The rapid enhancement of large-scale next-generation sequencing (NGS) technologies and their application have led to various computational challenges including genomic/epigenomic feature selection or new feature extraction, metagenomic classification, gene module identification, sub-graph clustering, gene signature detection, multi-omics data integration, spatial transcriptomic and other downstream analysis for big biomedical sequence data. NGS relevance has created a demand for sophisticated data analysis incorporating systems biology and various learning algorithms to solve those challenges for understanding disease pathology and single-cell heterogeneity.
Biomedical imaging encompasses various imaging modalities and procedures (viz., X-ray, CT-scan, MRI, Ultrasound, etc.) to capture the internal organs of the human body for potential diagnostic, treatment planning, follow-up and surgery.
Under these circumstances, Deep Learning (DL), one of the emerging learning techniques that rely on multi-layered artificial neural networks (ANNs) for unsupervised, supervised, and semi-supervised learning has gained a lot of attention for applications to NGS and imaging, even for applications comprising of both of these data for the same set of patients.
The major research phenomena for biomedical research are heterogeneity, feature dimension reduction, and disease classification. Artificial Intelligence (AI) techniques tend to resolve those issues in different ways. In particular, DL has a greater effect on handling big data alone or when combined with other techniques such as machine learning (ML) or pattern recognition. DL is also useful for the NGS and medical imaging data, or the integrated data, providing both of these profiles for the same set of patients. Various data repositories such as TCGA, ICGC, gene omnibus are available to collect those kinds of multi-omics NGS and biomedical data separately or together.
Different types of DL models are available to help researchers in disease prediction using NGS data as well as biomedical images of different organs of the body. The exposure of DL models in disease prediction using NGS and biomedical imaging provides unparallel opportunities to make the resulting outcomes of radiologists/doctors/biomedical researchers, reduce the time for detection and diagnosis of the disease, reduce treatment cost, and make a positive impact on public health.
To showcase the relevant features of DL algorithms and their application in the field of disease prediction and classification using transcriptomic/genomics/epigenomic NGS and biomedical imaging data, the scope of this issue broadly covers (but is not limited to) the following aspects:
• Application of various DL techniques on RNA-seq/scRNA-seq/CHIP-seq data.
• Multi-omics data integration using ML, DL, feature selection and bio-statistical techniques.
• Semi-supervised Generative Adversarial Networks (GAN) for the prediction of oncogenic variants in unlabeled genomic data profiles.
• Supervised GAN for labelled genomic/epigenomic data.
• 2D/3D/4D medical image classification.
• Organ, region and landmark localization.
• Object or lesion detection, segmentation and classification.
• Registration for disease progression.
• Development of computer-aided detection/diagnosis (CADe/CADx) system.
• Prediction of anti-cancer-based drug response transforming tabular array data into images through convolutional neural networks.
• Disease classification and biomarker discovery for Integrated data consisting of NGS and imaging data for the same set of patients.
• Classification model for physiological data segmentation and further application to disease progression and diagnosis.
• DL/ML applications in pharmacogenomics.
We would like to acknowledge Dr Yashika Rustagi, Dana–Farber Cancer Institute who has acted as a coordinator and have contributed to the preparation of the proposal for this Research Topic.
The rapid enhancement of large-scale next-generation sequencing (NGS) technologies and their application have led to various computational challenges including genomic/epigenomic feature selection or new feature extraction, metagenomic classification, gene module identification, sub-graph clustering, gene signature detection, multi-omics data integration, spatial transcriptomic and other downstream analysis for big biomedical sequence data. NGS relevance has created a demand for sophisticated data analysis incorporating systems biology and various learning algorithms to solve those challenges for understanding disease pathology and single-cell heterogeneity.
Biomedical imaging encompasses various imaging modalities and procedures (viz., X-ray, CT-scan, MRI, Ultrasound, etc.) to capture the internal organs of the human body for potential diagnostic, treatment planning, follow-up and surgery.
Under these circumstances, Deep Learning (DL), one of the emerging learning techniques that rely on multi-layered artificial neural networks (ANNs) for unsupervised, supervised, and semi-supervised learning has gained a lot of attention for applications to NGS and imaging, even for applications comprising of both of these data for the same set of patients.
The major research phenomena for biomedical research are heterogeneity, feature dimension reduction, and disease classification. Artificial Intelligence (AI) techniques tend to resolve those issues in different ways. In particular, DL has a greater effect on handling big data alone or when combined with other techniques such as machine learning (ML) or pattern recognition. DL is also useful for the NGS and medical imaging data, or the integrated data, providing both of these profiles for the same set of patients. Various data repositories such as TCGA, ICGC, gene omnibus are available to collect those kinds of multi-omics NGS and biomedical data separately or together.
Different types of DL models are available to help researchers in disease prediction using NGS data as well as biomedical images of different organs of the body. The exposure of DL models in disease prediction using NGS and biomedical imaging provides unparallel opportunities to make the resulting outcomes of radiologists/doctors/biomedical researchers, reduce the time for detection and diagnosis of the disease, reduce treatment cost, and make a positive impact on public health.
To showcase the relevant features of DL algorithms and their application in the field of disease prediction and classification using transcriptomic/genomics/epigenomic NGS and biomedical imaging data, the scope of this issue broadly covers (but is not limited to) the following aspects:
• Application of various DL techniques on RNA-seq/scRNA-seq/CHIP-seq data.
• Multi-omics data integration using ML, DL, feature selection and bio-statistical techniques.
• Semi-supervised Generative Adversarial Networks (GAN) for the prediction of oncogenic variants in unlabeled genomic data profiles.
• Supervised GAN for labelled genomic/epigenomic data.
• 2D/3D/4D medical image classification.
• Organ, region and landmark localization.
• Object or lesion detection, segmentation and classification.
• Registration for disease progression.
• Development of computer-aided detection/diagnosis (CADe/CADx) system.
• Prediction of anti-cancer-based drug response transforming tabular array data into images through convolutional neural networks.
• Disease classification and biomarker discovery for Integrated data consisting of NGS and imaging data for the same set of patients.
• Classification model for physiological data segmentation and further application to disease progression and diagnosis.
• DL/ML applications in pharmacogenomics.
We would like to acknowledge Dr Yashika Rustagi, Dana–Farber Cancer Institute who has acted as a coordinator and have contributed to the preparation of the proposal for this Research Topic.
