Ventilation-perfusion (V/Q) scintigraphy has a major role to play in the diagnosis of pulmonary embolism (PE). Objective criteria exist for diagnosing PE on both V/Q planar and SPECT; however, reporting physicians ultimately incorporate their own subjective judgement into a final diagnosis. Therefore, this imaging modality is a promising candidate for standardizing and automating image interpretation with artificial intelligence (AI). Early studies from the 1990s and early 2000s with this aim report promising results but now rely on outdated machine learning techniques. Since then, there has been little work in this domain with recent investigations shifting focus to computed tomography pulmonary angiography. There is therefore a huge potential to resurrect and modernize this field with state-of-the-art deep learning approaches. The ultimate objective of this research project is to increase the value of V/Q scans as a nuclear medicine procedure and promote its use by means of various supervised and unsupervised AI solutions for automatic diagnosis and improved workflow with convolutional neural networks. Such solutions should benefit nuclear medicine physicians and their patients, both as a diagnostic aid for the experienced physicians as well as a training tool for physicians in training.