Transcript Lec5_Vis_In_Radiology - Visual Computing for Medicine

Medical Visualization in Radiology
Radiology
technician
Relevant images
Reporting
Bernhard Preim
1
Outline
•
•
•
•
Radiology Departments
Storage of Medical Image Data
Conventional Film-Based Diagnosis
Soft-Copy Reading
• Digital Hanging Protocols
• 3D Visualizations in Radiology
Bernhard Preim
2
Radiology Departments
• Large variety of imaging devices (CT, MR, Computed X-Ray, …)
• Large variety of output devices (printer, CRT-monitors with
different spatial and grey-value resolution)
• Archival of image data in PACS-systems
• Huge amounts of data
• Imaging devices, output devices, workstations and PACS are
interconnected.
• Performance problems due to the large amounts of data are
frequent and annoying. → Spatial resolution is often reduced to
avoid performance problems.
Bernhard Preim
3
Radiology Departments
Radiology technician
Relevant images
Reporting
• General radiological workflow
Image Courtesy Jens Breitenborn,
MeVis Diagnostics
Bernhard Preim
4
Radiology Departments
Resulting requirements?
• Interoperability of devices from different manufacturers.
• Standardized format and description of image data.
• Data protection, security and reliable identification of patient and
body part, e.g. left leg
• „Correct“ or optimal display at different output devices
Bernhard Preim
5
Storage of Medical Image Data
• DICOM (Digital Imaging and Communications in Medicine)
• Established by NEMA (National Electronics Manufacturer‘s
Association) in 1993
• Enables digital communication between imaging devices,
information systems and referring physicians (other dept.s of the
hospital or external)
• Enables exchange of data between devices from different
manufacturers
• Conformance to the DICOM standard is formally certified.
Bernhard Preim
6
Storage of Medical Image Data
Scope of DICOM
• 20 parts, some 2200 pages
• 26 working groups discuss extensions
• Working group 24: Surgical DICOM
• Questions, such as: How to represent 3D meshes in DICOM
including different representations of an object, application of
textures, etc?
• Working group 26: Extensions for Digital Pathology
• Other working groups: Extensions for representing implant
geometries, radiation treatment planning, …
Bernhard Preim
7
Storage of Medical Image Data
CT and MRI Data in DICOM
• A series of individual DICOM files, each representing a slice.
• DICOM files belonging to one dataset are characterized by the
same patient-id.
• Important tags are included in the legend of a viewer (examples are MeVisLab
2D Viewers).
Bernhard Preim
8
Storage of Medical Image Data
Mandatory and optional tags summarized to groups
• Patient data:
Name, birth date, sex, optional: weight
• Image data:
study-id, series number, image group
• Image presentation: slice distance, pixel spacing, default
presentation (window width, window height)
• Acquisition parameters: sequence name, special description,
reconstruction filter, position and orientation of the patient
• Special parameters for a certain modality: e.g. in MR field
strength, used coils, echo and repetition time
Bernhard Preim
9
Conventional Film-based Diagnosis
Cooperation between radiologist and radiology technician:
• Radiologist decides which modality and scanning
parameters are used
• Radiology technician performs the
procedure, operates the device
and
arranges and prepares the
image data for
the diagnosis
Image Courtesy Sebastian Meyer, MeVis
Diagnostics
Bernhard Preim
10
Conventional Film-based Diagnosis
Result:
• Written report, e.g. using voice recorder
• Complex cases: Demonstration for the referring physician and/or
interdisciplinary discussion, e.g. tumor board
Digital solutions should support these processes!
Bernhard Preim
11
Written report
•Bernhard Preim
12
Conventional Film-based Diagnosis
•
•
•
•
•
Magnifying glasses are used intensively.
Simultaneous reading of „old“ and „new“ images
Careful and efficient arrangement of image data
Plenty of space for viewing
Films: High spatial and gray level resolution
Drawbacks:
• Quantitative analysis is heavily restricted.
• Diagnostic processes are rather subjective.
Bernhard Preim
13
Conventional Film-based Diagnosis
Why is it important?
• Radiologists are accustomed to this proces. Digital processes
should be somehow „similar“.
• Radiologists are extremely efficient in film-based diagnosis. Digital
processes must be at least that efficient.
• Radiologists cooperate with radiology technicians and focus on the
high-level tasks. Thus, the perspective of the radiology technicians
is also essential for adopting digital solutions.
Bernhard Preim
14
Soft Copy Reading
LCD and CRT monitors:
High requirements for CT and MRI viewing, w.r.t.
• Display size,
• Spatial resolution,
• Maximum contrast,
• Light intensity,
• Quality (very low geometric distortion, very few defect
pixels)
Careful regular tests including recalibration
Bernhard Preim
15
Soft Copy Reading
Special monitors with high spatial
resolution (2000x2500 pixels)
and high gray value resolution
(10 bit) are used for reading Xray images.
• Layout corresponds to the
habits in conventional filmbased diagnosis
• Enables easy comparison
between 1-year old and current
data.
Image Courtesy, MeVis BreastCare, 2002
Bernhard Preim
16
Soft Copy Reading
Image Courtesy, MeVis BreastCare, 2012
Bernhard Preim
•17/62
Soft Copy Reading
Quantitative analysis of gray values in a ROI for assessing the
severity of diseases, such as emphysema and fibrotic disease and
bone mineral density (osteoporosis)
Analysis of cross-sectional areas (severity of a stenosis)
Bernhard Preim
18
Soft Copy Reading
• Digital Lightbox (BrainLab)
Bernhard Preim
17
3D Visualizations
3D Visualizations in radiology
• Are rarely used at all, since radiologists are trained to infer
spatial relations from cross-sectional images.
• Support an overview, not an in-depth analyis
• Primarily in case of rare anatomic variants, complex fractures.
• The radiologist „only“ describes the data.
• Are used whenever the referring physician requires it.
Preferred viewing modes:
• Maximum-intensity projection (often as Movie)
• Oblique Multi-planar reformations
• Slab rendering (trade-off between 2D and 3D)
• Direct volume rendering with predefined TFs
Bernhard Preim
20
3D Visualizations
Image Courtesy
B. Terwey, Bremen
MIP renderings of cerebral MR angiography data. Diagnostic task:
search for vascular malformations, e.g. aneurysms
Bernhard Preim
21
3D Visualizations
Oblique MPRs and slices
Image Courtesy Tobias Boskamp, MeVis
Research
Bernhard Preim
22
3D Visualization
Orientation cube (size and placement is adjustable). Also sticky
figures may be used.
Bernhard Preim
•23/62
3D Visualization
Radiation treatment planning (Brainlab)
Bernhard Preim
•24/62
3D Visualization
Radiation treatment planning (Brainlab)
Bernhard Preim
•25/62
3D Visualizations
Image Courtesy Volker Dicken, MeVis
Research
Thin Slab-Rendering of CT thorax data. Diagnostic task: search
for lung nodules (suspicious of being malignant)
Bernhard Preim
26
3D Visualizations
Image Courtesy, SIEMENS Medical
Solutions
Volume rendering with pre-defined transfer functions (gallery
concept) based on high-resolution image data.
Bernhard Preim
27
3D Visualizations
Synchronized combinations of
2D and 3D visualizations are
useful for treatment planning.
Bernhard Preim
28
3D Visualizations
In dentistry, Digital Volume Tomography is employed. Images are
also presented in the layout with 4 views.
Bernhard Preim
•29/62
Fused Multimodal Visualizations
Clipping planes and lenses in CT data provide a window to
underlying MRI data. Also radiation doses are overlaid.
Bernhard Preim
•30/62
Interventional Radiology
Bernhard Preim
31
Summary
• The introduction of new techniques in radiology requires careful
integration in the infrastructure of radiological departments.
• DICOM is essential, also for communicating results.
• Analysis of conventional film-based diagnosis is inspiring.
• Try to improve on the drawbacks, but also to maintain the
positive aspects.
• Talk to users, analyze their needs and distinguish between
radiologists, radiology technicians and referring physicians.
Outlook:
• Similar changes occur currently in pathology with much larger
microscopy data.
Bernhard Preim
32
Links
• Integrating the health care enterprise
http://www.ihe.net/
• DICOM http://medical.nema.org/
• DICOMscope - DICOM-Viewer
http://dicom.offis.de/dscope.php.de
• David Clunie's dicom3tools
http://www.dclunie.com/dicom3tools.html
Bernhard Preim
33