web analytics

A Neural Network to Quantify in Situ Adaptive Immunity in Autoimmune Disorders

Interests: Imaging
Published:
Lead Inventor: Marcus Clark
Print or Save to PDF

SUMMARY

  • Presently, the use of single cell methods such as RNA sequencing can identify rare inflammatory cellular populations that can be used for the diagnosis of autoimmune disorders. However, the scope of current technologies is limited in that they do not convey information on the architecture of the cellular inflammation and the organization of inflammatory cellular networks.
  • The inventors developed a deep convolutional neural network that can analyze stained confocal microscopy images to determine immune cell proximity and shape, which are in turn revealing of a patient’s autoimmune pathophysiology. The neural network can be used to identify antigen specific interactions between T-cells and antigen presenting cells in high throughput. 
  • The invention is a digital pathology software package that allows for the automated analysis of stained biopsy samples to reveal the nature of T cell: dendritic cell interactions. Identification of maladaptive interactions between these cell types can be used for diagnosing autoimmune disorders.
  • A proof-of-concept study showed that the invention can classify T cell: dendritic cell interactions with a comparable accuracy to two-photon excitation microscopy. 

 

FIGURE

T cell: dendritic cell relative distance for both wild type mouse T cell samples (black line) and antigen specific mouse T cell samples (grey line). T cells were pulsed with antigen loaded dendritic cells and distances were computed using the deep convolutional neural network.  

ADVANTAGES

ADVANTAGES

  • Higher throughput than two-photon excitation microscopy
  • More informative than current single cell RNA-seq methods
  • Compatible with human samples  

APPLICATIONS

  • Autoimmune disorder diagnostic (lupus nephritis)
  • Cancer immunotherapy companion diagnostic
  • Research tools

 

PUBLICATIONS

 

 

 

 

 

 

 

 

 

 

 

  • PCT/US18/29426