Quantitative 3D digital pathology image analysis for tissue characterization on prostate cancer imaging

点击翻译按钮获取中文摘要

Medical imaging technologies such as magnetic resonance imaging and computed tomography are important to our understanding of prostate cancer. Effective use these technologies requires knowledge of what the images tell us about disease severity. Men often undergo prostate removal surgery for cancer treatment; assessment of the removed tissue under a microscope gives reliable cancer severity information. Most of what we know about the relationship between medical image appearance and cancer disease status was learned by comparing pre-surgical medical images to the post-surgical microscopic assessment of disease status. However, there is a lack of proper tools to assist experts in quantitatively and repeatably making these assessments, and this has hindered our understanding of the most effective ways to use prostate medical imaging to help patients. This lack of tools makes it difficult for experts to consistently report on the precise locations, shapes, sizes, and appearances of cancerous tumours, so instead these attributes are reported in an inexact way or not at all. The main reason for the lack of necessary tools is that images of microscopic tissue samples are extremely large and until recently, computer hardware was not sufficient to store and process them. Experts have therefore been limited to making assessments by looking through microscopes at tissue mounted on glass slides - this makes tasks like outlining the detailed boundaries of tumours on the tissue impossible to perform accurately and quickly. Recently, this has changed, and digital pathology imaging systems now enable tissue assessments on the computer screen. This presents us with an opportunity to develop new software technologies to provide assistive tools that allow viewing of tissue samples not only in two dimensions, but also in three dimensions (3D), and allow the accurate determination of the 3D size, shape, appearance, and location of each tumour. The development of these software technologies is the theme of this proposal, which leverages the shift to the digital pathology paradigm to develop transformative technologies enabling a deeper understanding of the meaning and effective use of medical imaging for prostate cancer diagnosis and therapy guidance.