About this Research Topic
Over the course of the last decade, the field of neuroimaging has witnessed considerable advancement in introduction of new machine learning (ML) techniques and improvement in the performance of algorithms for lesion segmentation & detection. This development has been greatly facilitated by a series of challenges in Medical Imaging (e.g. Brain Tumor Segmentation (BRATS) Challenge) that have provided a framework for unbiased evaluation and comparison of the developed methods that are proposed to address a clinical problem. In particular, deep learning algorithms (e.g. U-Net architecture) have proven to be effective to solve learning problems involving the segmentation & detection of brain lesions. Such algorithms have a wide variety of potential applications in Neuroradiology ranging from quantification of disease progression in brain tumors and multiple sclerosis to differential diagnosis & prognosis in neurovascular and neuroinflammatory diseases.
Despite the encouraging body of research on ML algorithms for brain lesion quantification, there is a substantial lack of methods which have been translated into routine clinical practice. The challenges in integration of brain lesion segmentation & detection tools with clinical workflows is often left unaddressed by researchers and hence remains poorly understood. When deployed in clinical routine, ML methods should ideally be robust to changes in input data and produce reliable results that can support clinical decision making. For ML-based brain lesion quantification methods to be successfully translated into clinical workflow, several current gaps, including domain adaptation, uncertainty quantification, and model validation, in current ML research need to be filled. In addition, new public datasets and software modules are needed to support the translation of ML algorithms from research into clinical practice.
The objective of the current Research Topic is to foster new research in areas relevant to the translation of ML-based brain lesion quantification methods into clinical routine. This includes but is not limited to the following topics:
· Domain Adaptation (e.g. detection & quantification of dataset shifts)
· Uncertainty quantification (e.g. calibration of segmentation & detection models, novel methods for uncertainty estimation, integration of uncertainty estimates in a clinical workflow)
· Model validation (e.g. research on new metrics for evaluation of brain lesion segmentation, validation studies of existing methods on large datasets, systematic reviews of results from brain lesion segmentation & detection challenges)
· Machine Learning & data privacy (e.g. federated learning or secure multi-party computation with applications in brain lesion quantification)
· Data contributions (e.g. new public datasets to support development of ML methods for brain lesion quantification)
· Software contributions (e.g. Python modules offering functionality to support integration of ML methods into clinical routine such as e.g. automated image quality rating)
· Perspectives on how ML-based brain lesion quantification has been integrated in the clinical workflow of individual’s institution
In particular, we consider the following article types: Original Research, Systematic Review, Data Report, Technology and Code, Perspective (choice varies depending on selected Frontiers journal).
Keywords: Machine Learning, Translation, Neuroimaging, Deep Learning, Segmentation
Important Note: All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.