Workflow

1
Workflow

• Stephen Aylward
• Luis Ibanez

2
Goals

• Identify 3 main challenges in this area
• Identify 3 specific problems that can be solved
  by a collaborative effort between academic and
  industry partners
• Identify a problem that the NCIGT can help
  address in the next year

3
Outline

• Scope
• Identify current projects
• How can NCIGT help
• Topics in workflow

4
Wikipedia Workflow

• Workflow at its simplest is the movement of
  documents and/or tasks through a work process.
• More specifically, workflow is the operational
  aspect of a work procedure how tasks are
  structured, who performs them, what their
  relative order is, how they are synchronized, how
  information flows to support the tasks and how
  tasks are being tracked.
• As the dimension of time is considered in
  Workflow, Workflow considers "throughput" as a
  distinct measure.
• Workflow problems can be modeled and analyzed
  using graph-based formalisms like Petri nets.

5
Workflow System

• 1. The workflow modeling component (sometimes
  called specification module, design environment
  or build time system) which enables
  administrators and analysts to define process and
  activities, analyze and simulate them, and assign
  them to people.
• 2. The workflow execution component, sometimes
  called the run-time system which most often
  consists of an
• execution interface seen by end-users and
• a workflow engine. The workflow engine is an
  execution environment which assists or performs
  the coordination of processes and activities.

6
Assumption

• A small number of workflow templates can capture
  the majority clinical cases Paggetti 2001
• Workflows are effectively modeled as State
  Diagrams, Petri Nets and Timed Petri Nets
• Each state has associated with it a finite set of
  events that invoke transitions to new states
• They provide an easy transition towards software
  implementation
• Workflows are the reality of the way humans
  perform activities
• Acknowledging its existence, when designing
  software, results in more realistic and safer
  software

7
Why

• Improved application design
• Simplified application design
• Produce safer software, better suited for
  clinical applications
• Basis for testing inputs/outputs/states
• Encapsulate technologies for research and
  integration
• Define a language that spans implementation
  toolkit
• Bridge IGSTK, Slicer, VolView,

8
Workflow granularity

• Effect-level (patient) workflow Guide probe to
  tumor
• Task-level (physician) workflow Acquire an
  ultrasound image of tumor
• Data-level (system) workflow Record tracker
  coordinates

9
Outline

• Scope
• Identify current projects
• How can NCIGT help
• Topics in workflow

10
What are the current/pending workflow-related
projects?

• Goals
• Methods
• Strengths / weaknesses
• Schedule

11
Outline

• Scope
• Identify current projects
• How can NCIGT help
• Topics in workflow

12
Per-topic questions

• Where is the field?
• Where can research have the most impact?
• Where is industry / academic partnering needed /
  beneficial?
• What is the timeline for the research(1 yr, 3
  yr, 5 yr)?

13
Topic 1

• What do we want from a workflow specification
• Non-programmatic description of a system /
  requirements
• Our system does
• Storyboard
• DICOM Workgroup
• Foundation for testing
• File format for creating applications
• An API standard for states/transitions for
  method/device sharing
• e.g., wrap IGSTK, Slicer
• GUI independentexecution model

• Where is the field?
• Where can research have the most impact?
• Industry / academic partnering needed /
  beneficial?
• What is the timeline for the research (1, 3, 5
  yr)?

14
Topic 2

• Users of workflow descriptions
• Application development
• For FDA approval
• In the clinic
• In research labs
• Testing
• New applications
• Existing applications
• Methods / Devices
• Research
• Developing methods / devices
• Developing new toolkits

• Where is the field?
• Where can research have the most impact?
• Industry / academic partnering needed /
  beneficial?
• What is the timeline for the research (1, 3, 5
  yr)?

15
Topic 3

• Role of patient safety / flexibility
• Solution do one thing well!!
• Error conditions enumerated
• Complexity vs transparency
• Determinism
• Guarantee each event is handled and transition
  occurs successfully
• Certain things should never happen explicit
  workflow make possible to forbid some paths
• Patient safety
• Vs ease-of-use (new applications)
• Vs ease-of-extension (new states)
• Vs ease-of-specialization (new components/devices)
  
• Vs ease-of-development

• Where is the field?
• Where can research have the most impact?
• Industry / academic partnering needed /
  beneficial?
• What is the timeline for the research (1, 3, 5
  yr)?

16
Topic 4

• Technical challenges
• Recording events in OR Lemke/Cleary
• Short lasting, repetitive, rapid,
  multi-parameter, simultaneous
• Deriving common states from multiple recordings
• Deriving common states from different procedures
• Quantifying physicians time and mental effort
• Workflow validation
• System testing using workflows
• Accruing error / uncertainty
• Accruing required accuracy
• Accruing risk to the patient / outcome
• Workflows vs C Not traditional programming
  style

• Where is the field?
• Where can research have the most impact?
• Industry / academic partnering needed /
  beneficial?
• What is the timeline for the research (1, 3, 5
  yr)?

17
Meeting Notes
18
Challenge 1

• Reference workflows
• Include clinicians in the process
• Use workflow for communication with clinicians
• Storyboard
• Recording workflow improved clinician performance
• Generating them
• Collection tools under development
• Data from Stealthlink directly
• Data from interviews
• Identify common workflows
• Choose a standard language
• Workflow management system
• Basis for teaching using new technology

19
Challenge 2

• Adapting workflows
• Specialize for site, physician, physician
  experience, patient
• Workflows should be very specific to balance
  flexibility with patient safety
• Limits creativity/problems in the OR?
• Dynamic programming?

20
Challenge 3

• Workflow for validation and comparison
• Workflows carry accuracy and time (outcome)
  predictions
• Workflows validated in the OR
• Workflows as a foundation for comparison of
  systems
• Compare workflows
• Time and accuracy

21
Challenge 4

• Implementing workflows
• Execution model is GUI independent
• Bridge IGTK, Slicer, and VolView
• C/QT/FLTK are ineffective for workflow-based
  programming
• Burden on developer
• Different models
• Lemke
• Service oriented architecture

22
Challenge 5

• Data Workflows
• No effective solution to process a large number
  of datasets
• Loni
• SIMULINK
• Kepler / Ptolomy
• Batchmake