Deep learning technologies have been developed for and applied to medical image analysis and show great potential to facilitate prostate gland, lung, breast, and liver segmentation, lesion detection, and classification, reduction in the impact incurred by inter-reader variabilities, and in mitigating the potential lack of expertise of less-experienced radiologists. Traditional shallow learning methods cannot meet the demand for accuracy needed to evaluate potential biological responses. Effective tumor prognostic indicators are conducive to the selection of reasonable personalized treatment for various cases to prevent excessive or inappropriate treatment. Deep learning methods are maturing, and this provides an opportunity to develop automatic analysis of medical images and to assist doctors in realizing high-precision intelligent diagnoses of diseases.

In clinical oncology, data on improving cancer treatment is rapidly increasing. With the development of deep learning and the advances in high-performance computing infrastructure, it is now feasible to integrate and analyze these growing multidimensional data to interpret models and predict prognosis, and to improve the joint decision-making of patients and clinicians. Although this approach has great potential, it still faces major challenges including the problem of inconsistency in data format and lack of labeled training data which need to be addressed. Existing deep learning approaches rely heavily on large, labeled data sets which are difficult to acquire. Several learning strategies have been adopted for medical image analysis. However, intelligence learning methods need to be more focused on enhancing prediction accuracy with decision understanding in solid tumor prognosis prediction. Therefore, more advanced technologies for the precise solid tumor prognosis elevation medical image analysis, target detection, target segmentation, signal diagnosis, and prediction of multimodal images are needed.

This Special Issue aims to bring together original research and review articles that provide new insights into solid tumor prognosis prediction. This Special Issue aims to collect the latest research progress and achievements in the field of solid tumor prognosis prediction based on deep learning approaches. We welcome research articles focusing on the modalities of medical imaging which include but are not limited to CT, MRI. PET, SPECT, etc. for solid tumor prognosis.

Potential topics include but are not limited to the following:

• Multitask learning for medical image reconstruction
• Unsupervised and semi-supervised learning for medical imaging
• Deep learning for medical image processing
• Multimodal machine learning for medical image fusion
• Medical image segmentation for medical image processing
• Radiomics analysis for solid tumor imaging