Effects of a comprehensive brain computed tomography deep learning model on radiologist detection accuracy.

OBJECTIVES: Non-contrast computed tomography of the brain (NCCTB) is commonly used to detect intracranial pathology but is subject to interpretation errors. Machine learning can augment clinical decision-making and improve NCCTB scan interpretation. This retrospective detection accuracy study assessed the performance of radiologists assisted by a deep learning model and compared the standalone performance of the model with that of unassisted radiologists. METHODS: A deep learning model was trained on 212,484 NCCTB scans drawn from a private radiology group in Australia. Scans from inpatient, outpatient, and emergency settings were included. Scan inclusion criteria were age ≥ 18 years and series slice thickness ≤ 1.5 mm. Thirty-two radiologists reviewed 2848 scans with and without the assistance of the deep learning system and rated their confidence in the presence of each finding using a 7-point scale. Differences in AUC and Matthews correlation coefficient (MCC) were calculated using a ground-truth gold standard. RESULTS: The model demonstrated an average area under the receiver operating characteristic curve (AUC) of 0.93 across 144 NCCTB findings and significantly improved radiologist interpretation performance. Assisted and unassisted radiologists demonstrated an average AUC of 0.79 and 0.73 across 22 grouped parent findings and 0.72 and 0.68 across 189 child findings, respectively. When assisted by the model, radiologist AUC was significantly improved for 91 findings (158 findings were non-inferior), and reading time was significantly reduced. CONCLUSIONS: The assistance of a comprehensive deep learning model significantly improved radiologist detection accuracy across a wide range of clinical findings and demonstrated the potential to improve NCCTB interpretation. CLINICAL RELEVANCE STATEMENT: This study evaluated a comprehensive CT brain deep learning model, which performed strongly, improved the performance of radiologists, and reduced interpretation time. The model may reduce errors, improve efficiency, facilitate triage, and better enable the delivery of timely patient care. KEY POINTS: • This study demonstrated that the use of a comprehensive deep learning system assisted radiologists in the detection of a wide range of abnormalities on non-contrast brain computed tomography scans. • The deep learning model demonstrated an average area under the receiver operating characteristic curve of 0.93 across 144 findings and significantly improved radiologist interpretation performance. • The assistance of the comprehensive deep learning model significantly reduced the time required for radiologists to interpret computed tomography scans of the brain.

Complete Genome Sequence of Streptomyces Phage ϕRKBJ001 Isolated from Prince Edward Island, Canada.

We reported here the complete genome sequence of Streptomyces phage ϕRKBJ001 that was isolated from a saltwater marsh on Prince Edward Island, Canada, using the Streptomyces sp. strain RKBHB0173. Based on electron microscopy and genomic analysis, this phage belongs to the Siphoviridae family and the BN Streptomyces phage cluster.

Effect of a comprehensive deep-learning model on the accuracy of chest x-ray interpretation by radiologists: a retrospective, multireader multicase study.

BACKGROUND: Chest x-rays are widely used in clinical practice; however, interpretation can be hindered by human error and a lack of experienced thoracic radiologists. Deep learning has the potential to improve the accuracy of chest x-ray interpretation. We therefore aimed to assess the accuracy of radiologists with and without the assistance of a deep-learning model. METHODS: In this retrospective study, a deep-learning model was trained on 821 681 images (284 649 patients) from five data sets from Australia, Europe, and the USA. 2568 enriched chest x-ray cases from adult patients (≥16 years) who had at least one frontal chest x-ray were included in the test dataset; cases were representative of inpatient, outpatient, and emergency settings. 20 radiologists reviewed cases with and without the assistance of the deep-learning model with a 3-month washout period. We assessed the change in accuracy of chest x-ray interpretation across 127 clinical findings when the deep-learning model was used as a decision support by calculating area under the receiver operating characteristic curve (AUC) for each radiologist with and without the deep-learning model. We also compared AUCs for the model alone with those of unassisted radiologists. If the lower bound of the adjusted 95% CI of the difference in AUC between the model and the unassisted radiologists was more than -0·05, the model was considered to be non-inferior for that finding. If the lower bound exceeded 0, the model was considered to be superior. FINDINGS: Unassisted radiologists had a macroaveraged AUC of 0·713 (95% CI 0·645-0·785) across the 127 clinical findings, compared with 0·808 (0·763-0·839) when assisted by the model. The deep-learning model statistically significantly improved the classification accuracy of radiologists for 102 (80%) of 127 clinical findings, was statistically non-inferior for 19 (15%) findings, and no findings showed a decrease in accuracy when radiologists used the deep-learning model. Unassisted radiologists had a macroaveraged mean AUC of 0·713 (0·645-0·785) across all findings, compared with 0·957 (0·954-0·959) for the model alone. Model classification alone was significantly more accurate than unassisted radiologists for 117 (94%) of 124 clinical findings predicted by the model and was non-inferior to unassisted radiologists for all other clinical findings. INTERPRETATION: This study shows the potential of a comprehensive deep-learning model to improve chest x-ray interpretation across a large breadth of clinical practice. FUNDING: Annalise.ai.

Herpesviruses and the Unfolded Protein Response.

Herpesviruses usurp cellular stress responses to promote viral replication and avoid immune surveillance. The unfolded protein response (UPR) is a conserved stress response that is activated when the protein load in the ER exceeds folding capacity and misfolded proteins accumulate. The UPR aims to restore protein homeostasis through translational and transcriptional reprogramming; if homeostasis cannot be restored, the UPR switches from "helper" to "executioner", triggering apoptosis. It is thought that the burst of herpesvirus glycoprotein synthesis during lytic replication causes ER stress, and that these viruses may have evolved mechanisms to manage UPR signaling to create an optimal niche for replication. The past decade has seen considerable progress in understanding how herpesviruses reprogram the UPR. Here we provide an overview of the molecular events of UPR activation, signaling and transcriptional outputs, and highlight key evidence that herpesviruses hijack the UPR to aid infection.

KSHV activates unfolded protein response sensors but suppresses downstream transcriptional responses to support lytic replication.

Herpesviruses usurp host cell protein synthesis machinery to convert viral mRNAs into proteins, and the endoplasmic reticulum (ER) to ensure proper folding, post-translational modification and trafficking of secreted and transmembrane viral proteins. Overloading ER folding capacity activates the unfolded protein response (UPR), whereby sensor proteins ATF6, PERK and IRE1 initiate a stress-mitigating transcription program that accelerates catabolism of misfolded proteins while increasing ER folding capacity. Kaposi's sarcoma-associated herpesvirus (KSHV) can be reactivated from latency by chemical induction of ER stress, which causes accumulation of the XBP1s transcription factor that transactivates the viral RTA lytic switch gene. The presence of XBP1s-responsive elements in the RTA promoter suggests that KSHV evolved a mechanism to respond to ER stress. Here, we report that ATF6, PERK and IRE1 were activated upon reactivation from latency and required for efficient KSHV lytic replication; genetic or pharmacologic inhibition of each UPR sensor diminished virion production. Despite UPR sensor activation during KSHV lytic replication, downstream UPR transcriptional responses were restricted; 1) ATF6 was cleaved to activate the ATF6(N) transcription factor but ATF6(N)-responsive genes were not transcribed; 2) PERK phosphorylated eIF2α but ATF4 did not accumulate; 3) IRE1 caused XBP1 mRNA splicing, but XBP1s protein did not accumulate and XBP1s-responsive genes were not transcribed. Ectopic expression of the KSHV host shutoff protein SOX did not affect UPR gene expression, suggesting that alternative viral mechanisms likely mediate UPR suppression during lytic replication. Complementation of XBP1s deficiency during KSHV lytic replication inhibited virion production in a dose-dependent manner in iSLK.219 cells but not in TREx-BCBL1-RTA cells. However, genetically distinct KSHV virions harvested from these two cell lines were equally susceptible to XBP1s restriction following infection of naïve iSLK cells. This suggests that cell-intrinsic properties of BCBL1 cells may circumvent the antiviral effect of ectopic XBP1s expression. Taken together, these findings indicate that while XBP1s plays an important role in reactivation from latency, it can inhibit virus replication at a later step, which the virus overcomes by preventing its synthesis. These findings suggest that KSHV hijacks UPR sensors to promote efficient viral replication while sustaining ER stress.

Viral activation of MK2-hsp27-p115RhoGEF-RhoA signaling axis causes cytoskeletal rearrangements, p-body disruption and ARE-mRNA stabilization.

Kaposi's sarcoma-associated herpesvirus (KSHV) is the infectious cause of several AIDS-related cancers, including the endothelial cell (EC) neoplasm Kaposi's sarcoma (KS). KSHV-infected ECs secrete abundant host-derived pro-inflammatory molecules and angiogenic factors that contribute to tumorigenesis. The precise contributions of viral gene products to this secretory phenotype remain to be elucidated, but there is emerging evidence for post-transcriptional regulation. The Kaposin B (KapB) protein is thought to contribute to the secretory phenotype in infected cells by binding and activating the stress-responsive kinase MK2, thereby selectively blocking decay of AU-rich mRNAs (ARE-mRNAs) encoding pro-inflammatory cytokines and angiogenic factors. Processing bodies (PBs) are cytoplasmic ribonucleoprotein foci in which ARE-mRNAs normally undergo rapid 5' to 3' decay. Here, we demonstrate that PB dispersion is a feature of latent KSHV infection, which is dependent on kaposin protein expression. KapB is sufficient to disperse PBs, and KapB-mediated ARE-mRNA stabilization could be partially reversed by treatments that restore PBs. Using a combination of genetic and chemical approaches we provide evidence that KapB-mediated PB dispersion is dependent on activation of a non-canonical Rho-GTPase signaling axis involving MK2, hsp27, p115RhoGEF and RhoA. PB dispersion in latently infected cells is likewise dependent on p115RhoGEF. In addition to PB dispersion, KapB-mediated RhoA activation in primary ECs caused actin stress fiber formation, increased cell motility and angiogenesis; these effects were dependent on the activity of the RhoA substrate kinases ROCK1/2. By contrast, KapB-mediated PB dispersion occurred in a ROCK1/2-independent manner. Taken together, these observations position KapB as a key contributor to viral reprogramming of ECs, capable of eliciting many of the phenotypes characteristic of KS tumor cells, and strongly contributing to the post-transcriptional control of EC gene expression and secretion.

Influenza a virus host shutoff disables antiviral stress-induced translation arrest.

Influenza A virus (IAV) polymerase complexes function in the nucleus of infected cells, generating mRNAs that bear 5' caps and poly(A) tails, and which are exported to the cytoplasm and translated by host machinery. Host antiviral defences include mechanisms that detect the stress of virus infection and arrest cap-dependent mRNA translation, which normally results in the formation of cytoplasmic aggregates of translationally stalled mRNA-protein complexes known as stress granules (SGs). It remains unclear how IAV ensures preferential translation of viral gene products while evading stress-induced translation arrest. Here, we demonstrate that at early stages of infection both viral and host mRNAs are sensitive to drug-induced translation arrest and SG formation. By contrast, at later stages of infection, IAV becomes partially resistant to stress-induced translation arrest, thereby maintaining ongoing translation of viral gene products. To this end, the virus deploys multiple proteins that block stress-induced SG formation: 1) non-structural protein 1 (NS1) inactivates the antiviral double-stranded RNA (dsRNA)-activated kinase PKR, thereby preventing eIF2α phosphorylation and SG formation; 2) nucleoprotein (NP) inhibits SG formation without affecting eIF2α phosphorylation; 3) host-shutoff protein polymerase-acidic protein-X (PA-X) strongly inhibits SG formation concomitant with dramatic depletion of cytoplasmic poly(A) RNA and nuclear accumulation of poly(A)-binding protein. Recombinant viruses with disrupted PA-X host shutoff function fail to effectively inhibit stress-induced SG formation. The existence of three distinct mechanisms of IAV-mediated SG blockade reveals the magnitude of the threat of stress-induced translation arrest during viral replication.

Kaposi's sarcoma-associated herpesvirus G-protein-coupled receptor prevents AU-rich-element-mediated mRNA decay.

During lytic Kaposi's sarcoma-associated herpesvirus (KSHV) infection, host gene expression is severely restricted by a process of global mRNA degradation known as host shutoff, which rededicates translational machinery to the expression of viral proteins. A subset of host mRNAs is spared from shutoff, and a number of these contain cis-acting AU-rich elements (AREs) in their 3' untranslated regions. AREs are found in labile mRNAs encoding cytokines, growth factors, and proto-oncogenes. Activation of the p38/MK2 signal transduction pathway reverses constitutive decay of ARE-mRNAs, resulting in increased protein production. The viral G-protein-coupled receptor (vGPCR) is thought to play an important role in promoting the secretion of angiogenic molecules from KSHV-infected cells during lytic replication, but to date it has not been clear how vGPCR circumvents host shutoff. Here, we demonstrate that vGPCR activates the p38/MK2 pathway and stabilizes ARE-mRNAs, augmenting the levels of their protein products. Using MK2-deficient cells, we demonstrate that MK2 is essential for maximal vGPCR-mediated ARE-mRNA stabilization. ARE-mRNAs are normally delivered to cytoplasmic ribonucleoprotein granules known as processing bodies (PBs) for translational silencing and decay. We demonstrate that PB formation is prevented during KSHV lytic replication or in response to vGPCR-mediated activation of RhoA subfamily GTPases. Together, these data show for the first time that vGPCR impacts gene expression at the posttranscriptional level, coordinating an attack on the host mRNA degradation machinery. By suppressing ARE-mRNA turnover, vGPCR may facilitate escape of certain target mRNAs from host shutoff and allow secretion of angiogenic factors from lytically infected cells.