Detection of Calcified Granulomas

Purpose

Detect calcified granulomas in lung parenchyma and abdominal solid organs as well as calcified lymph nodes.

Tag(s)

 

Panel

Thoracic Panel

Define-AI ID

19080008

Originator

External Use Case Submitter, University of British Columbia
Lead External Use Case Submitter, University of British Columbia

Panel Chair

Eric Stern

Panel Reviewers

Thoracic Panel

License

Creative Commons 4.0
Status Public Commenting
                               

Clinical Implementation


Value Proposition

An algorithm that could detect and characterize granulomas would help clinicians recommend further testing, if necessary, and improve patient care. This could help obviate additional imaging if an algorithm could help concretely determine a nodule is a benign calcification.

There are multiple ways to effectively utilize this algorithm for improvements in patient care. For instance, the algorithm could assess prior imaging to determine which patients have latent TB infection/prior TB exposure. This could help groups involved with immigration screening. The algorithm may also prove useful in settings where patients may require future immunomodulatory drugs such as chemo, monoclonal antibodies or HAART and are vulnerable to reactivation.

Narrative(s)

Patient present with lung nodule. Algorithm suggests the nodule is calcified granuloma as a result from a prior infection of TB. Radiologist confirms this interpretation from the pattern of the calcification and clinical background.

Workflow Description

The image is obtained from modality and sent to PACS and the AI engine. The image is analyzed by the engine. The system then detects calcified granulomas. An alert message is sent to PACS from the engine with the information, identification, and graphic highlighting the granulomas.

Considerations for Dataset Development



Procedures(s)

Chest, CT

View(s)


Sex at Birth

{Male,Female}

Age

[0,90]

BMI

varied

Diabetes

Present, Absent

Hypertension

Present, Absent

Smoker

Yes, No

Technical Specifications


 

Inputs

DICOM Study

Procedure

Chest CT

Views

Coronal, sagittal, axial

Data Type

DICOM

Modality

CT

Body Region

Chest

Anatomic Focus

Lung


Primary Outputs

Granulomas in lung parenchyma

RadElement ID

to be defined

Definition

presence of granuloma in lung parenchyma

Data Type

Categorical

Value Set

0- Granuloma absent in lung parenchyma

1- Granuloma present in lung parenchyma

2- Unknown

Units

N/A


Granuloma in lymph nodes

RadElement ID

to be defined

Definition

presence of granuloma in lymph node

Data Type

Categorical

Value Set

0- Granuloma absent in lymph nodes

1- Granuloma present in lymph nodes

Units

N/A


Secondary Outputs

Prior TB exposure

RadElement ID


Definition

granuloma consistent with prior TB exposure

Data Type

Categorical

Value Set

0- No granuloma consistent with TB exposure

1- Present granuloma, consistent with TB exposure

2- Unknown

Units

N/A



Granuloma segmentation

RadElement ID


Definition

segment the detect granuloma(s)

Data Type

Coordinate

Value Set

N/A

Units

N/A



Count of granulomas

RadElement ID


Definition

number of granulomas present

Data Type

Numeric

Value Set

N/A

Units

granulomas