Surrogate Biomedical Data Generation

Dear Colleagues,

Recently, there have been tremendous advances in data-driven techniques, mainly based on deep learning (DL) as a powerful tool in artificial intelligence (AI). Computing systems perform tasks such as image recognition and speech transcription, achieving super-human results. Huge amounts of data, principally images including medical images, text, speech and multimodal biomedical data, are needed to train these systems, many of which are based on deep neural networks (DNN). Such data may be stored in the Cloud, allowing for the identification of trends for patient remote monitoring, enabling speedy clinical intervention. Consequently, it makes the estimation of diagnostic parameters more accurate.

Scarcity of patient one-dimensional (i.e., time series) or two-dimensional (i.e., image) data is the major barrier to optimally using the power of machine learning, particularly data-driven and deep learning-based systems for feature estimation, learning, clustering and classification. This consequently affects the use of important diagnostic and patient monitoring information for clinical applications. For example, non-invasive brain analysis through deep learning approaches requires big data analytics and in silico simulation for explaining brain function and the associated pathologies. As another example, to develop more effective planning for human stroke rehabilitation, extensive data for each state of rehabilitative assessment are essential.  

Therefore, the generation of suitable biomedical data which can be combined with real data for identification of patient state is highly desirable. The methods for generation of surrogate (also known as augmented or synthetic) data are different depending on the methods for their assessment and analysis. Each surrogate data generation or augmentation method has different effects depending on the model and dataset too. One is to find a source with similar conditions or parameters and use those data in modeling. Another method is to focus on patterns of the underlying system, and to search for a similar pattern within the related data sources. Some methods attempt to preserve the underlying statistical properties and some others stress on maintaining the shape of some components of data within time, frequency or other domains. These synthetically generated data are used for training the machine learning systems and improving their performance when compared to the cases where only small data sizes are available. Therefore, various adaptive and statistical signal processing or machine learning systems, including deep learning, may be designed for this purpose. On the other hand, to avoid generating undesired surrogates, there are linear and non-linear testing methods and hypotheses. Therefore, this Special Issue is expected to receive the outcome of surrogate biomedical data generation research. The related topics include (but are not limited to):

• One-dimensional or time series surrogate biodata generation;
• Two-dimensional or medical image surrogate data generation;
• Data augmentation for edge systems;
• Deep learning for surrogate biodata generation;
• Dynamical systems for surrogate biomedical data generation;
• Group surrogate biomedical data generation;
• Multimodal surrogate biodata generation;
• Quantification of generated data quality;
• Shape preserving surrogate biodata generation;
• Surrogate biomedical data testing;
• Surrogate data for patient assessment.

Prof. Dr. Saeid Sanei
Dr. Tracey KM Lee
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Bioengineering is an international peer-reviewed open access monthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2700 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

• surrogate
• augmentation
• deep learning
• machine learning
• biomedical images
• biomedical signals
• edge systems
• testing