Computer aided diagnosis for radiology and histopathology images
Principal Investigator
Name
zhenhua xu
Degrees
M.D
Institution
Independent Researcher, not applicable
Position Title
Independent Researcher
Email
About this CDAS Project
Study
NLST
(Learn more about this study)
Project ID
NLST-293
Initial CDAS Request Approval
Mar 15, 2017
Title
Computer aided diagnosis for radiology and histopathology images
Summary
Radiology and histopathology images are very common and helpful in hospitals to diagnose the patients. Normally radiology can help to quickly identify whether a patient has cancer or not and histopathology can help to finally identify the grading (eg. 0-4) and stage of cancer and also as a golden-truth evidence. However, in some developing countries like china and india, there are few radiologists compared to large size of population. Radiologist have very heavy workload to check the radiology and histopathology images everyday. What's more, some work is really very repeatable and boring, this increase the chance of making error. So it's very urgent to call for a robust CAD system to help the radiologists and histopathologists improve their efficiency and accuracy.
We have developed an aided system to do some automatic segmentation and classification on a smaller dataset collected from some china hospitals, using state of art machine learning algorithms. The goal of this project is to improve the sensitivity, accuracy and generalization of current system and to build up a more advanced model using the large scale and multi-dimensional data (both radiology and pathology) from NLST.
We have developed an aided system to do some automatic segmentation and classification on a smaller dataset collected from some china hospitals, using state of art machine learning algorithms. The goal of this project is to improve the sensitivity, accuracy and generalization of current system and to build up a more advanced model using the large scale and multi-dimensional data (both radiology and pathology) from NLST.
Aims
1. Automatic outline the area of pulmonary modules and compute the area.
2. Detect and segment the nuclei and gland in the histopathological images.
Collaborators
None