Marc Kohli

As artificial intelligence (AI) tools are increasingly integrated into imaging workflows, understanding how AI results are generated and presented to the end user can equip radiologists to optimize interactions with AI results in practice. Although AI solutions are marketed as high-performing options that promise efficiency and diagnostic gains, issues arising at the radiologist-AI interface can lead to diminished returns due to unintentional cognitive burdens or misalignments with clinical workflows. A foundational understanding of how images are processed by AI solutions, presented in imaging workflows, and documented can allow radiologists to troubleshoot shortcomings in practice after clinical deployment. This article is the second in a primer series providing a foundation in AI literacy for radiologists. Building on the first article in this primer series, this article addresses questions raised by end users while using AI in practice. It covers how to consider the intended use of AI solutions, the process through which AI results are generated, the reasons why results may not be available at the time of study interpretation, and how to align the presentation of AI results with end user workflows. The discussion also explores emerging topics and challenges, including considerations for storing AI results and the related medicolegal considerations.

A foundational understanding of core principles for artificial intelligence (AI) development enables radiologists to function as effective users of AI in their practice.

Purpose of Review The purpose of this review is to summarize the current status of artificial intelligence applied to prostate cancer MR imaging. Recent Findings Artificial intelligence has been applied to prostate cancer MR imaging to improve its diagnostic accuracy and reproducibility of interpretation. Multiple models have been tested for gland segmentation and volume calculation, automated lesion detection, localization, and characterization, as well as prediction of tumor aggressiveness and tumor recurrence. Summary Studies show, for example, that very robust automated gland segmentation and volume calculations can be achieved and that lesions can be detected and accurately characterized. Although results are promising, we should view these with caution. Most studies included a small sample of patients from a single institution and most models did not undergo proper external validation. More research is needed with larger and well-design studies for the development of reliable artificial intelligence tools.

“Just Accepted” papers have undergone full peer review and have been accepted for publication in Radiology: Artificial Intelligence. This article will undergo copyediting, layout, and proof review ...

Purpose: To create an algorithm able to accurately detect IVC filters on radiographs without human assistance, capable of being used to screen radiographs to identify patients needing IVC filter retrieval. Methods: A primary dataset of 5225 images, 30% of which included IVC filters, was assembled and annotated. 85% of the data was used to train a Cascade R-CNN (Region Based Convolutional Neural Network) object detection network incorporating a pre-trained ResNet-50 backbone. The remaining 15% of the data, independently annotated by three radiologists, was used as a test set to assess performance. The algorithm was also assessed on an independently constructed 1424-image dataset, drawn from a different institution than the primary dataset. Results: On the primary test set, the algorithm achieved a sensitivity of 96.2% (95% CI 92.7–98.1%) and a specificity of 98.9% (95% CI 97.4–99.5%). Results were similar on the external test set: sensitivity 97.9% (95% CI 96.2–98.9%), specificity 99.6 (95% CI 98.9–99.9%). Conclusion: Fully automated detection of IVC filters on radiographs with high sensitivity and excellent specificity required for an automated screening system can be achieved using object detection neural networks. Further work will develop a system for identifying patients for IVC filter retrieval based on this algorithm. Graphical

While many case studies have described the implementation of self-scheduling tools, which allow patients to schedule visits and imaging studies asynchronously online, none have explored the impact of self-scheduling on equitable access to care. ¹ Using an electronic health record patient portal, University of California San Francisco deployed a self-scheduling tool that allowed patients to self-schedule diagnostic imaging studies. We analyzed electronic health record data for the imaging modalities with the option to be self-scheduled from January 1, 2021 to September 1, 2021. We used descriptive statistics to compare demographic characteristics and created a multivariable logistic regression model to identify predictors of patient self-scheduling utilization. Among all active patient portal users, Latinx, Black/African American, and non-English speaking patients were less likely to self-schedule studies. Patients with Medi-Cal, California’s Medicaid program, and Medicare insurance were also less likely to self-schedule when compared with commercially insured patients. Efforts to facilitate use of patient portal-based applications are necessary to increase equitability and decrease disparities in access.

The objective is to determine patients’ utilization rate of radiology image viewing through an online patient portal and to understand its impact on radiologists. IRB approval was waived. In this two-part, multi-institutional study, patients’ image viewing rate was retrospectively assessed, and radiologists were anonymously surveyed for the impact of patient imaging access on their workflow. Patient access to web-based image viewing via electronic patient portals was enabled at 3 institutions (all had open radiology reports) within the past 5 years. The number of exams viewed online was compared against the total number of viewable imaging studies. An anonymized survey was distributed to radiologists at the 3 institutions, and responses were collected over 2 months. Patients viewed 14.2% of available exams – monthly open rate varied from 7.3 to 41.0%. A total of 254 radiologists responded to the survey (response rate 32.8%); 204 were aware that patients could view images. The majority (155/204; 76.0%) felt no impact on their role as radiologists; 11.8% felt negative and 9.3% positive. The majority (63.8%) were never approached by patients. Of the 86 who were contacted, 46.5% were contacted once or twice, 46.5% 3–4 times a year, and 4.7% 3–4 times a month. Free text comments included support for healthcare transparency (71), concern for patient confusion and anxiety (45), and need for attention to radiology reports and image annotations (15). A small proportion of patients viewed their radiology images. Overall, patients’ image viewing had minimal impact on radiologists. Radiologists were seldom contacted by patients. While many radiologists feel supportive, some are concerned about causing patient confusion and suggest minor workflow modifications.