Development of 3D Pathology Techniques for Prostate Cancer

Advancements in imaging for the detection, staging, grading, and guidance of focal treatment of prostate cancer have produced several novel imaging approaches, such as multi-parametric, contrast-enhanced magnetic resonance imaging (MRI); computed tomography perfusion; microbubble contrast-enhanced 3D ultrasound; radiotracers for positron emission tomography imaging including 18F-choline; and MRI-based metabolic imaging using hyperpolarized 13C. These techniques and other new ones are being investigated in the OICR SIP-Prostate Project. Our understanding of the imaging physics underlying these modalities allows the calculation of parameters of potential clinical interest (e.g., hypervascularity, radiotracer uptake). Critically needed is a deeper understanding of how the 3D imaging data relate to disease status in the underlying tissue. For small specimens collected as part of animal studies, the gold standard measure of disease status can be based on an assessment of 2D images of tissue sections taken from a resected specimen with small inter-section spacing (e.g., 4–100 μm). For clinical specimens, the tissue samples obtained for assessment are usually sparser (e.g., one section every 3–5 mm from human prostate specimens), and the disease status is reflected in clinical pathology reports. However, traditional pathology reports on these specimens provide coarse 2D localization of lesions on histopathology slides. Furthermore, many parameters of clinical interest to oncologists treating patients are currently qualitatively and inconsistently assessed. For example, in prostate cancer, these parameters could include lesion volume, extent of extra-prostatic extension (EPE), and margin status. In both the small animal specimen and clinical specimen scenarios, lacking is a means for 3D histopathology reconstruction, visualization, and quantification. Such tools are of utmost importance to the scientific investigation into the biological and clinical meaning of data from in vivo imaging, and to the optimization of clinical pathologists’ workflow to render their assessments more quantitative and consistent, better informing follow-up treatment selection and guidance.

To address this issue, we propose to develop and test a computational tool set for quantitative digital pathology image analysis to calculate 3D pathology maps of prostate cancer in an accurate and repeatable manner. This proposal leverages the emergence of digital pathology imaging – a disruptive technology that is poised to push the discipline of pathology into a transition from the traditional approach of reading glass slides using microscopes, to the digital imaging paradigm, in which tissue sections are digitized at high resolution (e.g., 0.5 μm/pixel) using specialized scanning equipment. This transition is in its infancy, with the current state of the art limited primarily to slide scanning, storage, display, basic annotation, and transmission. The primary challenges in developing a quantitative pathology image analysis tool set lie within the areas of imaging science, biomedical engineering, and mathematics.