Radiology reports contain numeric values which represent important quantitative measurements, such as lesion size, that provide key information for diagnosis and treatment. Automated extraction of these numeric values from radiology reports would enable structured storage and analysis of this information to support applications like computer-aided diagnosis, personalized treatment planning, and population health management. Previous research has utilized rule-based systems with regular expressions and machine learning models like conditional random fields to identify and classify measurements in radiology reports. These techniques have achieved high performance, with precision and recall over 0.99, as well as classification accuracy over 0.95. However, building high-performing systems requires substantial manual engineering and annotation, limiting scalability. More recent work has applied neural networks for measurement extraction from radiology reports. Convolutional neural networks have been used to detect spans of text containing measurements, while recurrent neural networks with long short-term memory cells have been used for measurement classification. These neural approaches require less manual feature engineering and can leverage unlabeled data for pretraining, enabling more scalable measurement extraction systems. Measurement extraction from radiology reports enables structured analysis of quantitative medical data that would otherwise be locked in unstructured text. By automatically extracting measurements like lesion size, these techniques can support applications that improve radiology reporting, computer-aided diagnosis, personalized medicine, and healthcare analytics. Although initial research has shown promise, continued progress in this area has the potential to unlock large amounts of quantitative data and enable a new generation of data-driven medical tools.