10. Anatomy Detection

Agora automatically detects which anatomy a Study shows and, for cardiac Studies, what kind of cardiac scan each Series is. The result is displayed in the detail sidebar, can be searched for, and can optionally be shown as a tag — without anyone having to label the data by hand.

10.1. What is detected

Anatomy of a Study

Each Study gets one anatomy value, for example HEAD, HEART, SPINE, KNEE, LIVER, BREAST, PROSTATE, or PHANTOM.

The anatomy is derived from the thumbnails of the Study’s Series: every Series is classified individually and the Study gets the value that the Series agree on most confidently. Reference scans, coil surveys, and localizers are left out, because they do not show the anatomy reliably. If the detection is not confident enough, the Study gets no value rather than a doubtful one.

Cardiac Series

For Studies detected as HEART, each Series is classified further. The individual properties are combined into one short cardiac label such as CINE-SAX, T1-4CH, or DIXON-WH:

Property

Values

View

Short axis (SAX), two-chamber (2CH), three-chamber (3CH), four-chamber (4CH), or other

Acquisition

CINE, FLOW, T1 map (T1), T2 map (T2), T2* map (T2S), water/fat (DIXON), or other

Temporal

Static or dynamic

Dimensionality

2D or 3D (3D is appended to the label)

Coverage

Whole heart (WH replaces the view in the label) or local

10.2. When the detection runs

  • After import — a Study is classified a few minutes after its last Series has arrived. The short delay lets all Series of a Study arrive first, so the Study is classified once on the complete data.

  • Every night — a nightly job classifies everything that was missed, including Studies that were imported before the feature was available.

A value that was set manually is never overwritten by the automatic detection.

10.3. Viewing and correcting the result

Select a Study to see its Anatomy in the detail sidebar, next to the rating. The icon behind the value tells where it comes from:

  • A robot icon — detected automatically. Hover over the value to see the confidence of the detection.

  • A user icon — set manually.

Users who are allowed to manage data in the project can change the result:

  • Click the value or to pick a different value from the list.

  • Click to clear the value.

  • If there is no value, click the Recalculate automatically icon to run the detection again for this Study.

For a cardiac Series, the detail sidebar shows the Cardiac label. Click the arrow next to it to show the individual properties (View, Temporal, Acquisition, Dimensionality, Coverage) and correct each of them in the same way.

10.4. Searching

Detected values are included in the quick search. Searching for heart, for example, also finds the Studies whose anatomy was detected as HEART.

10.5. Showing the anatomy as a tag

By default the detected anatomy is only shown in the detail sidebar. Project managers can make it more visible in Project Settings → General, in the Anatomy section:

  • Show detected anatomy as a tag — Studies show an extra tag-styled badge with the detected anatomy in the Tags column of the study list, and cardiac Series show their cardiac label in the same way. This is display-only: no tag is created, and the badge cannot be edited or used as a tag filter.

  • Automatically create a real tag from the detected anatomy — a real tag is created for each Study’s detected anatomy and for each cardiac Series’ label. These tags behave like any other tag: they can be filtered on, used in queries, and removed. They are collected in the tag groups Anatomy and Cardiac. When the option is switched on, the Studies already in the project are tagged as well.

If an administrator has enabled one of these options for all projects, the corresponding checkbox is replaced by a note saying so.

Note

Administrators can switch off the automatic detection for the whole server under Settings → Server → General → Classification. When it is switched off, no new values are computed and the Anatomy and Cardiac entries are hidden everywhere. See Administration Settings.