Iris Tan
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Iris Tan has a total of 17 co-authors. Here are the top co-authors by number of shared publications:
- Arun James Thirunavukarasu
International Centre for Eye Health, London School of Hygiene and Tropical Medicine, London, UK
2 shared publications - Daniel Shu Wei Ting
Singapore Eye Research Institute, Singapore National Eye Center, Singapore, Singapore
2 shared publications - Edward Korot
Retina Specialists of Michigan, Grand Rapids, Michigan
2 shared publications - Kabilan Elangovan
Artificial Intelligence Office, SingHealth, Singapore, Singapore
2 shared publications - Laura Gutierrez
Artificial Intelligence and Digital Innovation Research Group, Singapore Eye Research Institute, Singapore, Singapore
2 shared publications - Pearse Andrew Keane
Institute of Ophthalmology, University College London, London, UK
2 shared publications
Latest Publications
- Open Access
- 12 October 2023
- Published by JMIR Publications Inc. in Journal of Medical Internet Research
- Vol. 25, e49949
Democratizing Artificial Intelligence Imaging Analysis With Automated Machine Learning: Tutorial
Deep learning–based clinical imaging analysis underlies diagnostic artificial intelligence (AI) models, which can match or even exceed the performance of clinical experts, having the potential to revolutionize clinical practice. A wide variety of automated machine learning (autoML) platforms lower the technical barrier to entry to deep learning, extending AI capabilities to clinicians with limited technical expertise, and even autonomous foundation models such as multimodal large language models. Here, we provide a technical overview of autoML with descriptions of how autoML may be applied in education, research, and clinical practice. Each stage of the process of conducting an autoML project is outlined, with an emphasis on ethical and technical best practices. Specifically, data acquisition, data partitioning, model training, model validation, analysis, and model deployment are considered. The strengths and limitations of available code-free, code-minimal, and code-intensive autoML platforms are considered. AutoML has great potential to democratize AI in medicine, improving AI literacy by enabling “hands-on” education. AutoML may serve as a useful adjunct in research by facilitating rapid testing and benchmarking before significant computational resources are committed. AutoML may also be applied in clinical contexts, provided regulatory requirements are met. The abstraction by autoML of arduous aspects of AI engineering promotes prioritization of data set curation, supporting the transition from conventional model-driven approaches to data-centric development. To fulfill its potential, clinicians must be educated on how to apply these technologies ethically, rigorously, and effectively; this tutorial represents a comprehensive summary of relevant considerations.
- 12 October 2023
- Published by JMIR Publications Inc. in Journal of Medical Internet Research
- Vol. 25, e49949
Open AccessDemocratizing Artificial Intelligence Imaging Analysis With Automated Machine Learning: Tutorial
Deep learning–based clinical imaging analysis underlies diagnostic artificial intelligence (AI) models, which can match or even exceed the performance of clinical experts, having the potential to revolutionize clinical practice. A wide variety of automated machine learning (autoML) platforms lower the technical barrier to entry to deep learning, extending AI capabilities to clinicians with limited technical expertise, and even autonomous foundation models such as multimodal large language models. Here, we provide a technical overview of autoML with descriptions of how autoML may be applied in education, research, and clinical practice. Each stage of the process of conducting an autoML project is outlined, with an emphasis on ethical and technical best practices. Specifically, data acquisition, data partitioning, model training, model validation, analysis, and model deployment are considered. The strengths and limitations of available code-free, code-minimal, and code-intensive autoML platforms are considered. AutoML has great potential to democratize AI in medicine, improving AI literacy by enabling “hands-on” education. AutoML may serve as a useful adjunct in research by facilitating rapid testing and benchmarking before significant computational resources are committed. AutoML may also be applied in clinical contexts, provided regulatory requirements are met. The abstraction by autoML of arduous aspects of AI engineering promotes prioritization of data set curation, supporting the transition from conventional model-driven approaches to data-centric development. To fulfill its potential, clinicians must be educated on how to apply these technologies ethically, rigorously, and effectively; this tutorial represents a comprehensive summary of relevant considerations.
- Open Access
- 14 June 2023
- Published by JMIR Publications Inc. in JMIR Preprints
Democratizing Artificial Intelligence Imaging Analysis With Automated Machine Learning: Tutorial (Preprint)
UNSTRUCTURED Deep learning–based clinical imaging analysis underlies diagnostic artificial intelligence (AI) models, which can match or even exceed the performance of clinical experts, having the potential to revolutionize clinical practice. A wide variety of automated machine learning (autoML) platforms lower the technical barrier to entry to deep learning, extending AI capabilities to clinicians with limited technical expertise, and even autonomous foundation models such as multimodal large language models. Here, we provide a technical overview of autoML with descriptions of how autoML may be applied in education, research, and clinical practice. Each stage of the process of conducting an autoML project is outlined, with an emphasis on ethical and technical best practices. Specifically, data acquisition, data partitioning, model training, model validation, analysis, and model deployment are considered. The strengths and limitations of available code-free, code-minimal, and code-intensive autoML platforms are considered. AutoML has great potential to democratize AI in medicine, improving AI literacy by enabling “hands-on” education. AutoML may serve as a useful adjunct in research by facilitating rapid testing and benchmarking before significant computational resources are committed. AutoML may also be applied in clinical contexts, provided regulatory requirements are met. The abstraction by autoML of arduous aspects of AI engineering promotes prioritization of data set curation, supporting the transition from conventional model-driven approaches to data-centric development. To fulfill its potential, clinicians must be educated on how to apply these technologies ethically, rigorously, and effectively; this tutorial represents a comprehensive summary of relevant considerations.
- 14 June 2023
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessDemocratizing Artificial Intelligence Imaging Analysis With Automated Machine Learning: Tutorial (Preprint)
UNSTRUCTURED Deep learning–based clinical imaging analysis underlies diagnostic artificial intelligence (AI) models, which can match or even exceed the performance of clinical experts, having the potential to revolutionize clinical practice. A wide variety of automated machine learning (autoML) platforms lower the technical barrier to entry to deep learning, extending AI capabilities to clinicians with limited technical expertise, and even autonomous foundation models such as multimodal large language models. Here, we provide a technical overview of autoML with descriptions of how autoML may be applied in education, research, and clinical practice. Each stage of the process of conducting an autoML project is outlined, with an emphasis on ethical and technical best practices. Specifically, data acquisition, data partitioning, model training, model validation, analysis, and model deployment are considered. The strengths and limitations of available code-free, code-minimal, and code-intensive autoML platforms are considered. AutoML has great potential to democratize AI in medicine, improving AI literacy by enabling “hands-on” education. AutoML may serve as a useful adjunct in research by facilitating rapid testing and benchmarking before significant computational resources are committed. AutoML may also be applied in clinical contexts, provided regulatory requirements are met. The abstraction by autoML of arduous aspects of AI engineering promotes prioritization of data set curation, supporting the transition from conventional model-driven approaches to data-centric development. To fulfill its potential, clinicians must be educated on how to apply these technologies ethically, rigorously, and effectively; this tutorial represents a comprehensive summary of relevant considerations.
- 1 September 2001
- Published by Springer Nature in Journal of Neurology
- Vol. 248 (9) , 789-794
Improved interobserver agreement for visual detection of active T2 lesions on serial MR scans in multiple sclerosis using image registration
The aim of this study was to analyse the effect of image registration on interobserver agreement in the visual detection of active multiple sclerosis (MS) lesions from serial magnetic resonance (MR) scans. T2W spin-echo MR scans (3-mm slices) of 16 MS patients participating in a treatment trial were selected. For each patient, two pairs of scans were used: an original (i. e., non-registered) and a registered pair. For the original pair, baseline and month 6 were used, and for the registered pair month 3 and 9. For registration an automatic matching algorithm based on Mutual Information was used. Six observers identified active lesions on both original and registered scans. Kappa values were calculated to assess interobserver agreement. Reslicing caused a slight blurring of the images, but near perfect registration. The kappa value of 0.35 ± 0.07 for new lesions on original images improved to 0.62 (± 0.06) by registration (p = 0.004). For enlarging lesions on original images it was extremely poor (κ 0.11 ± 0.05), and did not benefit much by registration (κ 0.20 ± 0.11). Thus, image registration improved interobserver agreement for visual detection of new lesions. For enlarging lesions, registration improved agreement but still not to a satisfactory level.
- 1 September 2001
- Published by Springer Nature in Journal of Neurology
- Vol. 248 (9) , 789-794
Improved interobserver agreement for visual detection of active T2 lesions on serial MR scans in multiple sclerosis using image registration
The aim of this study was to analyse the effect of image registration on interobserver agreement in the visual detection of active multiple sclerosis (MS) lesions from serial magnetic resonance (MR) scans. T2W spin-echo MR scans (3-mm slices) of 16 MS patients participating in a treatment trial were selected. For each patient, two pairs of scans were used: an original (i. e., non-registered) and a registered pair. For the original pair, baseline and month 6 were used, and for the registered pair month 3 and 9. For registration an automatic matching algorithm based on Mutual Information was used. Six observers identified active lesions on both original and registered scans. Kappa values were calculated to assess interobserver agreement. Reslicing caused a slight blurring of the images, but near perfect registration. The kappa value of 0.35 ± 0.07 for new lesions on original images improved to 0.62 (± 0.06) by registration (p = 0.004). For enlarging lesions on original images it was extremely poor (κ 0.11 ± 0.05), and did not benefit much by registration (κ 0.20 ± 0.11). Thus, image registration improved interobserver agreement for visual detection of new lesions. For enlarging lesions, registration improved agreement but still not to a satisfactory level.