Machine Learning in Differentiating Gliomas from Primary CNS Lymphomas: A Systematic Review, Reporting Quality, and Risk of Bias Assessment

Abstract

BACKGROUND: Differentiating gliomas and primary CNS lymphoma represents a diagnostic challenge with important therapeutic ramifications. Biopsy is the preferred method of diagnosis, while MR imaging in conjunction with machine learning has shown promising results in differentiating these tumors. PURPOSE: Our aim was to evaluate the quality of reporting and risk of bias, assess data bases with which the machine learning classification algorithms were developed, the algorithms themselves, and their performance. DATA SOURCES: Ovid EMBASE, Ovid MEDLINE, Cochrane Central Register of Controlled Trials, and the Web of Science Core Collection were searched according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.STUDY SELECTION: From 11,727 studies, 23 peer-reviewed studies used machine learning to differentiate primary CNS lymphoma from gliomas in 2276 patients.DATA ANALYSIS: Characteristics of data sets and machine learning algorithms were extracted. A meta-analysis on a subset of studies was performed. Reporting quality and risk of bias were assessed using the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) and Prediction Model Study Risk Of Bias Assessment Tool. DATA SYNTHESIS:The highest area under the receiver operating characteristic curve (0.961) and accuracy (91.2%) in external validation were achieved by logistic regression and support vector machines models using conventional radiomic features. Meta-analysis of machine learning classifiers using these features yielded a mean area under the receiver operating characteristic curve of 0.944 (95% CI, 0.898-0.99). The median TRIPOD score was 51.7%. The risk of bias was high for 16 studies.LIMITATIONS: Exclusion of abstracts decreased the sensitivity in evaluating all published studies. Meta-analysis had high heterogeneity.CONCLUSIONS: Machine learning-based methods of differentiating primary CNS lymphoma from gliomas have shown great potential, but most studies lack large, balanced data sets and external validation. Assessment of the studies identified multiple deficiencies in reporting quality and risk of bias. These factors reduce the generalizability and reproducibility of the findings.ABBREVIATIONS: AI ¼ artificial intelligence; AUC ¼ area under the receiver operating characteristic curve; CNN ¼ convolutional neural network; ML ¼ machine learning; PCNSL ¼ primary CNS lymphoma; PRISMA ¼ Preferred Reporting Items for Systematic Reviews and Meta-Analyses; PROBAST ¼ Prediction model study Risk Of Bias Assessment Tool; TRIPOD ¼ Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis G liomas are the most common primary malignancy of the CNS. 1 An important differential diagnosis for gliomas is primary CNS lymphoma (PCNSL), a more uncommon but highly malignant neoplasia. 2 Correct differentiation of these tumor entities is an important challenge for clinicians because

Classifying primary central nervous system lymphoma from glioblastoma using deep learning and radiomics based machine learning approach - a systematic review and meta-analysis

“…Although radiomics and deep learning algorithms have been used for a multitude of neurological conditions ( 31 – 34 ) its use in classifying malignant conditions and differentiating them is of paramount importance as the therapy and prognosis changes across the spectrum of brain tumours. The present study highlights the use of ML and DL algorithms for discriminating PCNSL and GBM on radiological imaging.…”

“…However, these positive results must be interpreted with caution as a multitude of factors such as small sample size, heterogenous imaging protocols, patient selection criteria into the training and the validation set may have led to overfitting of the data at the time of model development. Overfitting is common in radiomic studies involving machine learning and deep earning classifiers that reduces its potential for immediate incorporation into clinical practise and use it for treatment decisions ( 29 – 31 ).…”

“…A recent systematic review from 23 studies also investigated the role of DL & ML in differentiating PCNSL from all grades (Grade-II-IV) of Gliomas ( 31 ). However, significant differences exist in the methodology and the search strategy of our metaanalysis.…”

Role of Positron Emission Tomography in Primary Central Nervous System Lymphoma

“…To further improve radiological diagnosis, machine learning has recently emerged as an important tool. Petersen et al conducted a meta-analysis of 23 papers, including 3 papers on 18 F-FDG, that published machine learning-based classification algorithms to distinguish between HGG and PCNSL [ 39 ]. The machine learning models appear to have high accuracy as the algorithms were able to replicate the results of a senior subspecialty-trained radiologist.…”

Machine Learning Tools for Image-Based Glioma Grading and the Quality of Their Reporting: Challenges and Opportunities

“…As an image interpretation support tool, ML importantly may improve diagnostic performance [ 17 , 18 ]. Prior works demonstrate that AI alone can approach the diagnostic accuracy of neuroradiologists and other sub-specialty radiologists [ 19 , 20 , 21 ].…”