Clinical impact of a deep learning system for automated detection of missed pulmonary nodules on routine body computed tomography including the chest region.

OBJECTIVES: To evaluate the clinical impact of a deep learning system (DLS) for automated detection of pulmonary nodules on computed tomography (CT) images as a second reader. METHODS: This single-centre retrospective study screened 21,150 consecutive body CT studies from September 2018 to February 2019. Pulmonary nodules detected by the DLS on axial CT images but not mentioned in initial radiology reports were flagged. Flagged images were scored by four board-certificated radiologists each with at least 5 years of experience. Nodules with scores of 2 (understandable miss) or 3 (should not be missed) were then categorised as unlikely to be clinically significant (2a or 3a) or likely to be clinically significant (2b or 3b) according to the 2017 Fleischner guidelines for pulmonary nodules. The miss rate was defined as the total number of studies receiving scores of 2 or 3 divided by total screened studies. RESULTS: Among 172 nodules flagged by the DLS, 60 (35%) missed nodules were confirmed by the radiologists. The nodules were further categorised as 2a, 2b, 3a, and 3b in 24, 14, 10, and 12 studies, respectively, with an overall positive predictive value of 35%. Missed pulmonary nodules were identified in 0.3% of all CT images, and one-third of these lesions were considered clinically significant. CONCLUSIONS: Use of DLS-assisted automated detection as a second reader can identify missed pulmonary nodules, some of which may be clinically significant. CLINICAL RELEVANCE/APPLICATION: Use of DLS to help radiologists detect pulmonary lesions may improve patient care. KEY POINTS: • DLS-assisted automated detection as a second reader is feasible in a large consecutive cohort. • Performance of combined radiologists and DLS was better than DLS or radiologists alone. • Pulmonary nodules were missed more frequently in abdomino-pelvis CT than the thoracic CT.

Is AI the Ultimate QA?

We are among the many that believe that artificial intelligence will not replace practitioners and is most valuable as an adjunct in diagnostic radiology. We suggest a different approach to utilizing the technology, which may help even radiologists who may be averse to adopting AI. A novel method of leveraging AI combines computer vision and natural language processing to ambiently function in the background, monitoring for critical care gaps. This AI Quality workflow uses a visual classifier to predict the likelihood of a finding of interest, such as a lung nodule, and then leverages natural language processing to review a radiologist's report, identifying discrepancies between imaging and documentation. Comparing artificial intelligence predictions with natural language processing report extractions with artificial intelligence in the background of computer-aided detection decisions may offer numerous potential benefits, including streamlined workflow, improved detection quality, an alternative approach to thinking of AI, and possibly even indemnity against malpractice. Here we consider early indications of the potential of artificial intelligence as the ultimate quality assurance for radiologists.

Rehospitalization and emergency department use rates before and after vagus nerve stimulation for epilepsy: use of state databases to provide longitudinal data across multiple clinical settings.

OBJECTIVES: Data regarding rehospitalization and emergency department (ED) visits following vagus nerve stimulation (VNS) present data analysis challenges. We present a method that uses California's multiple databases to more completely assay VNS efficacy. MATERIALS AND METHODS: The Healthcare Cost and Utilization Project's California Inpatient and Ambulatory Surgery databases were assayed for all VNS surgeries from 2005 to 2009. Patients were selected by epilepsy diagnosis codes and VNS procedure codes. Patients (total N = 629) were tracked across multiple databases using unique identifiers. Thirty-day and one-year post-implantation rates of VNS complication and healthcare visits were abstracted, along with one-year preoperative hospital and ED use. Statistics included correction for multiple comparisons. RESULTS: The one-year reoperation rate for adult patients (N = 536) was 3.9%; during the second year, an additional 3.2% of patients had reoperations. Within the first 30 days, <2% of patients experienced a complication. Four percent of patients were readmitted to a hospital, and 11.6% of patients visited an ED. The most common reason for rehospitalization or ED visit was seizure. In the first year after VNS, total seizure-related visits (hospitalization and ED) were 17% lower (2.12 visits per year to 1.71; p = 0.03). In the second year following VNS, seizure-related visits were 42% lower (2.21 visits per year to 1.27, p = 0.01). Pediatric patients (N = 93) had comparable results. CONCLUSIONS: VNS surgery has low rates of complications and reoperations and is associated with reduced incidence of seizure-related ED visits and hospital admissions in the first and second postoperative years.

Ethnic differences in viral dominance patterns in patients with hepatitis B virus and hepatitis C virus dual infection.

UNLABELLED: Studies of hepatitis B virus (HBV)/hepatitis C virus (HCV) dual infection are limited. Most are small, conducted outside the United States, and compare dual infection with HCV monoinfection. The goal of this study was to characterize HBV/HCV dual infection in a large multiethnic, matched, case-control study of dual-infected and HBV-monoinfected patients at two United States centers. Using an International Classification of Disease Version 9 electronic query and chart review, we identified 115 HBV/HCV dual-infected patients with serial HBV DNA, HCV RNA, and alanine aminotransferase (ALT) levels. As a control, 115 HBV-monoinfected patients were chosen randomly and matched with cases by age ±10 years, sex, Asian versus non-Asian ethnicity, and study site. Both groups had similar sex, ethnic, and age distributions (68% male, 83% Asian, age 52 ± 14 years). The median follow-up times were 33 and 38 months for the dual-infected and monoinfected groups, respectively. More monoinfected patients received HBV antiviral therapy than dual-infected patients (43% versus 24%; P = 0.002). No significant difference was detected between the proportion of monoinfected versus dual-infected patients with ALT above 40 U/L at presentation or during follow-up. Dual infection patients exhibited very little HBV/HCV codominance at baseline and throughout follow-up: patients had either HBV viremia with low or absent HCV RNA or detectable HCV RNA with low or absent HBV DNA. Asian ethnicity was predictive of HBV dominance after adjusting for sex, age, and baseline ALT elevation (odds ratio 7.35; P = 0.01). CONCLUSION: HBV/HCV dual-infected and HBV-monoinfected patients had similar clinical characteristics. Asian ethnicity is a major independent predictor of HBV-dominant disease, and HCV dominance with undetectable HBV DNA is more common in non-Asian individuals. Larger studies are needed to further characterize the natural history of HBV/HCV dual infection in Asian and non-Asian individuals.

Is routine histological examination of mastectomy scars justified? An analysis of 619 scars.

BACKGROUND: The increasing incidence of breast cancer is paralleled by an increasing demand for post-mastectomy breast reconstruction. At the time of breast reconstruction routine submission of mastectomy scars has been considered appropriate clinical practice to ensure that no residual cancer exists. However, this practice has been challenged by some and has become the topic of controversy. In a retrospective analysis we wished to assess whether routine submission of mastectomy scars altered treatment. METHODS: Utilizing the Stanford Translational Research Integrated Database Environment (STRIDE) all patients who underwent implant-based breast reconstruction with routine histological analysis of mastectomy scars were identified. The following parameters were retrieved and analyzed: age, cancer histology, cancer stage (according to the American Joint Committee on Cancer staging system), receptor status (estrogen receptor [ER], progesterone receptor [PR], Her2neu), time interval between mastectomy and reconstruction, and scar histology. RESULTS: A total of 442 patients with a mean age of 45.9 years (range, 22-73 years) were included in the study. Mastectomy with subsequent reconstruction was performed for in-situ disease and invasive cancer in 83 and 359 patients, respectively. A total of 619 clinically unremarkable mastectomy scars were sent for histological analysis, with the most common finding being unremarkable scar tissue (i.e. collagen fibers). Of note, no specimen revealed the presence of carcinoma. CONCLUSION: According to published reports routine histological examination of mastectomy scars may detect early local recurrence. However, we were not able to detect this benefit in our patient population. As such, particularly in the current health-care climate the cost-effectiveness of this practice deserves further attention. A more selective use of histological analysis of mastectomy scars in patients with tumors that display poor prognostic indicators may be a more reasonable utilization of resources.