (clinicaltrials.gov)

scholarly journals Using Support Vector Machine Detection of Breast Cancer in Early stage

2020 ◽  
pp. 213-216
Keyword(s):  
Breast Cancer ◽  
Machine Learning ◽  
Support Vector Machine ◽  
Early Stage ◽  
Breast Cancer Diagnosis ◽  
Support Vector ◽  
Svm Classifier ◽  
K Nearest Neighbors ◽  
Data Set ◽  
Sensitivity Specificity

The Breast Cancer is disease which tremendously increased in women’s nowadays. Mammography is technique of low-powered X-ray diagnosis approach for detection and diagnosis of cancer diseases at early stage. The proposed system shows the solution of two problems. First shows to detect tumors as suspicious regions with a weak contrast to their background and second shows way to extract features which categorize tumors. Hence this classification can be done with SVM, a great method of statistical learning has made significant achievement in various field. Discovered in the early 90’s, which led to an interest in machine learning? Here the different types of tumor like Benign, Malignant, or Normal image are classified using the SVM classifier. This techniques shows how easily we can detect region of tumor is present in mammogram images with more than 80% of accuracy rates for linear classification using SVM. The 10-fold cross validation to get an accurate outcome is been used by proposed system. The Wisconsin breast cancer diagnosis data set is referred from UCI machine learning repository. The considering accuracy, sensitivity, specificity, false discovery rate, false omission rate and Matthews’s correlation coefficient is appraised in the proposed system. This Provides good result for both training and testing phase. The techniques also shows accuracy of 98.57% and 97.14% by use of Support Vector Machine and K-Nearest Neighbors

scholarly journals PC3 - 219 Evaluation of delayed FDG-PET in differentiating progressive disease from post-treatment radiation effect in brain tumors

2016 ◽  
Vol 43 (S4) ◽  
pp. S23-S23
Author(s):  
R.A. Harrison ◽  
A. Masood ◽  
O. Mawlawi ◽  
D. Schellingerhout ◽  
B.A. Chasen ◽  
...  
Keyword(s):  
Brain Tumors ◽  
Fdg Pet ◽  
Progressive Disease ◽  
Radiation Effect ◽  
Post Treatment ◽  
Delayed Imaging ◽  
Time Points ◽  
Delayed Time ◽  
Time Point

Traditional and advanced magnetic resonance imaging techniques are often unable to differentiate progressive central nervous system neoplasm from post-treatment radiation effect (PTRE). 18F-fluorodeoxyglucose positron emission tomography (FDG-PET) with delayed imaging has been shown to increase the specificity of PET imaging for cerebral neoplasm in small studies. We sought to further evaluate the potential diagnostic benefits of delayed imaging at 5 hours versus standard imaging at 1 hour to differentiate progressive disease (PD) from PTRE in patients with primary or metastatic brain tumors treated with radiation therapy.Ten patients with primary (n=4) and metastatic (n=6) brain tumors were identified, with diagnostic confirmation of PD or PTRE provided by pathology or>3 month clinical and radiographic follow-up. Maximum standard uptake values (SUV) were calculated for suspicious areas of abnormal contrast enhancement (lesion) and compared to contralateral normal appearing brain (background) at both early and delayed time points. Seven patients were classified as having PD and 3 as having PTRE based pathology or clinical/radiographic follow up. The average lesion to background ratio (L/B) at the early time point (1.16+0.50) was significantly different than L/B for the later time point (1.72+1.10), p=0.030. The mean L/B for PD was 2.17+1.01 at the later time point compared to 0.65+0.06 for PTRE (p=0.010). For the earlier time point, L/B for PD was 1.40+0.42, compared to the L/B for PTRE which was 0.61+0.10 (p=0.003).L/B ratios at early and delayed time points successfully differentiated between patients with PD and PTRE, with significantly greater L/B ratios seen at delayed time points. These initial results are promising and further investigation is underway to evaluate the contribution of delayed imaging in differentiating PD from PTRE.

Sign in / Sign up

Export Citation Format

Share Document