Материал: part17

Figure E.1-2. Summary of Detections and Analyses Levels of Mammography CAD SR Content Tree​

The Summary of Detections and Summary of Analyses sub-trees identify the algorithms used and the work done by the CAD device,​ and whether or not each process was performed on one or more entire images or selected regions of images. The findings of the​ detections and analyses are not encoded in the summary sub-trees, but rather in the CAD Processing and Findings Summary sub-​ tree. CAD processing may produce no findings, in which case the sub-trees of the CAD Processing and Findings Summary sub-tree​ are incompletely populated. This occurs in the following situations:​

a.​All algorithms succeeded, but no findings resulted​

b.​Some algorithms succeeded, some failed, but no findings resulted​

c.​All algorithms failed​

Note​

1.​IfthetreecontainsnoIndividualImpression/Recommendationnodesandallattempteddetectionsandanalysessucceeded​ then the mammography CAD device made no findings.​

2.​Detections and Analyses that are not attempted are not listed in the Summary of Detections and Summary of Analyses​ trees.​

3.​If the code value of the Summary of Detections or Summary of Analyses codes in TID 4000 “Mammography CAD​ Document Root” is "Not Attempted" then no detail is provided as to which algorithms were not attempted.​

- Standard -​

Figure E.1-3. Example of Individual Impression/Recommendation Levels of Mammography CAD SR​

Content Tree​

TheshadedareainFigureE.1-3demarcatesinformationresultingfromDetection,whereastheunshadedareaisinformationresulting​ fromAnalysis.ThisdistinctionisusedindeterminingwhethertoplacealgorithmidentificationinformationintheSummaryofDetections​ or Summary of Analyses sub-trees.​

The clustering of calcifications within a single image is considered to be a Detection process that results in a Single Image Finding.​ The spatial correlation of a calcification cluster in two views, resulting in a Composite Feature, is considered Analysis. The clustering​ of calcifications in a single image is the only circumstance in which a Single Image Finding can result from the combination of other​ Single Image Findings, which must be Individual Calcifications.​

OnceaSingleImageFindingorCompositeFeaturehasbeeninstantiated,itmaybereferencedbyanynumberofCompositeFeatures​ higher in the tree.​

E.2 Mammography CAD SR Observation Context Encoding​

•​Any content item in the Content tree that has been inserted (i.e., duplicated) from another SR object instance has a HAS OBS​ CONTEXT relationship to one or more content items that describe the context of the SR object instance from which it originated.​ This mechanism may be used to combine reports (e.g., Mammography CAD 1, Mammography CAD 2, Human).​

•​By-reference relationships within Single Image Findings and Composite Features paraphrased from prior Mammography CAD SR​ objects need to be updated to properly reference Image Library Entries carried from the prior object to their new positions in the​ present object.​

The Impression/Recommendation section of the SR Document Content tree of a Mammography CAD SR IOD may contain a mixture​ ofcurrentandpriorsingleimagefindingsandcompositefeatures.Thecontentitemsfromcurrentandpriorcontextsaretargetcontent​ items that have a by-value INFERRED FROM relationship to a Composite Feature content item. Content items that come from a​ context other than the Initial Observation Context have a HAS OBS CONTEXT relationship to target content items that describe the​ context of the source document.​

- Standard -​

In Figure E.2-1, Composite Feature and Single Image Finding are current, and Single Image Finding (from Prior) is duplicated from​ a prior document.​

Figure E.2-1. Example of Use of Observation Context​

E.3 Mammography CAD SR Examples​

The following is a simple and non-comprehensive illustration of an encoding of the Mammography CAD SR IOD for Mammography​ computer aided detection results. For brevity, some Mandatory content items are not included, such as several acquisition context​ content items for the images in the Image Library.​

E.3.1 Example 1: Calcification and Mass Detection With No Findings​

AmammographyCADdeviceprocessesatypicalscreeningmammographycase,i.e.,therearefourfilmsandnocancer.Mammography​ CAD runs both density and calcification detection successfully and finds nothing. The mammograms resemble:​

Figure E.3-1. Mammograms as Described in Example 1​

- Standard -​

E.3.2 Example 2: Calcification and Mass Detection With Findings​

AmammographyCADdeviceprocessesascreeningmammographycasewithfourfilmsandamassintheleftbreast.Mammography​ CAD runs both density and calcification detection successfully. It finds two densities in the LCC, one density in the LMLO, a cluster​ of two calcifications in the RCC and a cluster of 20 calcifications in the RMLO. It performs two clustering algorithms. One identifies​ individual calcifications and then clusters them, and the second simply detects calcification clusters. It performs mass correlation and​ combinesoneoftheLCCdensitiesandtheLMLOdensityintoamass;theotherLCCdensityisflaggedNotforPresentation,therefore​ not intended for display to the end-user. The mammograms resemble:​

- Standard -​

Figure E.3-2. Mammograms as Described in Example 2​

The content tree structure in this example is complex. Structural illustrations of portions of the content tree are placed within the​ content tree table to show the relationships of data within the tree. Some content items are duplicated (and shown in boldface) to fa-​ cilitate use of the diagrams.​

Mammography

CAD Report

CONTAINER

CONTAINS CONTAINS CONTAINS CONTAINS

Figure E.3-3. Content Tree Root of Example 2 Content Tree​

- Standard -​