Machine learning methods for MRI biomarkers analysis of pediatric posterior fossa tumors

Deciphering Machine Learning Decisions to Distinguish between Posterior Fossa Tumor Types Using MRI Features: What Do the Data Tell Us?

Machine learning (ML) models have become capable of making critical decisions on our behalf. Nevertheless, due to complexity of these models, interpreting their decisions can be challenging, and humans cannot always control them. This paper provides explanations of decisions made by ML models in diagnosing four types of posterior fossa tumors: medulloblastoma, ependymoma, pilocytic astrocytoma, and brainstem glioma. The proposed methodology involves data analysis using kernel density estimations with Gaussian distributions to examine individual MRI features, conducting an analysis on the relationships between these features, and performing a comprehensive analysis of ML model behavior. This approach offers a simple yet informative and reliable means of identifying and validating distinguishable MRI features for the diagnosis of pediatric brain tumors. By presenting a comprehensive analysis of the responses of the four pediatric tumor types to each other and to ML models in a single source, this study aims to bridge the knowledge gap in the existing literature concerning the relationship between ML and medical outcomes. The results highlight that employing a simplistic approach in the absence of very large datasets leads to significantly more pronounced and explainable outcomes, as expected. Additionally, the study also demonstrates that the pre-analysis results consistently align with the outputs of the ML models and the clinical findings reported in the existing literature.

Machine Learning in the Classification of Pediatric Posterior Fossa Tumors: A Systematic Review

Background: Posterior fossa tumors (PFTs) are a morbid group of central nervous system tumors that most often present in childhood. While early diagnosis is critical to drive appropriate treatment, definitive diagnosis is currently only achievable through invasive tissue collection and histopathological analyses. Machine learning has been investigated as an alternative means of diagnosis. In this systematic review and meta-analysis, we evaluated the primary literature to identify all machine learning algorithms developed to classify and diagnose pediatric PFTs using imaging or molecular data. Methods: Of the 433 primary papers identified in PubMed, EMBASE, and Web of Science, 25 ultimately met the inclusion criteria. The included papers were extracted for algorithm architecture, study parameters, performance, strengths, and limitations. Results: The algorithms exhibited variable performance based on sample size, classifier(s) used, and individual tumor types being investigated. Ependymoma, medulloblastoma, and pilocytic astrocytoma were the most studied tumors with algorithm accuracies ranging from 37.5% to 94.5%. A minority of studies compared the developed algorithm to a trained neuroradiologist, with three imaging-based algorithms yielding superior performance. Common algorithm and study limitations included small sample sizes, uneven representation of individual tumor types, inconsistent performance reporting, and a lack of application in the clinical environment. Conclusions: Artificial intelligence has the potential to improve the speed and accuracy of diagnosis in this field if the right algorithm is applied to the right scenario. Work is needed to standardize outcome reporting and facilitate additional trials to allow for clinical uptake.

Differentiation of paediatric posterior fossa tumours by the multiregional and multiparametric MRI radiomics approach: a study on the selection of optimal multiple sequences and multiregions

OBJECTIVE

To evaluate the diagnostic performance of a radiomics model based on multiregional and multiparametric MRI to classify paediatric posterior fossa tumours (PPFTs), explore the contribution of different MR sequences and tumour subregions in tumour classification, and examine whether contrast-enhanced T₁ weighted (T1C) images have irreplaceable added value.

METHODS

This retrospective study of 136 PPFTs extracted 11,958 multiregional (enhanced, non-enhanced, and total tumour) features from multiparametric MRI (T₁- and T₂ weighted, T1C, fluid-attenuated inversion recovery, and diffusion-weighted images). These features were subjected to fast correlation-based feature selection and classified by a support vector machine based on different tasks. Diagnostic performances of multiregional and multiparametric MRI features, different sequences, and different tumoral regions were evaluated using multiclass and one-vs-rest strategies.

RESULTS

The established model achieved an overall area under the curve (AUC) of 0.977 in the validation cohort. The performance of PPFTs significantly improved after replacing T1C with apparent diffusion coefficient maps added into the plain scan sequences (AUC from 0.812 to 0.917). When oedema features were added to contrast-enhancing tumour volume, the performance did not significantly improve.

CONCLUSION

The radiomics model built by multiregional and multiparametric MRI features allows for the excellent distinction of different PPFTs and provides valuable references for the rational adoption of MR sequences.

ADVANCES IN KNOWLEDGE

This study emphasized that T1C has limited added value in predicting PPFTs and should be cautiously adopted. Selecting optimal MR sequences may help guide clinicians to better allocate acquisition sequences and reduce medical costs.

Applications of Artificial Intelligence in Pediatric Oncology: A Systematic Review

PURPOSE

There is a need for an improved understanding of clinical and biologic risk factors in pediatric cancer to improve patient outcomes. Machine learning (ML) represents the application of computational inference from advanced statistical methods that can be applied to increasing amount of data available for study in pediatric oncology. The goal of this systematic review was to systematically characterize the state of ML in pediatric oncology and highlight advances and opportunities in the field.

METHODS

We conducted a systematic review of the Embase, Scopus, and MEDLINE databases for applications of ML in pediatric oncology. Query results from all three databases were aggregated and duplicate studies were removed.

RESULTS

A total of 42 unique articles that examined the applications of ML in pediatric oncology met inclusion criteria for review. We identified 20 studies of CNS tumors, 13 of solid tumors, and nine of leukemia. ML tasks included classification, prediction of treatment response, and dose optimization with a variety of methods being used including neural network, k-nearest neighbor, random forest, naive Bayes, and support vector machines. Strengths of the identified studies included matching or outperforming physician comparators via automated analysis and predicting therapeutic response. Common limitations included significant heterogeneity in reporting standards, clinical applicability, small sample sizes, and missing external validation cohorts.

CONCLUSION

We identified areas where ML can enhance clinical care in ways that may not otherwise be achievable. Although ML promises enormous potential in improving diagnostics, decision making, and monitoring for children with cancer, the field remains in early stages and future work will be aided by standards and guidelines to ensure rigorous methodologic design and maximizing clinical utility.

Artificial Intelligence Applications in Pediatric Brain Tumor Imaging: A Systematic Review

OBJECTIVE

Artificial intelligence (AI) has facilitated the analysis of medical imaging given increased computational capacity and medical data availability in recent years. Although many applications for AI in the imaging of brain tumors have been proposed, their potential clinical impact remains to be explored. A systematic review was performed to examine the role of AI in the analysis of pediatric brain tumor imaging.

METHODS

PubMed, Embase, and Scopus were searched for relevant articles up to January 27, 2021.

RESULTS

Literature search identified 298 records, of which 22 studies were included. The most commonly studied tumors were posterior fossa tumors including brainstem glioma, ependymoma, medulloblastoma, and pilocytic astrocytoma (15, 68%). Tumor diagnosis was the most frequently performed task (14, 64%), followed by tumor segmentation (3, 14%) and tumor detection (3, 14%). Of the 6 studies comparing AI to clinical experts, 5 demonstrated superiority of AI for tumor diagnosis. Other tasks including tumor segmentation, attenuation correction of positron emission tomography scans, image registration for patient positioning, and dose calculation for radiotherapy were performed with high accuracy comparable with clinical experts. No studies described use of the AI tool in routine clinical practice.

CONCLUSIONS

AI methods for analysis of pediatric brain tumor imaging have increased exponentially in recent years. However, adoption of these methods in clinical practice requires further characterization of validity and utility. Implementation of these methods may streamline clinical workflows by improving diagnostic accuracy and automating basic imaging analysis tasks.

The Challenge of Choosing the Best Classification Method in Radiomic Analyses: Recommendations and Applications to Lung Cancer CT Images

Radiomics uses high-dimensional sets of imaging features to predict biological characteristics of tumors and clinical outcomes. The choice of the algorithm used to analyze radiomic features and perform predictions has a high impact on the results, thus the identification of adequate machine learning methods for radiomic applications is crucial. In this study we aim to identify suitable approaches of analysis for radiomic-based binary predictions, according to sample size, outcome balancing and the features-outcome association strength. Simulated data were obtained reproducing the correlation structure among 168 radiomic features extracted from Computed Tomography images of 270 Non-Small-Cell Lung Cancer (NSCLC) patients and the associated to lymph node status. Performances of six classifiers combined with six feature selection (FS) methods were assessed on the simulated data using AUC (Area Under the Receiver Operating Characteristics Curves), sensitivity, and specificity. For all the FS methods and regardless of the association strength, the tree-based classifiers Random Forest and Extreme Gradient Boosting obtained good performances (AUC ≥ 0.73), showing the best trade-off between sensitivity and specificity. On small samples, performances were generally lower than in large-medium samples and with larger variations. FS methods generally did not improve performances. Thus, in radiomic studies, we suggest evaluating the choice of FS and classifiers, considering specific sample size, balancing, and association strength.

Identification of Tumor-Specific MRI Biomarkers Using Machine Learning (ML)

The identification of reliable and non-invasive oncology biomarkers remains a main priority in healthcare. There are only a few biomarkers that have been approved as diagnostic for cancer. The most frequently used cancer biomarkers are derived from either biological materials or imaging data. Most cancer biomarkers suffer from a lack of high specificity. However, the latest advancements in machine learning (ML) and artificial intelligence (AI) have enabled the identification of highly predictive, disease-specific biomarkers. Such biomarkers can be used to diagnose cancer patients, to predict cancer prognosis, or even to predict treatment efficacy. Herein, we provide a summary of the current status of developing and applying Magnetic resonance imaging (MRI) biomarkers in cancer care. We focus on all aspects of MRI biomarkers, starting from MRI data collection, preprocessing and machine learning methods, and ending with summarizing the types of existing biomarkers and their clinical applications in different cancer types.