Error and Accountability

In this chapter we will explore the issues related to individual and organizational accountability for error, particularly when a patient suffers harm that is attributable to physician error. We will review the blameless culture and “just culture” models, as well as related issues of peer-review, regulatory compliance, medicolegal, and ethical aspects of error in this context. We will discuss the ethical duty to provide open disclosure of all errors and lapses directly to patients and their families, regardless of cause (and separated from the issue of blame) and in some circumstances coupled with financial or other compensation for any harm done.

Taxonomy and Epidemiology of Error in Radiology

This chapter provides an overview of the prevalence and classification of error types in radiology, including the frequency and types of errors made by radiologists. We will review the relative contribution of perceptual error—in which findings are simply not seen—as compared to other common types of error. This error epidemiology will be considered in the light of the underlying variability and uncertainties present in the radiological process. The role of key cognitive biases will also be reviewed, including anchoring bias, confirmation bias, and availability bias. The role of attentional focus, working memory, and problems caused by fatigue and interruption will also be explored. Finally, the problem of radiologist error will be considered in the context of the overall problem of diagnostic error in medicine.

Artificial Intelligence and Machine Learning in Radiology

This final chapter, which assumes no prior reader knowledge of the topic, reviews the promise of artificial intelligence (AI), especially machine learning and deep learning in radiology. We initially discuss key concepts in the field of AI and gain a broad overview of the field and its potential, as well as the impact it is having on multiple areas of human endeavor. Subsequently, we focus on current and projected aspects of AI as applied to diagnostic radiology, specifically on how AI might provide an avenue for error prevention and remediation in radiology. The possible impact of AI in changing the radiologist’s role and basic job description is also considered.

Uncertainty and Interpretive Error

Uncertainty pervades medical diagnosis and can rarely be entirely eliminated. Diagnostic imaging is meant to reduce that uncertainty, ideally to the point where a clinician feels confident enough to choose a course of action. But the process of diagnostic imaging is itself prone to high variability and error. Sources of variability include technical, procedural, and anatomic variation, the variable use of language to describe and report radiological abnormalities, and the range of variability in the manifestations of disease processes Cognitive biases and varying understanding of the prevalence and likelihood of disease among radiologists can also lead to interpretive error. This chapter explores the sources of error and the sources of uncertainty in the radiological process. There is considerable overlap between the two.

Visual and Perceptual Science and Radiology

This chapter discusses the science of human visual perception, including recently published experimental results that directly explore the underlying visual perceptual processes involved in diagnostic radiology. The amazing properties of human visual perception have been found to involve a (perhaps unexpectedly) complex interplay between the eye and brain and can be affected by a large number of factors in surprising and unexpected ways. We will discuss what is different about a radiologist’s brain in this regard, compared to the brains of people who are not similarly trained and experienced. The chapter reviews the most common radiologist errors, as these may yield key insights into the process and lead to more effective strategies for error prevention.

Reducing Error in Radiology

This chapter reviews a new “theory of error,” which is a way of classifying errors based on three principal types of error causes. Mechanisms and solutions are proposed for each of these major error types. The strategies for error reduction that we discuss include those aimed at systems factors, those aimed at individual and cognitive factors, and those aimed at overcoming the flaws that are inherent to the basic neurobiological makeup of human beings and that predispose us to making certain types of error in this setting. We conclude by outlining a “roadmap” for reducing error in radiology in the future.