Bony Metastatic Disease Detection

Purpose

Detect changes in the bones during cancer staging.

Tag(s)

 

Panel

Oncology

Define-AI ID

19070002

Originator

Christopher P. Gange
Lead Christopher P. Gange

Panel Chair

Reid Thompson

Panel Reviewers

Oncology Panel

License

Creative Commons 4.0
Status Public Commenting
                               

Clinical Implementation


Value Proposition

Accurate cancer staging is essential for decisions regarding timing and use of cancer treatments and for assessing effectiveness in clinical trials. Bones are important site of metastatic disease but are not focus of many types of scans. An algorithm that could detect bony lesions could improve the accuracy of radiologists staging cancer, which directly affects treatment decisions. Decreasing the error rate for detecting these lesions makes staging more accurate and leads to better treatment decisions for cancer patients.

Narrative(s)

A 65 year old patient with lung cancer who was not a surgical candidate was receiving a chemotherapy regimen for lung cancer. On a routine restaging chest CT, the thoracic radiologist described that the lung mass was stable and the mediastinal lymph nodes had decreased in size, but did not notice a new small lesion in the patient’s thoracic spine. This lesion was not discovered until 2 months later when the patient went to the ED with back pain and the bone lesion was much larger. Upon learning of the new lesion, the oncologist changed the chemotherapy regimen and the patient had a good response. Had this lesion been detected initially, this treatment change could have been made earlier, and may have prevented the ED visit and patient discomfort.

Workflow Description

Image obtained from modality and sent to PACS and the AI engine. Image analyzed by engine. System detects and describes skeletal metastases. This includes data on the size and classification of the metastases. An alert message is sent to PACS from the engine with the information and graphic highlighting the lesion(s).



Considerations for Dataset Development



Age

Varied

Sex at birth

Male, female

BMI

Varied

Primary tumor location

Prostate, lung, bowel, bladder, breast, uterus/cervix/ovary, melanoma

Pain

Yes, no

Location

Vertebrae, pelvis, femur, humerus, skull, appendicular skeleton

Characteristics

diffuse, focal, expansile

Disease

lytic, blastic, mixed

Technical Specifications


Inputs

DICOM Study

Procedure

Chest CT, Abdomen/pelvis CT, or Neck CT

Views

All

Data Type

DICOM

Modality

CT

Body Region

Chest, Abdomen, Neck

Anatomic Focus

N/A

Pharmaceutical

N/A

Scenario

N/A






Primary Outputs

Presence of skeletal metastasis

RadElement ID


Definition

presence of skeletal metastasis

Data Type

Categorical

Value Set

0-Skeletal metastasis absent

1-Skeletal metastasis present

2-Unknown

Units

N/A


Segmentation of skeletal metastasis

RadElement ID


Definition

segmentation of skeletal metastasis

Data Type

Coordinates

Value Set

N/A

Units

N/A


Max width of skeletal metastasis

RadElement ID


Definition

maximum width of the skeletal metastasis

Data Type

Numeric

Value Set

N/A

Units

mm


Disease classification

RadElement ID


Definition

classification of bony metastatic disease

Data Type

Categorical

Value Set

0-lytic disease

1-blastic disease

2-mixed

Units

N/A