Smartphone Augmented Reality Outperforms Conventional CT Guidance for Composite Ablation Margins in Phantom Models.

PURPOSE: To develop and evaluate a smartphone augmented reality (AR) system for a large 50-mm liver tumor ablation with treatment planning for composite overlapping ablation zones. MATERIALS AND METHODS: A smartphone AR application was developed to display tumor, probe, projected probe paths, ablated zones, and real-time percentage of the ablated target tumor volume. Fiducial markers were attached to phantoms and an ablation probe hub for tracking. The system was evaluated with tissue-mimicking thermochromic phantoms and gel phantoms. Four interventional radiologists performed 2 trials each of 3 probe insertions per trial using AR guidance versus computed tomography (CT) guidance approaches in 2 gel phantoms. Insertion points and optimal probe paths were predetermined. On Gel Phantom 2, serial ablated zones were saved and continuously displayed after each probe placement/adjustment, enabling feedback and iterative planning. The percentages of tumor ablated for AR guidance versus CT guidance, and with versus without display of recorded ablated zones, were compared among interventional radiologists with pairwise t-tests. RESULTS: The means of percentages of tumor ablated for CT freehand and AR guidance were 36% ± 7 and 47% ± 4 (P = .004), respectively. The mean composite percentages of tumor ablated for AR guidance were 43% ± 1 (without) and 50% ± 2 (with display of ablation zone) (P = .033). There was no strong correlation between AR-guided percentage of ablation and years of experience (r < 0.5), whereas there was a strong correlation between CT-guided percentage of ablation and years of experience (r > 0.9). CONCLUSIONS: A smartphone AR guidance system for dynamic iterative large liver tumor ablation was accurate, performed better than conventional CT guidance, especially for less experienced interventional radiologists, and enhanced more standardized performance across experience levels for ablation of a 50-mm tumor.

Challenges, Barriers, and Successes of Standardized Report Templates: Results of a Society of Interventional Radiology Survey.

Registry data are being increasingly used to establish treatment guidelines, set benchmarks, allocate resources, and make payment decisions. Although many registries rely on manual data entry, the Society of Interventional Radiology (SIR) is using automated data extraction for its VIRTEX registry. This process relies on participants using consistent terminology with highly structured data in physician-developed standardized reports (SR). To better understand barriers to adoption, a survey was sent to 3,178 SIR members. Responses were obtained from 451 interventional radiology practitioners (14.2%) from 92 unique academic and 151 unique private practices. Of these, 75% used structured reports and 32% used the SIR SR. The most common barriers to the use of these reports include SR length (35% of respondents), lack of awareness about the SR (31%), and lack of agreement on adoption within practices (27%). The results demonstrated insights regarding barriers in the use and/or adoption of SR and potential solutions.

A History and Modern Framework for Quality Improvement in Interventional Radiology.

Quality improvement (QI) initiatives have benefited patients as well as the broader practice of medicine. Large-scale QI has been facilitated by multi-institutional data registries, many of which were formed out of national or international medical society initiatives. With broad participation, QI registries have provided benefits that include but are not limited to establishing treatment guidelines, facilitating research related to uncommon procedures and conditions, and demonstrating the fiscal and clinical value of procedures for both medical providers and health systems. Because of the benefits offered by these databases, Society of Interventional Radiology (SIR) and SIR Foundation have committed to the development of an interventional radiology (IR) clinical data registry known as VIRTEX. A large IR database with participation from a multitude of practice environments has the potential to have a significant positive impact on the specialty through data-driven advances in patient safety and outcomes, clinical research, and reimbursement. This article reviews the current landscape of societal QI programs, presents a vision for a large-scale IR clinical data registry supported by SIR, and discusses the anticipated results that such a framework can produce.

Phase 1 trial to model primary, secondary, and tertiary dengue using a monovalent vaccine.

BACKGROUND: The four co-circulating and immunologically interactive dengue virus serotypes (DENV1-4) pose a unique challenge to vaccine design because sub-protective immunity can increase the risk of severe dengue disease. Existing dengue vaccines have lower efficacy in DENV seronegative individuals but higher efficacy in DENV exposed individuals. There is an urgent need to identify immunological measures that are strongly associated with protection against viral replication and disease following sequential exposure to distinct serotypes. METHODS/DESIGN: This is a phase 1 trial wherein healthy adults with neutralizing antibodies to zero (seronegative), one non-DENV3 (heterotypic), or more than one (polytypic) DENV serotype will be vaccinated with the live attenuated DENV3 monovalent vaccine rDEN3Δ30/31-7164. We will examine how pre-vaccine host immunity influences the safety and immunogenicity of DENV3 vaccination in a non-endemic population. We hypothesize that the vaccine will be safe and well tolerated, and all groups will have a significant increase in the DENV1-4 neutralizing antibody geometric mean titer between days 0 and 28. Compared to the seronegative group, the polytypic group will have lower mean peak vaccine viremia, due to protection conferred by prior DENV exposure, while the heterotypic group will have higher mean peak viremia, due to mild enhancement. Secondary and exploratory endpoints include characterizing serological, innate, and adaptive cell responses; evaluating proviral or antiviral contributions of DENV-infected cells; and immunologically profiling the transcriptome, surface proteins, and B and T cell receptor sequences and affinities of single cells in both peripheral blood and draining lymph nodes sampled via serial image-guided fine needle aspiration. DISCUSSION: This trial will compare the immune responses after primary, secondary, and tertiary DENV exposure in naturally infected humans living in non-endemic areas. By evaluating dengue vaccines in a new population and modeling the induction of cross-serotypic immunity, this work may inform vaccine evaluation and broaden potential target populations. TRIAL REGISTRATION: NCT05691530 registered on January 20, 2023.

Woodchuck Hepatic Anatomy and Vascular Alterations Due to Hepatocellular Carcinoma with Angiographic Atlas of the Abdomen and Pelvis.

PURPOSE: To characterize the hepatic and abdominal angiographic anatomy of woodchucks and vascular changes associated with hepatocellular carcinoma (HCC). MATERIALS AND METHODS: Twenty-nine woodchucks (23 with viral-associated HCC, 6 without) underwent multiphasic computed tomography (CT). Fourteen woodchucks (8 with HCC) also underwent diagnostic angiography. Hepatic arterial diameters were measured on the CT scans. Woodchucks were divided into 3 groups: non-tumor-bearing, largest tumor supplied by the right hepatic artery (RHA), and largest tumor supplied by the left hepatic artery (LHA). Statistical analysis with a repeated measures model was performed to determine the effects of tumor location (right, left), vessel measured (RHA, LHA), and interaction between the 2 on vessel diameter. Lobar arteries supplying HCC were compared with those that did not. RESULTS: CT anatomy and normal and variant vascular anatomy were defined. In woodchucks with HCC, LHA and RHA supplying tumors had mean diameters of 2.0 mm ± 0.3 and 1.6 mm ± 0.3 versus 1.5 mm ± 0.3 and 1.1 mm ± 0.2 for non-tumor-supplying arteries (P = .0002 and P < .0001), respectively. Lobar arteries supplying tumors were similarly ectatic. The right lateral lobe artery had the most profound increase in the mean diameter when supplying tumors, measuring 1.7 mm ± 0.1 versus 1.0 mm ± 0.1 in the non-tumor-supplying artery (P < .0001). There were no differences in the diameters of the aorta and celiac, common, and proper hepatic arteries between tumor- and non-tumor-bearing woodchucks. An angiographic atlas of the abdominal vessels was generated. CONCLUSIONS: HCC tumoral vasculature in woodchucks was ectatic compared with normal vasculature. This phenomenon recapitulates human HCC and may facilitate investigation of transcatheter and drug delivery therapies in an HCC animal model.

CT and clinical assessment in asymptomatic and pre-symptomatic patients with early SARS-CoV-2 in outbreak settings.

OBJECTIVES: The early infection dynamics of patients with SARS-CoV-2 are not well understood. We aimed to investigate and characterize associations between clinical, laboratory, and imaging features of asymptomatic and pre-symptomatic patients with SARS-CoV-2. METHODS: Seventy-four patients with RT-PCR-proven SARS-CoV-2 infection were asymptomatic at presentation. All were retrospectively identified from 825 patients with chest CT scans and positive RT-PCR following exposure or travel risks in outbreak settings in Japan and China. CTs were obtained for every patient within a day of admission and were reviewed for infiltrate subtypes and percent with assistance from a deep learning tool. Correlations of clinical, laboratory, and imaging features were analyzed and comparisons were performed using univariate and multivariate logistic regression. RESULTS: Forty-eight of 74 (65%) initially asymptomatic patients had CT infiltrates that pre-dated symptom onset by 3.8 days. The most common CT infiltrates were ground glass opacities (45/48; 94%) and consolidation (22/48; 46%). Patient body temperature (p < 0.01), CRP (p < 0.01), and KL-6 (p = 0.02) were associated with the presence of CT infiltrates. Infiltrate volume (p = 0.01), percent lung involvement (p = 0.01), and consolidation (p = 0.043) were associated with subsequent development of symptoms. CONCLUSIONS: COVID-19 CT infiltrates pre-dated symptoms in two-thirds of patients. Body temperature elevation and laboratory evaluations may identify asymptomatic patients with SARS-CoV-2 CT infiltrates at presentation, and the characteristics of CT infiltrates could help identify asymptomatic SARS-CoV-2 patients who subsequently develop symptoms. The role of chest CT in COVID-19 may be illuminated by a better understanding of CT infiltrates in patients with early disease or SARS-CoV-2 exposure. KEY POINTS: • Forty-eight of 74 (65%) pre-selected asymptomatic patients with SARS-CoV-2 had abnormal chest CT findings. • CT infiltrates pre-dated symptom onset by 3.8 days (range 1-5). • KL-6, CRP, and elevated body temperature identified patients with CT infiltrates. Higher infiltrate volume, percent lung involvement, and pulmonary consolidation identified patients who developed symptoms.

The impact of introducing a no oral contrast abdominopelvic CT examination (NOCAPE) pathway on radiology turn around times, emergency department length of stay, and patient safety.

This investigation evaluates the impact of the no oral contrast abdominopelvic CT examination (NOCAPE) on radiology turn around time (TAT), emergency department (ED) length of stay (LOS), and patient safety metrics. During a 12-month period at two urban teaching hospitals, 6,409 ED abdominopelvic (AP) CTs were performed to evaluate acute abdominal pain. NOCAPE represented 70.9 % of all ED AP CT examinations with intravenous contrast. Data collection included patient demographics, use of intravenous (IV) and/or oral contrast, order to complete and order to final interpretation TAT, ED LOS, admission, recall and bounce back rates, and comparison and characterization of impressions. The NOCAPE pathway reduced median order to complete TAT by 32 min (22.9 %) compared to IV and oral contrast AP CT examinations (traditional pathway) (P < 0.001). Median order to final TAT was 2.9 h in NOCAPE patients and 3.5 h in the traditional pathway, a 36-min (17.1 %) reduction (P < 0.001). Overall, the NOCAPE pathway reduced ED LOS by a median of 43 min (8.8 %) compared to the traditional pathway (8.2 vs 7.5 h) (P = 0.003). Recall and bounce back rates were 3.2 %, and only one patient had change in impression after oral contrast CT was repeated. The NOCAPE pathway is associated with decreased radiology TAT and ED LOS metrics. The authors suggest that NOCAPE implementation in the ED setting is safe and positively impacts both radiology and emergency medicine workflow.

Smart goggles augmented reality CT-US fusion compared to conventional fusion navigation for percutaneous needle insertion.

PURPOSE: Targeting accuracy determines outcomes for percutaneous needle interventions. Augmented reality (AR) in IR may improve procedural guidance and facilitate access to complex locations. This study aimed to evaluate percutaneous needle placement accuracy using a goggle-based AR system compared to an ultrasound (US)-based fusion navigation system. METHODS: Six interventional radiologists performed 24 independent needle placements in an anthropomorphic phantom (CIRS 057A) in four needle guidance cohorts (n = 6 each): (1) US-based fusion, (2) goggle-based AR with stereoscopically projected anatomy (AR-overlay), (3) goggle AR without the projection (AR-plain), and (4) CT-guided freehand. US-based fusion included US/CT registration with electromagnetic (EM) needle, transducer, and patient tracking. For AR-overlay, US, EM-tracked needle, stereoscopic anatomical structures and targets were superimposed over the phantom. Needle placement accuracy (distance from needle tip to target center), placement time (from skin puncture to final position), and procedure time (time to completion) were measured. RESULTS: Mean needle placement accuracy using US-based fusion, AR-overlay, AR-plain, and freehand was 4.5 ± 1.7 mm, 7.0 ± 4.7 mm, 4.7 ± 1.7 mm, and 9.2 ± 5.8 mm, respectively. AR-plain demonstrated comparable accuracy to US-based fusion (p = 0.7) and AR-overlay (p = 0.06). Excluding two outliers, AR-overlay accuracy became 5.9 ± 2.6 mm. US-based fusion had the highest mean placement time (44.3 ± 27.7 s) compared to all navigation cohorts (p < 0.001). Longest procedure times were recorded with AR-overlay (34 ± 10.2 min) compared to AR-plain (22.7 ± 8.6 min, p = 0.09), US-based fusion (19.5 ± 5.6 min, p = 0.02), and freehand (14.8 ± 1.6 min, p = 0.002). CONCLUSION: Goggle-based AR showed no difference in needle placement accuracy compared to the commercially available US-based fusion navigation platform. Differences in accuracy and procedure times were apparent with different display modes (with/without stereoscopic projections). The AR-based projection of the US and needle trajectory over the body may be a helpful tool to enhance visuospatial orientation. Thus, this study refines the potential role of AR for needle placements, which may serve as a catalyst for informed implementation of AR techniques in IR.

Timing, incidence and management of delayed bleeding after partial nephrectomy in patients at risk for recurrent, bilateral, multifocal renal tumors.

INTRODUCTION: Delayed bleeding is a potentially serious complication after partial nephrectomy (PN), with reported rates of 1%-2%. Patients with multiple renal tumors, including those with hereditary forms of kidney cancer, are often managed with resection of multiple tumors in a single kidney which may increase the risk of delayed bleeding, though outcomes have not previously been reported specifically in this population. The objective of this study was to evaluate the incidence and timing of delayed bleeding as well as the impact of intervention on renal functional outcomes in a cohort primarily made up of patients at risk for bilateral, multifocal renal tumors. METHODS: A retrospective review of a prospectively maintained database of patients with known or suspected predisposition to bilateral, multifocal renal tumors who underwent PN from 2003 to 2023 was conducted. Patients who presented with delayed bleeding were identified. Patients with delayed bleeding were compared to those without. Comparative statistics and univariate logistic regression were used to determine potential risk factors for delayed bleeding. RESULTS: A total of 1256 PN were performed during the study period. Angiographic evidence of pseudoaneurysm, AV fistula and/or extravasation occurred in 24 cases (1.9%). Of these, 21 were symptomatic presenting with gross hematuria in 13 (54.2%), decreasing hemoglobin in 4(16.7%), flank pain in 2(8.3%), and mental status change in 2 (8.3%), while 3 patients were asymptomatic. Median number of resected tumors was 5 (IQR 2-8). All patients underwent angiogram with super-selective embolization. Median time to bleed event was 13.5 days (IQR 7-22). Factors associated with delayed bleeding included open approach (OR 2.2, IQR(1.06-5.46), P = 0.04 and left-sided surgery (OR 4.93, IQR(1.67-14.5), P = 0.004. Selective embolization had little impact on ultimate renal functional outcomes, with a median change of 11% from the baseline eGFR after partial nephrectomy and embolization. One patient required total nephrectomy for refractory bleeding after embolization. CONCLUSIONS: Delayed bleeding after PN in a cohort of patients with multifocal tumors is an infrequent event, with similar rates to single tumor series. Patients should be counseled regarding timing and symptoms of delayed bleeding and multidisciplinary management with interventional radiology is critical for timely diagnosis and treatment.

Hydrogel drug delivery systems for minimally invasive local immunotherapy of cancer.

Although systemic immunotherapy has achieved durable responses and improved survival for certain patients and cancer types, low response rates and immune system-related systemic toxicities limit its overall impact. Intratumoral (intralesional) delivery of immunotherapy is a promising technique to combat mechanisms of tumor immune suppression within the tumor microenvironment and reduce systemic drug exposure and associated side effects. However, intratumoral injections are prone to variable tumor drug distribution and leakage into surrounding tissues, which can compromise efficacy and contribute to toxicity. Controlled release drug delivery systems such as in situ-forming hydrogels are promising vehicles for addressing these challenges by providing improved spatio-temporal control of locally administered immunotherapies with the goal of promoting systemic tumor-specific immune responses and abscopal effects. In this review we will discuss concepts, applications, and challenges in local delivery of immunotherapy using controlled release drug delivery systems with a focus on intratumorally injected hydrogel-based drug carriers.

Functional Imaging of Liver Cancer (FLIC): Study protocol of a phase 2 trial of 18F-DCFPyL PET/CT imaging for patients with hepatocellular carcinoma.

BACKGROUND: While prostate specific membrane antigen (PSMA) is overexpressed in high-grade prostate cancers, it is also expressed in tumor neovasculature and other malignancies, including hepatocellular carcinoma (HCC). Importantly, no functional imaging for HCC is clinically available, making diagnosis and surveillance following local therapies particularly challenging. 18F-DCFPyL binds with high affinity to PSMA yet clears rapidly from the blood pool. PET imaging with 18F-DCFPyL may represent a new tool for staging, surveillance and assessment of treatment response in HCC. The purpose of this Functional Imaging Liver Cancer (FLIC) trial is to assess the ability of 18F-DCFPyL-PET/CT to detect sites of HCC. METHODS: This is a phase II multi-site prospective imaging trial with a plan to enroll 50 subjects with suspected HCC on standard of care CT or MRI and eligible for standard local treatment. Participants will undergo a baseline 18F-DCFPyL-PET/CT, prior to therapy. Subjects will also be scanned with 18F-FDG-PET/CT within 2 weeks of 18F-DCFPyL-PET/CT. Participants will undergo histopathologic assessment and standard of care local treatment for HCC within a multidisciplinary team context. Participants with histopathologic confirmation of HCC and a positive baseline 18F-DCFPyL-PET/CT will undergo a post-treatment 18F-DCFPyL-PET/CT during the first routine follow-up, typically within 4-8 weeks. Subjects with negative baseline 18F-DCFPyL-PET/CT will not be re-scanned after treatment but will remain in follow-up. Participants will be followed for 5-years to assess for progression-free-survival. The primary endpoint is the positive predictive value of 18F-DCFPyL-PET for HCC as confirmed by histopathology. Secondary endpoints include comparison of 18F-DCFPyL-PET/CT with CT, MRI, and 18F-FDG-PET/CT, and evaluation of the value of 18F-DCFPyL-PET/CT in assessing treatment response following local treatment. Exploratory endpoints include next generation sequencing of tumors, and analysis of extracellular vesicles to identify biomarkers associated with response to therapy. DISCUSSION: This is a prospective imaging trial designed to evaluate whether PSMA-PET/CT imaging with 18F-DCFPyL can detect tumor sites, assess local treatment response in HCC patients, and to eventually determine whether PSMA-PET/CT could improve outcomes of patients with HCC receiving standard of care local therapy. Importantly, this trial may help determine whether PSMA-selective radiopharmaceutical therapies may be beneficial for patients with HCC. CLINICAL TRIAL REGISTRATION: NIH IND#133631. Submission date: 04-07-2021. Safe-to-proceed letter issued by FDA: 05.07.2021. NIH IRB #00080. ClinicalTrials.gov Identifier NCT05009979. Date of Registry: 08-18-2021. Protocol version date: 01-07-2022.

Image-Guided Robotics for Standardized and Automated Biopsy and Ablation.

Image-guided robotics for biopsy and ablation aims to minimize procedure times, reduce needle manipulations, radiation, and complications, and enable treatment of larger and more complex tumors, while facilitating standardization for more uniform and improved outcomes. Robotic navigation of needles enables standardized and uniform procedures which enhance reproducibility via real-time precision feedback, while avoiding radiation exposure to the operator. Robots can be integrated with computed tomography (CT), cone beam CT, magnetic resonance imaging, and ultrasound and through various techniques, including stereotaxy, table-mounted, floor-mounted, and patient-mounted robots. The history, challenges, solutions, and questions facing the field of interventional radiology (IR) and interventional oncology are reviewed, to enable responsible clinical adoption and value definition via ergonomics, workflows, business models, and outcome data. IR-integrated robotics is ready for broader adoption. The robots are coming!

Comparison of Smartphone Augmented Reality, Smartglasses Augmented Reality, and 3D CBCT-guided Fluoroscopy Navigation for Percutaneous Needle Insertion: A Phantom Study.

PURPOSE: To compare needle placement performance using an augmented reality (AR) navigation platform implemented on smartphone or smartglasses devices to that of CBCT-guided fluoroscopy in a phantom. MATERIALS AND METHODS: An AR application was developed to display a planned percutaneous needle trajectory on the smartphone (iPhone7) and smartglasses (HoloLens1) devices in real time. Two AR-guided needle placement systems and CBCT-guided fluoroscopy with navigation software (XperGuide, Philips) were compared using an anthropomorphic phantom (CIRS, Norfolk, VA). Six interventional radiologists each performed 18 independent needle placements using smartphone (n = 6), smartglasses (n = 6), and XperGuide (n = 6) guidance. Placement error was defined as the distance from the needle tip to the target center. Placement time was recorded. For XperGuide, dose-area product (DAP, mGy*cm2) and fluoroscopy time (sec) were recorded. Statistical comparisons were made using a two-way repeated measures ANOVA. RESULTS: The placement error using the smartphone, smartglasses, or XperGuide was similar (3.98 ± 1.68 mm, 5.18 ± 3.84 mm, 4.13 ± 2.38 mm, respectively, p = 0.11). Compared to CBCT-guided fluoroscopy, the smartphone and smartglasses reduced placement time by 38% (p = 0.02) and 55% (p = 0.001), respectively. The DAP for insertion using XperGuide was 3086 ± 2920 mGy*cm2, and no intra-procedural radiation was required for augmented reality. CONCLUSIONS: Smartphone- and smartglasses-based augmented reality reduced needle placement time and radiation exposure while maintaining placement accuracy compared to a clinically validated needle navigation platform.

Imaging, Pathology, and Immune Correlates in the Woodchuck Hepatic Tumor Model.

BACKGROUND: Woodchucks chronically infected with woodchuck hepatitis virus (WHV), which resembles human hepatitis B virus, develop spontaneous hepatic tumors and may be an important biological and immunological model for human HCC. Nonetheless, this model requires further validation to fully realize its translational potential. METHODS: Woodchucks infected at birth with WHV that had developed HCC (n=12) were studied. Computed tomography, ultrasound, and magnetic resonance imaging were performed under anesthesia. LI-RADS scoring and correlative histologic analysis of sectioned tissues were performed. For immune characterization of tumors, CD3 (T cells), CD4 (T helpers), NCAM (Natural killers), FOXP3 (T-regulatory), PDL-1 (inhibitory checkpoint protein), and the human hepatocellular carcinoma (HCC) biomarker alpha-fetoprotein (AFP) immunohistochemical stains were performed. RESULTS: Forty tumors were identified on imaging of which 29 were confirmed to be HCC with 26 categorized as LR-4 or 5. The remainder of the tumors had benign histology including basophilic foci, adenoma, and lipidosis as well as pre-malignant dysplastic foci. LR-4 and LR-5 lesions showed high sensitivity (90%) and specificity (100%) for malignant and pre-malignant tumors. Natural killers count was found to be 2-5 times lower in tumors relative to normal parenchyma while other immune cells were located in the periphery of tumors. Tumors expressed AFP and did not express PD-L1. CONCLUSION: Woodchucks chronically infected with WHV developed diverse hepatic tumor types with diagnostic imaging, pathology, and immune patterns comparable to that in humans. This unique animal model may provide a valuable tool for translation and validation of novel image-guided and immune-therapeutic investigations.

Generalized chest CT and lab curves throughout the course of COVID-19.

A better understanding of temporal relationships between chest CT and labs may provide a reference for disease severity over the disease course. Generalized curves of lung opacity volume and density over time can be used as standardized references from well before symptoms develop to over a month after recovery, when residual lung opacities remain. 739 patients with COVID-19 underwent CT and RT-PCR in an outbreak setting between January 21st and April 12th, 2020. 29 of 739 patients had serial exams (121 CTs and 279 laboratory measurements) over 50 ± 16 days, with an average of 4.2 sequential CTs each. Sequential volumes of total lung, overall opacity and opacity subtypes (ground glass opacity [GGO] and consolidation) were extracted using deep learning and manual segmentation. Generalized temporal curves of CT and laboratory measurements were correlated. Lung opacities appeared 3.4 ± 2.2 days prior to symptom onset. Opacity peaked 1 day after symptom onset. GGO onset was earlier and resolved later than consolidation. Lactate dehydrogenase, and C-reactive protein peaked earlier than procalcitonin and leukopenia. The temporal relationships of quantitative CT features and clinical labs have distinctive patterns and peaks in relation to symptom onset, which may inform early clinical course in patients with mild COVID-19 pneumonia, or may shed light upon chronic lung effects or mechanisms of medical countermeasures in clinical trials.

Determination of disease severity in COVID-19 patients using deep learning in chest X-ray images.

PURPOSE: Chest X-ray plays a key role in diagnosis and management of COVID-19 patients and imaging features associated with clinical elements may assist with the development or validation of automated image analysis tools. We aimed to identify associations between clinical and radiographic features as well as to assess the feasibility of deep learning applied to chest X-rays in the setting of an acute COVID-19 outbreak. METHODS: A retrospective study of X-rays, clinical, and laboratory data was performed from 48 SARS-CoV-2 RT-PCR positive patients (age 60±17 years, 15 women) between February 22 and March 6, 2020 from a tertiary care hospital in Milan, Italy. Sixty-five chest X-rays were reviewed by two radiologists for alveolar and interstitial opacities and classified by severity on a scale from 0 to 3. Clinical factors (age, symptoms, comorbidities) were investigated for association with opacity severity and also with placement of central line or endotracheal tube. Deep learning models were then trained for two tasks: lung segmentation and opacity detection. Imaging characteristics were compared to clinical datapoints using the unpaired student's t-test or Mann-Whitney U test. Cohen's kappa analysis was used to evaluate the concordance of deep learning to conventional radiologist interpretation. RESULTS: Fifty-six percent of patients presented with alveolar opacities, 73% had interstitial opacities, and 23% had normal X-rays. The presence of alveolar or interstitial opacities was statistically correlated with age (P = 0.008) and comorbidities (P = 0.005). The extent of alveolar or interstitial opacities on baseline X-ray was significantly associated with the presence of endotracheal tube (P = 0.0008 and P = 0.049) or central line (P = 0.003 and P = 0.007). In comparison to human interpretation, the deep learning model achieved a kappa concordance of 0.51 for alveolar opacities and 0.71 for interstitial opacities. CONCLUSION: Chest X-ray analysis in an acute COVID-19 outbreak showed that the severity of opacities was associated with advanced age, comorbidities, as well as acuity of care. Artificial intelligence tools based upon deep learning of COVID-19 chest X-rays are feasible in the acute outbreak setting.