2014 Summer Project Week:CAD Toolbox for Neurological Disorders

From NAMIC Wiki
Revision as of 03:28, 27 June 2014 by Sliu7418 (talk | contribs)
Jump to: navigation, search
Home < 2014 Summer Project Week:CAD Toolbox for Neurological Disorders


Key Investigators

  • USYD: Sidong Liu, Siqi Liu, Fan Zhang, Yang Song, Weidong (Tom) Cai
  • SPL-BWH: Sonia Pujol, Ron Kikinis


Project Description

We have accumulated a wealth of knowledge and experience in computer-aided-diagnosis (CAD) of neurological disorders, such as Alzheimer’s Disease (AD), from our previous studies [1-3]. In this project, we will develop a Slicer module that is capable of estimating the AD evolvement. Due to the computational cost of the pre-processing pipeline, the system will be decomposed into a client-server pattern. This system will also be adaptive to CAD applications for other neurological disorders in the future.

Objective

The goal of this project is to provide estimations of AD progression through a pre-trained deep-learning model to aid the decision making for the neurologists.

Approach, Plan

There are roughly three key steps of the pipeline: client interface, pre-processing pipeline and decision making module. Slicer 4.3 will be used as the client side to remotely access the CAD server. The client will be responsible of providing a user interface and some early stage preprocessing. The loaded imaging data (MRI in this project) will be transferred to the CAD server with necessary complementary information of the patient. At the server end, brain descriptors will be computed from the MRI scans processed with FreeSurfer pipeline, and then fed into a pre-trained deep neural network. The outputs of the network will be used to estimate the probabilities of the different AD stages, and the diagnostic results will be sent to the users via emails. Considering the high computational cost of the CAD pipeline, some of the concurrent requests will be queued.



Resources

  • 3D Slicer is a well-established medical image computing and visualization platform, and is also an ideal tool for image content analysis in personalized medicine.
  • FreeSurfer is a set of tools that contains a fully automatic structural imaging stream for processing cross sectional and longitudinal data.


Delivery Mechanism

This work will be delivered to the NA-MIC Kit as a (please select the appropriate options by noting YES against them below)

  1. ITK Module
  2. Slicer Module
    1. Built-in
    2. Extension -- commandline
    3. Extension -- scripted ("YES")
  3. Other (Please specify)


Progress

  1. - CAD server setup (done)
  2. - Machine learning module (done)
  3. Pre-processing pipeline (done)
  4. - Database (done)
  5. - Module interface (done)
  1. - Policy (in development)
  2. - Client (in development)
  3. - Security improvement (in development)

Acknowledgements

This work is supported in part by the ARC, AADRF, NA-MIC (U54EB005149), and NAC (P41EB015902).


References

[1] Siqi Liu, Sidong Liu, Weidong Cai, Sonia Pujol, Ron Kikinis, Dagan Feng, “Early Diagnosis of Alzheimer's Disease with Deep Learning”, ISBI 2014, pp1015-1018, 2014.
[2] Hangyu Che, Sidong Liu, Weidong Cai, Sonia Pujol, Ron Kikinis, Dagan Feng, “Co-neighbor Multi-view Spectral Embedding for Medical content-based Retrieval”, ISBI 2014, pp911-914, 2014.
[3] Sidong Liu, Weidong Cai, Lingfeng Wen, David Feng, Sonia Pujol, Ron Kikinis, Michael Fulham, Stefan Eberl, “Multi-Channel Neurodegenerative Pattern Analysis and Its Application in Alzheimer’s Disease Characterization”, Computerized Medical Imaging and Graphics, 2014.