Automated Detection of Metastatic Bone Disease on Bone Scintigraphy Scans
Study Details
Study Description
Brief Summary
Bone scintigraphy scans are two dimensional medical images that are used heavily in nuclear medicine. The scans detect changes in bone metabolism with high sensitivity, yet it lacks the specificity to underlying causes. Therefore, further imaging would be required to confirm the underlying cause. The aim of this study is to investigate whether deep learning can improve clinical decision based on bone scintigraphy scans.
Condition or Disease | Intervention/Treatment | Phase |
---|---|---|
|
Study Design
Arms and Interventions
Arm | Intervention/Treatment |
---|---|
BS-UKA Patients who underwent bone scintigraphy scanning between 2010 and 2018 at RTWH Aachen university hospital, and had a bone scan report that indicates the presence or absence of metastatic bone disease. |
Other: Deep learning based detection of metastatic bone disease on bone scintigraphy scans.
The aim is to investigate whether deep learning algorithms can detect bone metastasis with high accuracy and specificity.
|
BS-Namur Patients who underwent bone scintigraphy scanning between 2010 and 2018 at Namur university hospital, and had a bone scan report that indicates the presence or absence of metastatic bone disease. |
Other: Deep learning based detection of metastatic bone disease on bone scintigraphy scans.
The aim is to investigate whether deep learning algorithms can detect bone metastasis with high accuracy and specificity.
|
BS-Aalborg Patients who underwent bone scintigraphy scanning between 2010 and 2018 at Aalborg university hospital, and had a bone scan report that indicates the presence or absence of metastatic bone disease. |
Other: Deep learning based detection of metastatic bone disease on bone scintigraphy scans.
The aim is to investigate whether deep learning algorithms can detect bone metastasis with high accuracy and specificity.
|
Outcome Measures
Primary Outcome Measures
- The classification performance of DL algorithm compared to the ground truth [June 2021]
Reporting the performance measures (Area under the curve, accuracy, specificity..etc)
Secondary Outcome Measures
- Comparing the classification performance of the DL algorithm to that of physicians [June 2021]
Correctness of the diagnosis of Dr versus AI (dichotomous variable: correct versus not correct) on a subset of the validation data, using a McNemar statistical test
Eligibility Criteria
Criteria
Inclusion Criteria:
- Patients who underwent a bone scintigraphy scan that is available with the radiologic report between 2010-2018
Exclusion Criteria:
- The lack of a bone scan, or corresponding radiologic report
Contacts and Locations
Locations
Site | City | State | Country | Postal Code | |
---|---|---|---|---|---|
1 | Maastricht University | Maastricht | Limburg | Netherlands | 6229ER |
Sponsors and Collaborators
- Maastricht University
- Aalborg University Hospital
- Centre Hospitalier Universitaire de Liege
- University Hospital, Aachen
- University of Namur
Investigators
None specified.Study Documents (Full-Text)
None provided.More Information
Publications
None provided.- MBDDL