Artificial intelligence (AI) is rapidly moving from an experimental phase to an implementation phase in many fields, including medicine. The combination of improved availability of large datasets, increasing computing power, and advances in learning algorithms has created major performance breakthroughs in the development of AI applications. In the last 5 years, AI techniques known as deep learning have delivered rapidly improving performance in image recognition, caption generation, and speech recognition. Radiology, in particular, is a prime candidate for early adoption of these techniques. It is anticipated that the implementation of AI in radiology over the next decade will significantly improve the quality, value, and depth of radiology's contribution to patient care and population health, and will revolutionize radiologists' workflows. The Canadian Association of Radiologists (CAR) is the national voice of radiology committed to promoting the highest standards in patient-centered imaging, lifelong learning, and research. The CAR has created an AI working group with the mandate to discuss and deliberate on practice, policy, and patient care issues related to the introduction and implementation of AI in imaging. This white paper provides recommendations for the CAR derived from deliberations between members of the AI working group. This white paper on AI in radiology will inform CAR members and policymakers on key terminology, educational needs of members, research and development, partnerships, potential clinical applications, implementation, structure and governance, role of radiologists, and potential impact of AI on radiology in Canada.

This is a condensed summary of an international multisociety statement on ethics of artificial intelligence (AI) in radiology produced by the ACR, European Society of Radiology, RSNA, Society for Imaging Informatics in Medicine, European Society of Medical Imaging Informatics, Canadian Association of Radiologists, and American Association of Physicists in Medicine. AI has great potential to increase efficiency and accuracy throughout radiology, but it also carries inherent pitfalls and biases. Widespread use of AI-based intelligent and autonomous systems in radiology can increase the risk of systemic errors with high consequence and highlights complex ethical and societal issues. Currently, there is little experience using AI for patient care in diverse clinical settings. Extensive research is needed to understand how to best deploy AI in clinical practice. This statement highlights our consensus that ethical use of AI in radiology should promote well-being, minimize harm, and ensure that the benefits and harms are distributed among stakeholders in a just manner. We believe AI should respect human rights and freedoms, including dignity and privacy. It should be designed for maximum transparency and dependability. Ultimate responsibility and accountability for AI remains with its human designers and operators for the foreseeable future. The radiology community should start now to develop codes of ethics and practice for AI that promote any use that helps patients and the common good and should block use of radiology data and algorithms for financial gain without those two attributes.

Osteoporotic fractures are a significant cause of morbidity in acute lymphoblastic leukemia (ALL). Our objective was to determine the incidence and predictors of fractures and recovery from osteoporosis in pediatric ALL over 6 years following glucocorticoid initiation. Vertebral fractures (VF) and vertebral body reshaping were assessed on annual spine radiographs, low-trauma non-VF were recorded at regular intervals and spine bone mineral density (BMD) was captured every 6 months for 4 years and then annually. A total of 186 children with ALL were enrolled (median age 5.3 years; range, 1.3 to 17.0 years). The cumulative fracture incidence was 32.5% for VF and 23.0% for non-VF; 39.0% of children with VF were asymptomatic. No fractures occurred in the sixth year and 71.3% of incident fractures occurred in the first 2 years. Baseline VF, cumulative glucocorticoid dose, and baseline lumbar spine (LS) BMD Z-score predicted both VF and non-VF. Vertebral body reshaping following VF was incomplete or absent in 22.7% of children. Those with residual vertebral deformity following VF were older compared to those without (median age 8.0 years at baseline [interquartile range {IQR}, 5.5 to 9.4] versus 4.8 years [IQR, 3.6 to 6.2], p = 0.04) and had more severe vertebral collapse (median maximum spinal deformity index 3.5 [IQR, 1.0 to 8.0] versus 0.5 [IQR, 0.0 to 1.0], p = 0.01). VF and low LS BMD Z-score at baseline as well as glucocorticoid exposure predicted incident VF and non-VF. Nearly 25% of children had persistent vertebral deformity following VF, more frequent in older children, and in those with more severe collapse. These results suggest the need for trials addressing interventions in the first 2 years of chemotherapy, targeting older children and children with more severe vertebral collapse, because these children are at greatest risk for incident VF and subsequent residual vertebral deformity. © 2018 American Society for Bone and Mineral Research.

SUMMARY: DWI reportedly accurately differentiates pediatric posterior fossa tumors, but anecdotal experience suggests limitations. In 3 years, medulloblastoma and JPA were differentiated by DWI alone in 23/26 cases (88%). Ependymoma (n ϭ 5) could not be reliably differentiated from medulloblastoma or JPA. A trend toward increased diffusion restriction in higher grade tumors (1/14 grade I, 7%; 9/12 grade IV, 75%) had too much overlap to predict the grade of individual cases. The overlap in ADC between tumor types appeared partly due to technical factors (in small, heterogeneous, calcific, or hemorrhagic tumors) but also likely reflected true histologic variability, given that our 3 overlap cases included a desmoplastic medulloblastoma, an anaplastic ependymoma, and a JPA with restricted diffusion in its nodule. Simple structural features (macrocystic tumor, location off midline) aided in distinguishing JPA from the other tumors in these cases.ABBREVIATIONS: ADC ϭ apparent diffusion coefficient; ADCmean ϭ mean value of ADC; ADCmin ϭ minimum value of ADC; DWI ϭ diffusion-weighted imaging; FLAIR ϭ fluid-attenuated inversion recovery; JPA ϭ juvenile pilocytic astrocytoma; WHO ϭ World Health Organization D WI might, in theory, effectively distinguish tumor types and histologic grades because higher grade tumors with more densely packed cells should have increasingly restricted diffusion (with a lower ADC).

No abstract

One quarter of children with ALL developed incident VF in the 4 years after diagnosis; most of the VF burden was in the first year. Over one third of children with incident VF were asymptomatic. Discrete clinical predictors of a VF were evident early in the patient's clinical course, including a VF at diagnosis.

This is a condensed summary of an international multisociety statement on ethics of artificial intelligence (AI) in radiology produced by the ACR, European Society of Radiology, RSNA, Society for Imaging Informatics in Medicine, European Society of Medical Imaging Informatics, Canadian Association of Radiologists, and American Association of Physicists in Medicine. AI has great potential to increase efficiency and accuracy throughout radiology, but also carries inherent pitfalls and biases. Widespread use of AI-based intelligent and autonomous systems in radiology can increase the risk of systemic errors with high consequence, and highlights complex ethical and societal issues. Currently, there is little experience using AI for patient care in diverse clinical settings. Extensive research is needed to understand how to best deploy AI in clinical practice. This statement highlights our consensus that ethical use of AI in radiology should promote well-being, minimize harm, and ensure that the benefits and harms are distributed among stakeholders in a just manner. We believe AI should respect human rights and freedoms, including dignity and privacy. It should be designed for maximum transparency and dependability. Ultimate responsibility and accountability for AI remains with its human designers and operators for the foreseeable future. The radiology community should start now to develop codes of ethics and practice for AI which promote any use that helps patients and the common good and should block use of radiology data and algorithms for financial gain without those two attributes.

Vertebral fractures are an important yet underrecognized manifestation of osteoporosis in children with chronic, glucocorticoidtreated illnesses. Our goal was to determine the incidence and clinical predictors of vertebral fractures in the 3 years following glucocorticoid initiation among pediatric patients with rheumatic disorders. Incident vertebral fractures were evaluated according to the Genant semiquantitative method on lateral radiographs at baseline and then annually in the 3 years following glucocorticoid initiation. Extended Cox models were used to assess the association between vertebral fractures and clinical risk predictors. A total of 134 children with rheumatic disorders were enrolled in the study (mean AE standard deviation (SD) age 9.9 AE 4.4 years; 65% girls). The unadjusted vertebral fracture incidence rate was 4.4 per 100 person-years, with a 3-year incidence proportion of 12.4%. The highest annual incidence occurred in the first year (6.0%; 95% confidence interval (CI) 2.9% to 11.7%). Almost one-half of the patients with fractures were asymptomatic. Every 0.5 mg/kg increase in average daily glucocorticoid (prednisone equivalents) dose was associated with a twofold increased fracture risk (hazard ratio (HR) 2.0; 95% CI 1.1 to 3.5). Other predictors of increased vertebral fracture risk included: (1) increases in disease severity scores between baseline and 12 months; (2) increases in body mass index Z-scores in the first 6 months of each 12-month period preceding the annual fracture assessment; and (3) decreases in lumbar spine bone mineral density Z-scores in the first 6 months of glucocorticoid therapy. As such, we observed that a clinically significant number of children with rheumatic disorders developed incident vertebral fractures in the 3 years following glucocorticoid initiation. Almost one-half of the children were asymptomatic and thereby would have been undiagnosed in the absence of radiographic monitoring. In addition, discrete clinical predictors of incident vertebral fractures were evident early in the course of glucocorticoid therapy.