design

Master Server

• Purpose: To handle the entire pipeline and communication between other classes in the system. Also manages disk I/O for now.
• Not persistent. Created during system initialization based on user input or config files.
• Class name: NeuralNexus

Attributes

• All other objects in the processing pipeline
• Data store: includes frames, logs

Methods

• For data store:
  • _startStore: start running plasma server in subprocess
  • _closeStore: kill subprocess running server
• Load objects: Based on user selection for acquisition, processing, etc
• Run: Multiprocessing of simultaneous/asynchronous running objects
• Update visualization

Test cases

• Unit and system integration tests

Front Panel

• Purpose: To handle all visualization and user input. Communicates with the server.

Attributes

• UI front panel
• ConfigParam

Methods

• Update handler (UI)
• Plot current frame
• Plot components
• Plot activity

Test cases

Data Store

• Purpose: Backend data store. Handled by server.
• Current implementation in Pyarrow Plasma. Class name Limbo.
• Can be implemented with different store types.

Attributes

• PlasmaClient to communicate with store
• dict to hold internal correspondence between simple object names and their IDs -- limited use

Methods

• connectStore: create a client to connect to store
• get: get items from store
• put: put items into store

Configuration Parameters

• Purpose: To contain any and all configuration parameters for the experiment.
• Logs (via the server) initial and changed states
• Class also wraps CNMFParams

Attributes

• CNMFParams
• List of objects selected for the experiement
• Timestamp

Methods

• Update: configuration change
• Save: to server to write to disk as log

Test cases

Image Acquisition

• Purpose: Acquire an image from a source to hand off to the server.
• Implements some kind of buffer to ensure no lost frames.
• Class wraps imagemp functionality
• Abstracted to allow multiple classes depending on how images/frames are acquired.

Attributes

• Buffer: holding certain number of frames
• Current frame

Methods

• Acquire_image: get next frame
• Save: hand off to server

Test cases

Image Processing

• Purpose: Given an image (from file or shared structure), produce neural activity information.
• Should include motion stabilization, neuron-finding, and spike identification.
• Class wraps CaImAn functionality (OnACID, CNMF)
• Is also abstracted

Attributes

• Current frame: frame being analyzed

Methods

• Get_next_frame: from server
• Process_frame: analyze current frame
• Save: hand off to server

Test cases

Activity

• Purpose: To contain the object describing neural activity
• Class name: NeuralExciton
• Class wraps CaImAn objects (Estimates)

Attributes

• Estimates

Methods

• Normalize components
• Filter components: take a subset, threshold, etc
• Compare components

Test cases

Inference

• Purpose: Given some neural activity, produce an online estimate of the effects of each neuron's response.
• Class name: Neurlock

Attributes

• Chosen method for analysis

Methods

• Evaluate components
• Compute residuals

Test cases

Stimulation

• Purpose: To send a stimulus to the subject under study.
• Class: not yet implemented