Cancers, life threats to human life, remain difficult to detect and identify at an early stage. Radiological images are common sources for cancer detection; but early signs of cancer are subtle, and it is easy to miss treatable cancer. Artificial intelligence (AI) has enormous potential as an assistive tool to aid radiologists in cancer detection. AI algorithms offer great performance for cancer identification, segmentation, assessments, etc. However, the black-box characteristics of the existing AI algorithms inevitably lead to less interpretability and verifiability in clinical predictions. There are several emerging challenges in applying AI to detect cancers.
At the data curation stage, existing publicly available datasets usually are of small size, contain partial labels, or come from various scanners or protocols, thus limiting their scopes. During the development stage, AI algorithms rely heavily on manual annotations by expert radiologists, and their performance affect when applied to data from other hospitals or protocols. In addition, current AI algorithms are weak in dealing with partial labels, noisy labels, long-tailed distribution, continual learning, etc.
More investigation is encouraged to facilitate the clinical usage of effective and user-friendly AI algorithms, which will be used as assistive aids to radiologists in a cooperative manner combining the best of human and AI expertise.
The aim of this Research Topic is to show the impact of novel AI applications using recent developments in computer vision, machine learning, and deep learning models for cancer detection, identification, and healthcare outcome improvements. We welcome the manuscripts covering the follow themes, but not limited to:
1. AI for diagnosis and interventions of cancer in the lung, bowel, prostate, breast, pancreas, esophagus, liver, bladder, brain, head and neck, etc.
2. AI for cancer modeling and simulation.
3. AI for tomographic cancer image reconstruction from CT, PET, SPECT, MRI, Ultrasound, etc.
4. AI for multimodal cancer visualisation.
5. AI for quantitative image analysis, including segmentation, registration, fusion, and synthesis.
We would like to acknowledge Drs. Linmin Pei, Zongwei Zhou, and Qiongmin Zhang, the Topic Coordinators, and Dr. Abhirup Banerjee have contributed to the preparation of the proposal for this Research Topic. Dr. Zhang’s research interest includes machine learning, medical image analysis, and computer vision.
Cancers, life threats to human life, remain difficult to detect and identify at an early stage. Radiological images are common sources for cancer detection; but early signs of cancer are subtle, and it is easy to miss treatable cancer. Artificial intelligence (AI) has enormous potential as an assistive tool to aid radiologists in cancer detection. AI algorithms offer great performance for cancer identification, segmentation, assessments, etc. However, the black-box characteristics of the existing AI algorithms inevitably lead to less interpretability and verifiability in clinical predictions. There are several emerging challenges in applying AI to detect cancers.
At the data curation stage, existing publicly available datasets usually are of small size, contain partial labels, or come from various scanners or protocols, thus limiting their scopes. During the development stage, AI algorithms rely heavily on manual annotations by expert radiologists, and their performance affect when applied to data from other hospitals or protocols. In addition, current AI algorithms are weak in dealing with partial labels, noisy labels, long-tailed distribution, continual learning, etc.
More investigation is encouraged to facilitate the clinical usage of effective and user-friendly AI algorithms, which will be used as assistive aids to radiologists in a cooperative manner combining the best of human and AI expertise.
The aim of this Research Topic is to show the impact of novel AI applications using recent developments in computer vision, machine learning, and deep learning models for cancer detection, identification, and healthcare outcome improvements. We welcome the manuscripts covering the follow themes, but not limited to:
1. AI for diagnosis and interventions of cancer in the lung, bowel, prostate, breast, pancreas, esophagus, liver, bladder, brain, head and neck, etc.
2. AI for cancer modeling and simulation.
3. AI for tomographic cancer image reconstruction from CT, PET, SPECT, MRI, Ultrasound, etc.
4. AI for multimodal cancer visualisation.
5. AI for quantitative image analysis, including segmentation, registration, fusion, and synthesis.
We would like to acknowledge Drs. Linmin Pei, Zongwei Zhou, and Qiongmin Zhang, the Topic Coordinators, and Dr. Abhirup Banerjee have contributed to the preparation of the proposal for this Research Topic. Dr. Zhang’s research interest includes machine learning, medical image analysis, and computer vision.
