Trust and stakeholder perspectives on the implementation of AI tools in clinical radiology.

OBJECTIVES: To define requirements that condition trust in artificial intelligence (AI) as clinical decision support in radiology from the perspective of various stakeholders and to explore ways to fulfil these requirements. METHODS: Semi-structured interviews were conducted with twenty-five respondents-nineteen directly involved in the development, implementation, or use of AI applications in radiology and six working with AI in other areas of healthcare. We designed the questions to explore three themes: development and use of AI, professional decision-making, and management and organizational procedures connected to AI. The transcribed interviews were analysed in an iterative coding process from open coding to theoretically informed thematic coding. RESULTS: We identified four aspects of trust that relate to reliability, transparency, quality verification, and inter-organizational compatibility. These aspects fall under the categories of substantial and procedural requirements. CONCLUSIONS: Development of appropriate levels of trust in AI in healthcare is complex and encompasses multiple dimensions of requirements. Various stakeholders will have to be involved in developing AI solutions for healthcare and radiology to fulfil these requirements. CLINICAL RELEVANCE STATEMENT: For AI to achieve advances in radiology, it must be given the opportunity to support, rather than replace, human expertise. Support requires trust. Identification of aspects and conditions for trust allows developing AI implementation strategies that facilitate advancing the field. KEY POINTS: • Dimensions of procedural and substantial demands that need to be fulfilled to foster appropriate levels of trust in AI in healthcare are conditioned on aspects related to reliability, transparency, quality verification, and inter-organizational compatibility. •Creating the conditions for trust to emerge requires the involvement of various stakeholders, who will have to compensate the problem's inherent complexity by finding and promoting well-defined solutions.

The clinical significance of the T2-FLAIR mismatch sign in grade II and III gliomas: a population-based study.

BACKGROUND: The T2-FLAIR mismatch sign is an imaging finding highly suggestive of isocitrate dehydrogenase mutated (IDH-mut) 1p19q non-codeleted (non-codel) gliomas (astrocytomas). In previous studies, it has shown excellent specificity but limited sensitivity for IDH-mut astrocytomas. Whether the mismatch sign is a marker of a clinically relevant subtype of IDH-mut astrocytomas is unknown. METHODS: We included histopathologically verified supratentorial lower-grade gliomas (LGG) WHO grade II-III retrospectively during the period 2010-2016. In the period 2017-2018, patients with suspected LGG radiologically were prospectively included, and in this cohort other diagnoses than glioma could occur. Clinical, radiological and molecular data were collected. For clinical evaluation we included all patients with IDH-mut astrocytomas. In the 2010-2016 cohort DNA methylation analysis with Infinium MethylationEPIC BeadChip (Illumina) was performed for patients with an IDH-mut astrocytoma with available tissue. We aimed to examine the association of the T2-FLAIR mismatch sign with clinical factors and outcomes. Additionally, we evaluated the diagnostic reliability of the mismatch sign and its relation to methylation profiles. RESULTS: Out of 215 patients with LGG, 135 had known IDH-mutation and 1p19q codeletion status. Fifty patients had an IDH-mut astrocytoma and 12 of these (24.0%) showed a mismatch sign. The sensitivity and specificity of the mismatch sign for IDH-mut detection were 26.4 and 97.6%, respectively. There were no differences between patients with an IDH-mut astrocytoma with or without mismatch sign when grouped according to T2-FLAIR mismatch sign with respect to baseline characteristics, clinical outcomes and methylation profiles. The overall interrater agreement between neuroradiologist and clinical neurosurgeons for the T2-FLAIR mismatch sign was significant when all 215 MRI examination assessed (κ = 0.77, p < 0.001, N = 215). CONCLUSION: The T2-FLAIR mismatch sign in patients with an IDH-mut astrocytoma is not associated with clinical presentation or outcome. It seems unlikely that the IDH-mut astrocytomas with mismatch sign represent a specific subentity. Finally, we have validated that the T2-FLAIR mismatch sign is a reliable and specific marker of IDH-mut astrocytomas.

Inborn errors of metabolism: combining clinical and radiologic clues to solve the mystery.

OBJECTIVE: Inborn errors of metabolism in children can be challenging to interpret because of the similarity of their appearances on imaging. There are important clues to the diagnosis based on clinical history, head circumference (e.g., macrocephaly), geographic distribution of lesions (e.g., subcortical vs deep white matter or frontal vs parietooccipital), and other imaging features (e.g., contrast enhancement, calcification, cysts, and cortical dysplasia). CONCLUSION: In this article, we present an algorithm-based approach to diagnosing pediatric metabolic disease with a discussion of key imaging features.

Spinal cord compression in relation to clinical symptoms in patients with spinal meningiomas.

OBJECTIVE: Spinal meningiomas are common primary tumors of the spinal canal and the resulting spinal cord compression (SCC) is intrinsically related to symptoms and outcome, but literature concerning this association is limited. We aimed to present data on both degree of SCC and tumor occupancy percentage in relation to neurological symptoms and outcome. METHODS: Patients ≥ 18 years with a histological diagnosis of spinal canal meningioma treated between 2000 and 2017 were retrospectively evaluated for symptoms and neurological outcome in relation to SCC (i.e. compression of spinal cord at maximal tumor compression compared to maximum area above/below compression) and tumor occupancy percentage (percentage of dural sac area occupied by tumor at maximal tumor compression). Area segmentation of spinal cord, tumor and dural sac (as marker of spinal canal) was performed manually on magnetic resonance imaging (MRI) scans. The neurological deficit was assessed pre- and postoperatively according to the McCormick score. A logistic regression was made with a training set to identify the cut-off level for motor deficit. RESULTS: The cohort included 111 patients with a mean age of 62.5 years and 77.5% were female. The dominating symptoms preoperatively were sensory disturbance (91.0%), motor deficit (80.2%) and gait disturbance (67.6%). Postoperatively 53.2% of patients, also in some of those with severe deficit and high tumor occupancy, improved their neurological deficit and 43.2% were unchanged. Patients with intradural meningioma and assessable MRI scans were included to evaluate SCC (n = 83). The mean extent of SCC was 50.6%. Exploration of tumor occupancy percentage identified a cut-off at 65% tumor occupancy to best discriminate between patients with or without motor deficit. CONCLUSION: Patients with an intradural tumor occupancy percentage of > 65% are more likely to have a preoperative symptom and deficit, validating previous findings. Therefore, we suggest that even in asymptomatic, otherwise fit, patients with tumor occupancy approaching 65% should be considered for surgery since there is a high risk of developing deficit with even minimal growth. Concerning recovery, patients with tumor both high tumor occupancy and significantly impaired function tended to improve their functional level postoperatively.

Identification and characterization of choline transporter-like protein 2, an inner ear glycoprotein of 68 and 72 kDa that is the target of antibody-induced hearing loss.

The Kresge Hearing Research Institute-3 (KHRI-3) antibody binds to a guinea pig inner ear supporting cell antigen (IESCA) and causes hearing loss. To gain insight into the mechanism of antibody-induced hearing loss, we used antibody immunoaffinity purification to isolate the IESCA, which was then sequenced by mass spectroscopy, revealing 10 guinea pig peptides identical to sequences in human choline transporter-like protein 2 (CTL2). Full-length CTL2 cDNA sequenced from guinea pig inner ear has 85.9% identity with the human cDNA. Consistent with its expression on the surface of supporting cells in the inner ear, CTL2 contains 10 predicted membrane-spanning regions with multiple N-glycosylation sites. The 68 and 72 kDa molecular forms of inner ear CTL2 are distinguished by sialic acid modification of the carbohydrate. The KHRI-3 antibody binds to an N-linked carbohydrate on CTL2 and presumably damages the organ of Corti by blocking the transporter function of this molecule. CTL2 mRNA and protein are abundantly expressed in human inner ear. Sera from patients with autoimmune hearing loss bind to guinea pig inner ear with the same pattern as CTL2 antibodies. Thus, CTL2 is a possible target of autoimmune hearing loss in humans.

Intracardiac migration of inferior vena cava filters: review of published data.

BACKGROUND: Intracardiac or intrapulmonary migration of inferior vena cava (IVC) filters is an uncommon although potentially life-threatening event that is poorly understood. METHODS: We searched the medical literature and extracted data detailing information concerning the event, including the cause and treatment of the filter migration. Our data were summarized with respect to the filter type, presenting symptoms, associated morbidity and mortality, and success and failure of the treatment provided. RESULTS: Ninety-eight cases of intracardiac or intrapulmonary migration of IVC filters were identified in 77 publications. CONCLUSIONS: Intracardiac and intrapulmonary migration of IVC filters is an uncommon event the etiology of which has been attributed to a variety of causes including operator error, device failure, and patient's physical attributes. Although there is no consensus on patient management, we offer that open thoracotomy has several advantages over the endovascular approach and may provide the best first option. Whenever the patient is not a surgical candidate, endovascular retrieval has been demonstrated to be a relatively safe and viable alternative.