Deep-learning based detection of eosinophilic esophagitis

Background and aims: For eosinophilic esophagitis (EoE) a substantial diagnostic delay is still a clinically relevant phenomenon. Deep learning-based algorithms have demonstrated potential in medical image analysis. Here we establish a convolutional neuronal network (CNN)-based approach that can distinguish EoE from normal findings and candida esophagitis. Methods: We trained and tested a CNN using 484 real-world endoscopic images from 134 subjects consisting of three classes (normal, EoE, and candidiasis). Images were split into two completely independent datasets. The proposed approach was evaluated against three trainee endoscopists on the test set. Model-explainability was enhanced by deep Taylor decomposition. Results: Global accuracy (0.915 [0.880-0.940]), sensitivity (0.871 [0.819-0.910]) and specificity (0.936 [0.910-0.955]) were significantly higher than for endoscopists on the test set. Global area under the ROC curve was 0.966 [0.954-0.975]. Results were highly reproducible. Explainability analysis found that the algorithm identified characteristic signs also used by endoscopists. Conclusions: Complex endoscopic classification tasks including more than two classes can be solved by CNN-based algorithms. Thus, our algorithm (https://ccb-test.cs.uni-saarland.de/EoE/) may assist clinicians in making the diagnosis of EoE.

1099. Opportunistic Infections Among Long Term Survivors of Kidney Transplantation: Defining Risk Factors

Background Opportunistic infections (OIs) in kidney transplant recipients (KTR) most commonly occur in the early post-transplant period or with increased immunosuppression, largely as a result of impaired T-cell function. Additionally, age confers susceptibility to infection independent of time post-transplant. The combined impact of cumulative immunosuppression and immunosenescence on infection risk of long-term KT survivors has not been well described. Methods We performed a retrospective chart review of patients age ≥ 18 years who underwent KT between 2003 to 2009 and who survived ≥ 10 years post-KT, in order to evaluate the risk factors for OIs. Demographics, comorbidities, immunosuppression, and clinical data for OIs occurring ≥ 10 years of KT were collected. AST ID Working Group on Infectious Disease Monitoring definitions for OIs was used. Risk factors for OIs were assessed by simple logistic regression. Results Of 332 KTR, 16 (4.8%) had an OI with 18 total episodes. Of 16 KTR, half were white, 10 (62.5%) were male, median age at time of transplant was 43 (range 25-72) and the median post-transplant follow-up was 14.2 years (range 10.3-37.6). The mean Charlson Comorbidity Index (CCI) at diagnosis was 5.6 (S.D. 3.6). Ten patients (62.5%) were on mycophenolate-based regimens. The mean absolute lymphocyte count (ALC) at the time of OI was 0.78 x 103/µL (S.D. 0.43). Two (12.5%) had acute rejection within 1 year of OI. Of 18 OI episodes, there were 6 PJP, 2 candida esophagitis, 3 CMV (2 viremia, 1 colitis), 2 cryptococcal infections (1 meningitis, 1 myositis/disseminated), 2 adenovirus (pneumonia, colitis), 2 VZV (herpes zoster) and 1 HSV (esophagitis). Two patients had 2 concurrent OIs (1 had PJP and cryptococcus and 1 had HSV and candida esophagitis). Three died within 30-days of OI diagnosis. OI incidence was associated with years from date of transplant [OR 1.3, p=0.002], cerebrovascular disease [OR 4.45, p=0.02], and lower ALC [OR 5.9, p < 0.05]. CCI also trended towards association [OR 1.24, p=0.09]. Table 1: Demographics, comorbidities, immunosuppression, and clinical data for patients with OIs Table 2: Detailed characteristics of each patient with opportunistic infections Conclusion OIs were infrequently observed beyond 10 years of transplant among long-term survivors of KT. However, OI incidence was associated with poor outcome. Low ALC and a higher burden of comorbidities were risk factors for very late occurrence of OIs in this population. Disclosures All Authors: No reported disclosures