This session contained four presentations on PACS-CAD integration.
Dr. Regge gave an outlook of a radiologist on how to use the CAD in the clinical workflow. He discussed the problems with double reading with increased recall rate of patients for biopsy and the cost involved in requiring two experts to review all datasets. This provides a good incentive to start using CAD as a second reader. Studies have shown similar performance with increased recall rates but also with reduced cost of the diagnostic process because of having only one reader and the decrease in reading time. Controversies are that radiologists might reject true positive polyps defined by CAD and the influence of CAD on the reading process. Inconclusion CAD is coming to maturity and can be used but still is controversial and the clinical value still needs to be assessed.
Dr. Suárez Cuenca presented on an integration of CAD into the PACS environment using a wide computing infrastructure. They aim to build a sytem for a whole region in Spain to make CAD assessable for multiple hosptials utilizing different PACSs and clinical workstations. They build a standards based platform with which users can request a CAD service and receive the results in their local PACS. The process is running through a webinterface that allows sending their (anonymized) data to the CAD service and receiving a DICOM object with the result.
Next a presentation on Enhancing clinical use of CAD systems in PACS with automation and open-source tools by dr. Summers. They built a system that allows a better integration of CAD into the normal workflow utilizing the clinical information already available in the PACS to automate the utilization of CAD in three different scenario's. Either all studies are processed automatically based on their properties (8 minutes per case), a radiologist selects a specific case and requests processing (6 min per case), or a list of cases is predefined in PACS and processed automatically with notification of completion to the radiologist by email or text message (9 min per case). This method allows non disruptive integration with the ability to utilize the full capabilities of the PACS workstation, such as direct comparison with older data.
The final presentation by dr. Behlen was entitled 'an unexamined assumptionnis not worth assuming: imaging data quality exposed in data migration'. this talk was about PACS data migration by Laitek, a company specializing in this area. They presented figures showing exceptions ranging from almost 0 to 35%. Majority of the migrations will only have 1-2% of exceptions. Many problems occur related to interpretation of the DICOM standard. One of the problems is the utilization of annotations, where everybody is implementing in a different way resulting in difficulties to transfer them from one to another PACS environment.
Dr. Regge gave an outlook of a radiologist on how to use the CAD in the clinical workflow. He discussed the problems with double reading with increased recall rate of patients for biopsy and the cost involved in requiring two experts to review all datasets. This provides a good incentive to start using CAD as a second reader. Studies have shown similar performance with increased recall rates but also with reduced cost of the diagnostic process because of having only one reader and the decrease in reading time. Controversies are that radiologists might reject true positive polyps defined by CAD and the influence of CAD on the reading process. Inconclusion CAD is coming to maturity and can be used but still is controversial and the clinical value still needs to be assessed.
Dr. Suárez Cuenca presented on an integration of CAD into the PACS environment using a wide computing infrastructure. They aim to build a sytem for a whole region in Spain to make CAD assessable for multiple hosptials utilizing different PACSs and clinical workstations. They build a standards based platform with which users can request a CAD service and receive the results in their local PACS. The process is running through a webinterface that allows sending their (anonymized) data to the CAD service and receiving a DICOM object with the result.
Next a presentation on Enhancing clinical use of CAD systems in PACS with automation and open-source tools by dr. Summers. They built a system that allows a better integration of CAD into the normal workflow utilizing the clinical information already available in the PACS to automate the utilization of CAD in three different scenario's. Either all studies are processed automatically based on their properties (8 minutes per case), a radiologist selects a specific case and requests processing (6 min per case), or a list of cases is predefined in PACS and processed automatically with notification of completion to the radiologist by email or text message (9 min per case). This method allows non disruptive integration with the ability to utilize the full capabilities of the PACS workstation, such as direct comparison with older data.
The final presentation by dr. Behlen was entitled 'an unexamined assumptionnis not worth assuming: imaging data quality exposed in data migration'. this talk was about PACS data migration by Laitek, a company specializing in this area. They presented figures showing exceptions ranging from almost 0 to 35%. Majority of the migrations will only have 1-2% of exceptions. Many problems occur related to interpretation of the DICOM standard. One of the problems is the utilization of annotations, where everybody is implementing in a different way resulting in difficulties to transfer them from one to another PACS environment.
No comments:
Post a Comment