Evaluation of clinical images is essential for diagnosis in many specialties. Therefore the development of computer vision algorithms to help analyze biomedical images will be important. In ophthalmology, optical coherence tomography (OCT) is critical for managing retinal conditions. We developed a convolutional neural network (CNN) that detects intraretinal fluid (IRF) on OCT in a manner indistinguishable from clinicians. Using 1,289 OCT images, the CNN segmented images with a 0.911 cross-validated Dice coefficient, compared with segmentations by experts. Additionally, the agreement between experts and between experts and CNN were similar. Our results reveal that CNN can be trained to perform automated segmentations of clinically relevant image features.

Despite advances in artificial intelligence (AI), its application in medical imaging has been burdened and limited by expert-generated labels. We used images from optical coherence tomography angiography (OCTA), a relatively new imaging modality that measures retinal blood flow, to train an AI algorithm to generate flow maps from standard optical coherence tomography (OCT) images, exceeding the ability and bypassing the need for expert labeling. Deep learning was able to infer flow from single structural OCT images with similar fidelity to OCTA and significantly better than expert clinicians (P < 0.00001). Our model allows generating flow maps from large volumes of previously collected OCT data in existing clinical trials and clinical practice. This finding demonstrates a novel application of AI to medical imaging, whereby subtle regularities between different modalities are used to image the same body part and AI is used to generate detailed inferences of tissue function from structure imaging.

ObjectiveTo identify trends in patient presentation and outcomes data that may guide the development of clinical algorithms on Merkel Cell Carcinoma (MCC).MethodsWe performed a retrospective cohort study searching in the National Cancer Institute's SEER registry for documented MCC cases from 1986-2013. No exclusion criteria were applied. We hereby identified 7,831 original MCC entries. Demographics, staging, and socioeconomic characteristics were identified and treatment modality likelihoods and survival data were calculated via logistic regression and Kaplan-Meier statistical modeling.ResultsConcerning tumor localization, 44.5% (n= 3,485) were located on the head and neck, and 47.8% were located on the trunk and extremities (n= 3,742). Male and younger patients are more likely to receive radiation than surgery with no differences seen among patient race. Caucasians and “Other” races both showed higher overall survival than African American patients. States with higher median household income levels demonstrated survival advantage. Income quartiles yielded no differences in surgical or radiotherapy interventions. Moreover, patients who forego radiotherapy had a poorer overall survival.LimitationsGeneralizability of SEER data, potential intrinsic coding inconsistencies, and limited information on patient comorbidities, sentinel lymph node and surgical margin status are major limitations. There is no information regarding medical intervention such as systemic chemotherapy or immunotherapy. Recoding efforts are inconclusive regarding variables such as tumor infiltrating lymphocytes, mutations, or immunosuppression status, which are well-documented for other cancers within the database.ConclusionMCC lesions of the head and neck region, lower income quartiles, and African American race are associated with higher mortality. MCC patients have a median household income that is significantly higher than national values with no significant difference in subsequent treatment modalities (surgery or radiotherapy) based on socioeconomic markers. A lack of radiotherapy is associated with higher mortality.

No abstract