QUANTITATIVE 3D ANATOMIC CHANGE DETECTION

Customer:
NIH: National Institute of Mental Health

Goal of Program:
We are developing quantitative image analysis software tools for detecting and precisely measuring anatomical changes in sequences of 3D medical image data, such as MRI or CT scans taken over time. Accurate structural change detection and identification will improve several medical applications, such as the experimental evaluation of drugs and treatments, precise monitoring of disease progression, and early disease diagnosis. The key objective of this research is to demonstrate that practical automated registration tools can be developed for handling the problems encountered in medical change detection applications: high accuracy and reproducibility requirements, structures of complex 3D shape with possible tissue deformation, varying image resolutions and fields-of-view, and sensor distortion.

Key Technology:
Automated 3D volume and surface registration techniques. Accurate image segmentation algorithm

Key Product:
The planned result of this program is an end-to-end system that precisely quantifies anatomical pathologies by detecting changes in sequential 3D images. In many diagnostic and patient monitoring applications the key events to detect are changes to anatomical structures of interest. Early indications of possible pathologies from single scans are often difficult to interpret since it may not be known what is ìnormalî for a particular patient. The identification of changes from multiple time points, on the other hand, pinpoints trends and improves detection sensitivity of the pathologies. Such early and accurate identification of anatomical changes often leads to reduced cost of treatment and improved patient care. Such change identification systems have the potential to positively impact a wide array of medical application areas, including: Disease diagnosis and progression studies, such as tracking multiple sclerosis lesion behavior and stroke damage; drug efficacy trials; and treatment monitoring.