Medical imaging is a core component in medical health care, and accounts for 40% of all UK hospital procedures (http://www.ahrq.gov/data/hcup/). Clinical screening programmes for breast and colorectal cancer rely exclusive on medical imaging. The vast amount of imaging data produced by modalities such as CT and MRI, and other imaging technologies are rapidly changing the radiological task.
Until the 1940’s radiologists assumed that the reports of trained specialists were a faithful representation of the images’ information content. In subsequent years the issue of observer error began to emerge as a critical factor. For example, it became clear that inter and intra observer variability far exceeded any variability in the scanning modality during research to determine the optimal imaging method to detect tuberculosis in mass chest imaging programmes. This problem remains today. Despite all the advances in technology, screening error rates remain persistently high in breast cancer and lung nodule detection. The use of Computer Aided Detection (i.e., CAD), which is often implemented without considering its impact on the performance of expert observers, has not resulted in a clear improvement in the diagnostic performance.
Although there have been great advances in understanding radiological expertise, in particular the type of errors made, there is still a limited understanding of the underlying processes that are involved in accurate medical image perception. New techniques and technologies are often implemented without understanding their impact on the human observer; and understanding the human observer is essential for education, for accurate modelling and the development of potentially useful computer algorithms. To date the medical image perception and vision science literatures have largely developed separately. Yet there is clearly scope to determine how visual processes are applied to the radiological task, assuming that the basic mechanisms that underlie visual perception must form the basis for medical image perception.
The aim of this Research Topic is to:
(1) Review the current state of play in our understanding of medical image perception.
(2) Determine the intersection between vision science and the radiological task.
(3) Map out the factors that contribute to radiological errors.
(4) Review evidence based studies on the guidance for enhancing training and practice.
(5) Review the relationship of CAD and human expertise.
(6) Evaluate the implications of recent research on imitation and observational learning for performance on perceptual detection tasks.
Medical imaging is a core component in medical health care, and accounts for 40% of all UK hospital procedures (http://www.ahrq.gov/data/hcup/). Clinical screening programmes for breast and colorectal cancer rely exclusive on medical imaging. The vast amount of imaging data produced by modalities such as CT and MRI, and other imaging technologies are rapidly changing the radiological task.
Until the 1940’s radiologists assumed that the reports of trained specialists were a faithful representation of the images’ information content. In subsequent years the issue of observer error began to emerge as a critical factor. For example, it became clear that inter and intra observer variability far exceeded any variability in the scanning modality during research to determine the optimal imaging method to detect tuberculosis in mass chest imaging programmes. This problem remains today. Despite all the advances in technology, screening error rates remain persistently high in breast cancer and lung nodule detection. The use of Computer Aided Detection (i.e., CAD), which is often implemented without considering its impact on the performance of expert observers, has not resulted in a clear improvement in the diagnostic performance.
Although there have been great advances in understanding radiological expertise, in particular the type of errors made, there is still a limited understanding of the underlying processes that are involved in accurate medical image perception. New techniques and technologies are often implemented without understanding their impact on the human observer; and understanding the human observer is essential for education, for accurate modelling and the development of potentially useful computer algorithms. To date the medical image perception and vision science literatures have largely developed separately. Yet there is clearly scope to determine how visual processes are applied to the radiological task, assuming that the basic mechanisms that underlie visual perception must form the basis for medical image perception.
The aim of this Research Topic is to:
(1) Review the current state of play in our understanding of medical image perception.
(2) Determine the intersection between vision science and the radiological task.
(3) Map out the factors that contribute to radiological errors.
(4) Review evidence based studies on the guidance for enhancing training and practice.
(5) Review the relationship of CAD and human expertise.
(6) Evaluate the implications of recent research on imitation and observational learning for performance on perceptual detection tasks.