UCLA researchers in the Department of Radiology have developed an algorithm for automated processing of medical images.
The increasing use of medical imaging techniques such as magnetic resonance imaging (MRI), computed tomography (CT), and positron emission tomography (PET) has generated a lot of medical data with nearly 20-40% increase every year. Currently, these medical images have to be processed manually by physicians and clinicians. There is a need for algorithms that can automatically read the patient images and classify them by type. Such approaches can be particularly useful for data mining and big data processing. However, previous attempts at developing such a system have utilized text-based algorithms that often fail between images with different classification systems.
UCLA researchers have developed an algorithm that can accurately identify different medical images. They have used their algorithm to successfully differentiate between CT images of brain, chest, and lungs. Their approach directly identifies the anatomical features of the image and therefore, does not require any classification code. It provides an important pre-processing classification step for data mining. The software can process any image and accurately classify it as a specific image category.
Algorithm developed and successfully tested for different CT scans.