Radiographers' knowledge, attitudes and expectations of artificial intelligence in medical imaging

INTRODUCTION

Artificial intelligence (AI) is increasingly utilised in medical imaging systems and processes, and radiographers must embrace this advancement. This study aimed to investigate perceptions, knowledge, and expectations towards integrating AI into medical imaging amongst a sample of radiographers and determine the current state of AI education within the community.

METHODS

A cross-sectional online quantitative study targeting radiographers based in Europe was conducted over ten weeks. Captured data included demographical information, participants' perceptions and understanding of AI, expectations of AI and AI-related educational backgrounds. Both descriptive and inferential statistical techniques were used to analyse the obtained data.

RESULTS

A total of 96 valid responses were collected. Of these, 64% correctly identified the true definition of AI from a range of options, but fewer (37%) fully understood the difference between AI, machine learning and deep learning. The majority of participants (83%) agreed they were excited about the advancement of AI, though a level of apprehensiveness remained amongst 29%. A severe lack of education on AI was noted, with only 8% of participants having received AI teachings in their pre-registration qualification.

CONCLUSION

Overall positive attitudes towards AI implementation were observed. The slight apprehension may stem from the lack of technical understanding of AI technologies and AI training within the community. Greater educational programs focusing on AI principles are required to help increase European radiography workforce engagement and involvement in AI technologies.

IMPLICATIONS FOR PRACTICE

This study offers insight into the current perspectives of European based radiographers on AI in radiography to help facilitate the embracement of AI technology and convey the need for AI-focused education within the profession.

Dendrites use mechanosensitive channels to proofread ligand-mediated neurite extension during morphogenesis

Ligand-receptor interactions guide axon navigation and dendrite arborization. Mechanical forces also influence guidance choices. However, the nature of such mechanical stimulations, the mechanosensor identity, and how they interact with guidance receptors are unknown. Here, we demonstrate that mechanosensitive DEG/ENaC channels are required for dendritic arbor morphogenesis in Caenorhabditis elegans. Inhibition of DEG/ENaC channels causes reduced dendritic outgrowth and branching in vivo, a phenotype that is alleviated by overexpression of the mechanosensitive channels PEZO-1/Piezo or YVC1/TrpY1. DEG/ENaCs trigger local Ca²⁺ transients in growing dendritic filopodia via activation of L-type voltage-gated Ca²⁺ channels. Anchoring of filopodia by dendrite ligand-receptor complexes is required for the mechanical activation of DEG/ENaC channels. Therefore, mechanosensitive channels serve as a checkpoint for appropriate chemoaffinity by activating Ca²⁺ transients required for neurite growth.

The metalloprotease ADM-4/ADAM17 promotes axonal repair

Axonal fusion is an efficient means of repair following axonal transection, whereby the regenerating axon fuses with its own separated axonal fragment to restore neuronal function. Despite being described over 50 years ago, its molecular mechanisms remain poorly understood. Here, we demonstrate that the Caenorhabditis elegans metalloprotease ADM-4, an ortholog of human ADAM17, is essential for axonal fusion. We reveal that animals lacking ADM-4 cannot repair their axons by fusion, and that ADM-4 has a cell-autonomous function within injured neurons, localizing at the tip of regrowing axon and fusion sites. We demonstrate that ADM-4 overexpression enhances fusion to levels higher than wild type, and that the metalloprotease and phosphatidylserine-binding domains are essential for its function. Last, we show that ADM-4 interacts with and stabilizes the fusogen EFF-1 to allow membranes to merge. Our results uncover a key role for ADM-4 in axonal fusion, exposing a molecular target for axonal repair.

Neuron-epidermal attachment protects hyper-fragile axons from mechanical strain

Axons experience significant strain caused by organismal development and movement. A combination of intrinsic mechanical resistance and external shielding by surrounding tissues prevents axonal damage, although the precise mechanisms are unknown. Here, we reveal a neuroprotective function of neuron-epidermal attachment in Caenorhabditis elegans. We show that a gain-of-function mutation in the epidermal hemidesmosome component LET-805/myotactin, in combination with a loss-of-function mutation in UNC-70/β-spectrin, disrupts the uniform attachment and subsequent embedment of sensory axons within the epidermis during development. This generates regions of high tension within axons, leading to spontaneous axonal breaks and degeneration. Completely preventing attachment, by disrupting HIM-4/hemicentin or MEC-5/collagen, eliminates tension and alleviates damage. Finally, we demonstrate that progressive neuron-epidermal attachment via LET-805/myotactin is induced by the axon during development, as well as during regeneration after injury. Together, these results reveal that establishment of uniform neuron-epidermal attachment is critical to protect axons from mechanical strain during development.

Epidermal control of axonal attachment via β-spectrin and the GTPase-activating protein TBC-10 prevents axonal degeneration

Neurons are subjected to strain due to body movement and their location within organs and tissues. However, how they withstand these forces over the lifetime of an organism is still poorly understood. Here, focusing on touch receptor neuron-epidermis interactions using Caenorhabditis elegans as a model system, we show that UNC-70/β-spectrin and TBC-10, a conserved GTPase-activating protein, function non-cell-autonomously within the epidermis to dynamically maintain attachment of the axon. We reveal that, in response to strain, UNC-70/β-spectrin and TBC-10 stabilize trans-epidermal hemidesmosome attachment structures which otherwise become lost, causing axonal breakage and degeneration. Furthermore, we show that TBC-10 regulates axonal attachment and maintenance by inactivating RAB-35, and reveal functional conservation of these molecules with their vertebrate orthologs. Finally, we demonstrate that β-spectrin functions in this context non-cell-autonomously. We propose a model in which mechanically resistant epidermal attachment structures are maintained by UNC-70/β-spectrin and TBC-10 during movement, preventing axonal detachment and degeneration.

6-OHDA-induced dopaminergic neurodegeneration in Caenorhabditis elegans is promoted by the engulfment pathway and inhibited by the transthyretin-related protein TTR-33

Oxidative stress is linked to many pathological conditions including the loss of dopaminergic neurons in Parkinson’s disease. The vast majority of disease cases appear to be caused by a combination of genetic mutations and environmental factors. We screened for genes protecting Caenorhabditis elegans dopaminergic neurons from oxidative stress induced by the neurotoxin 6-hydroxydopamine (6-OHDA) and identified the transthyretin-related gene ttr-33. The only described C. elegans transthyretin-related protein to date, TTR-52, has been shown to mediate corpse engulfment as well as axon repair. We demonstrate that TTR-52 and TTR-33 have distinct roles. TTR-33 is likely produced in the posterior arcade cells in the head of C. elegans larvae and is predicted to be a secreted protein. TTR-33 protects C. elegans from oxidative stress induced by paraquat or H₂O₂ at an organismal level. The increased oxidative stress sensitivity of ttr-33 mutants is alleviated by mutations affecting the KGB-1 MAPK kinase pathway, whereas it is enhanced by mutation of the JNK-1 MAPK kinase. Finally, we provide genetic evidence that the C. elegans cell corpse engulfment pathway is required for the degeneration of dopaminergic neurons after exposure to 6-OHDA. In summary, we describe a new neuroprotective mechanism and demonstrate that TTR-33 normally functions to protect dopaminergic neurons from oxidative stress-induced degeneration, potentially by acting as a secreted sensor or scavenger of oxidative stress.

Animals employ multiple mechanisms to prevent their cells from damage by reactive oxygen species, chemically reactive molecules containing oxygen. Oxidative stress, caused by the overabundance of reactive oxygen species or a decreased cellular defence against these chemicals, is linked to a variety of neurodegenerative conditions, including the loss of dopaminergic neurons in Parkinson’s disease. In this study, we discovered a novel protective molecule that functions to prevent dopaminergic neurodegeneration caused by oxidative stress induced by the neurotoxin 6-hydroxydopamine (6-OHDA). We used the nematode C. elegans, a well-characterised model in which mechanisms can be studied on an organismal level. When C. elegans is exposed to 6-OHDA, its dopaminergic neurons gradually die. Our major findings include (i) mutations of the transthyretin-related gene ttr-33 causes highly increased dopaminergic neurodegeneration after 6-OHDA exposure; (ii) TTR-33 is likely produced and secreted by several cells in the head of the animal; (iii) TTR-33 protects against oxidative stress induced by other compounds; (iv) mutations in the KGB-1 MAP kinase stress pathway alleviate dopaminergic neuron loss in the ttr-33 mutant; and (v) the cell corpse engulfment pathway is required for dopaminergic neurodegeneration. We hypothesise that TTR-33 protects dopaminergic neurons against 6-OHDA-induced oxidative stress by acting as an oxygen sensor or scavenger.

Rapid and permanent neuronal inactivation in vivo via subcellular generation of reactive oxygen with the use of KillerRed

Inactivation of selected neurons in vivo can define their contribution to specific developmental outcomes, circuit functions, and behaviors. Here, we show that the optogenetic tool KillerRed selectively, rapidly, and permanently inactivates different classes of neurons in C. elegans in response to a single light stimulus, through the generation of reactive oxygen species (ROS). Ablation scales from individual neurons in single animals to multiple neurons in populations and can be applied to freely behaving animals. Using spatially restricted illumination, we demonstrate that localized KillerRed activation in either the cell body or the axon triggers neuronal degeneration and death of the targeted cell. Finally, targeting KillerRed to mitochondria results in organelle fragmentation without killing the cell, in contrast to the cell death observed when KillerRed is targeted to the plasma membrane. We expect this genetic tool to have wide-ranging applications in studies of circuit function and subcellular responses to ROS.

A multi-channel device for high-density target-selective stimulation and long-term monitoring of cells and subcellular features in C. elegans

Selective cell ablation can be used to identify neuronal functions in multicellular model organisms such as Caenorhabditis elegans. The optogenetic tool KillerRed facilitates selective ablation by enabling light-activated damage of cell or subcellular components in a temporally and spatially precise manner. However, the use of KillerRed requires stimulating (5 min-1 h), culturing (~24 h) and imaging (often repeatedly) a large number of individual animals. Current manual manipulation methods are limited by their time-consuming, labor-intensive nature, and their usage of anesthetics. To facilitate large-scale selective ablation, culturing, and repetitive imaging, we developed a densely-packed multi-channel device and used it to perform high-throughput neuronal ablation on KillerRed-expressing animals. The ability to load worms in identical locations with high loading efficiency allows us to ablate selected neurons in multiple worms simultaneously. Our device also enables continuous observation of animals for 24 h following KillerRed activation, and allows the animals to be recovered for behavioural assays. We expect this multi-channel device to facilitate a broad range of long-term imaging and selective illumination experiments in neuroscience.

A dominant mutation in mec-7/β-tubulin affects axon development and regeneration in Caenorhabditis elegans neurons

Microtubules are the basic elements of the cytoskeleton. This study demonstrates that a specific mutation in mec-7/β-tubulin is necessary for the correct number of neurites a neuron extends in vivo and the neuron’s capacity for axonal regeneration following injury.

Microtubules have been known for decades to be basic elements of the cytoskeleton. They form long, dynamic, rope-like structures within the cell that are essential for mitosis, maintenance of cell shape, and intracellular transport. More recently, in vitro studies have implicated microtubules as signaling molecules that, through changes in their stability, have the potential to trigger growth of axons and dendrites in developing neurons. In this study, we show that specific mutations in the Caenorhabditis elegans mec-7/β-tubulin gene cause ectopic axon formation in mechanosensory neurons in vivo. In mec-7 mutants, the ALM mechanosensory neuron forms a long ectopic neurite that extends posteriorly, a phenotype that can be mimicked in wild-type worms with a microtubule-stabilizing drug (paclitaxel), and suppressed by mutations in unc-33/CRMP2 and the kinesin-related gene, vab-8. Our results also reveal that these ectopic neurites contain RAB-3, a marker for presynaptic loci, suggesting that they have axon-like properties. Interestingly, in contrast with the excessive axonal growth observed during development, mec-7 mutants are inhibited in axonal regrowth and remodeling following axonal injury. Together our results suggest that MEC-7/β-tubulin integrity is necessary for the correct number of neurites a neuron generates in vivo and for the capacity of an axon to regenerate.

A phase II trial of first-line bevacizumab in combination with pemetrexed and carboplatin in advanced nonsquamous non-small cell lung cancer (NSCLC)

e19091

Background: Combining bevacizumab (Bev) with platinum doublet yields a survival benefit in previously untreated patients with advanced nonsquamous non-small cell lung cancer. Carboplatin with pemetrexed is an efficient combination that is potentially less toxic. This phase II trial evaluates the efficacy and the safety of adding bevacizumab to carboplatin and pemetrexed in advanced non-small cell lung cancer. Methods: This is a single arm, open-label phase II trial. Eligibility stipulated enrollment of patients with stage IIIB/IV non-squamous non-small cell lung cancer with an ECOG PS 0–1 and no other contraindications to Bev. Treatment consisted of pemetrexed 500 mg/m2, carbo AUC 6 and Bev 15mg/kg on day 1, repeated every 3 weeks. If clinical benefit was proven, patients received Bev maintenance until progression. The response was evaluated every two cycles. The primary endpoint was time to progression (TTP); secondary endpoints included overall survival, response rate (RR) and grade 3–4 toxicities. Results: As of September 2008, 27 out of 42 projected patients were enrolled of which, 14 were males and 13 females; median age 68 (72% older than 60); IIIB/IV 5/22; median number of cycles 9 (range 2–33). At the time of analysis, 25 patients were evaluable. Preliminary results show a median time to progression of 7.25 months (95% CI, 6.4 to 8.2). 23 patients (92%) had disease control (95%, CI 82–96%) with an overall response rate of 36% (95%, CI 25–47%).14 patients had stable disease (56%), 8 had partial response (32%) and one achieved a complete response (4 %). Median survival has not yet been reached. Grade 3/4 adverse events include: Cerebral ischemic event (N=1), Anorexia (Gr 3: N= 5/ Gr 4: N= 4), gastrointestinal fistula (N=1), hypertension (Gr3: N=3), epistaxis (Gr 3 N=1), vomiting (Gr 3; N= 3), fever (Gr 4: N=1) and neuropathy (Gr 3: N=2). Conclusion: The addition of bevacizumab to carboplatin and pemetrexed appears to be an active first line treatment with a high disease control rate and an acceptable toxicity profile in the setting of advanced non-small cell lung cancer.

No significant financial relationships to disclose.

Fungal peritonitis in pediatric patients

Fungal peritonitis (FP) is a rare complication of peritoneal dialysis (PD). Although treatment with fluconazole (FCZ) has improved catheter survival and preservation of the peritoneal membrane, FP still carries a high morbidity and mortality in pediatrics. High-risk factors for FP include previous usage of systemic antibiotics and recurrent bacterial peritonitis. A prospective experience in the treatment of FP was conducted at the University of Miami/Jackson Children's Hospital from 1992 to 1997. All patients received either oral or intravenous loading dose of FCZ (5-7 mg/kg) followed by intraperitoneal (i.p.) FCZ (75 mg/L). Amphotericin B (amp B) was added when clinical sepsis was present. A total of 6 patients had FP (all Candida sp.; mean age: 6 years). Two of these patients were neonates with Tenckhoff-catheter placement at less than 1 week of age. Five patients achieved sterilization of the peritoneal fluid. One patient required catheter removal (C. tropicalis). The 2 neonates were infection free for 29 and 41 days, respectively, but both died of superimposed bacterial sepsis. The remaining 4 patients survived and completed 6 weeks of FCZ treatment. Two have had preservation of the peritoneal membrane for more than 1 year. The other 2 were switched to hemodialysis. We conclude that FCZ is an effective treatment for fungal peritonitis in pediatric patients. Adjunct therapy with amp B is usually necessary if sepsis is present. Although eradication of the fungus is possible in a majority of cases, neonates and immunocompromised hosts remain at high risk for morbidity and mortality.