Atypical pulmonary carcinoids (PCa) are believed to progress from the less aggressive typical carcinoids. In contrast, it has been widely accepted that PCa have unique clinico-histopathological traits with no causative relationship or genetic, epidemiologic, or clinical traits in common with the aggressive lung neuroendocrine cancers (NECs), large-cell neuroendocrine carcinomas (LCNECs), and small-cell lung cancers. However, recent evidence in the literature has suggested that the molecular link between PCa and NECs might be subtler than initially thought. Supporting this hypothesis, we have recently identified a group of PCa (named supra-carcinoids) that exhibit genuine carcinoid-like morphology but molecular and clinical features of LCNEC. The objective of this thesis project is the analysis of histopathological images of carcinoids of the lung by computer vision techniques, to facilitate diagnosis and define the histopathological characteristics of supra-carcinoids. The overall strategy to achieve this objective is as follows: (i) development of an annotated carcinoid image database; (ii) implementation of a convolutional neural network (CNN) to classify and extract image features; and (iii) extraction of computational histopathological features to morphologically and genomically interpret the matrices output from the CNN. These results may reveal relevant attributes to facilitate the diagnosis and characterization of supracarcinoids, which will ultimately improve the management of patients with the most severe forms of carcinoids.