Week Seven - Medical Data Storage & Retrieval Flashcards by Meg Flannery

What does PACS stand for?

Picture archiving and communication systems.

How well did you know this?

Not at all

Perfectly

What did it originate as?

An image management system for improving the efficiency of the radiology practice.

How well did you know this?

Not at all

Perfectly

What does RIS stand for?

Radiology information system

How well did you know this?

Not at all

Perfectly

What does HIS stand for?

Hospital information system

How well did you know this?

Not at all

Perfectly

What does EMR stand for?

Electronic medical record

How well did you know this?

Not at all

Perfectly

What is DICOM?

Digital imaging and communications in medicine. The universal format for PACS image storage and transfer.

How well did you know this?

Not at all

Perfectly

What is HL7?

Health Level 7 - electronic data exchange standard in health care

How well did you know this?

Not at all

Perfectly

What is PACS?

A medical imaging technology system which provides economical storage and convenient access to images from multiple modalities.

How well did you know this?

Not at all

Perfectly

What are the four major components in a PACS?

Image modalities
Secure network (acquisition gateway)
Workstations
Archives

How well did you know this?

Not at all

Perfectly

Why do we need PACS?

Eliminate the need to manually file, retrieve, or transport medical files.

Reduce medical errors, allow for larger image sizes, EMR improve continuity of care and shorten patient stays.

How well did you know this?

Not at all

Perfectly

What are 6 challenges with PACS?

Confidentiality of records
Reliability of components
Connectivity between components
Expensive
Training
Data migration

How well did you know this?

Not at all

Perfectly

What are some advantages of using a PACS? (4)

Simultaneous availability of images at different sites.
Faster access to archived images
Minimal space needed
Reduced incidence of lost data

How well did you know this?

Not at all

Perfectly

What is the future of PACS? (5)

Decision support tool
Education tool
Treatment planning (virtual simulation)
Image assisted surgery
Large scale integration of locally managed PACS

How well did you know this?

Not at all

Perfectly

What is CBIR?

Content based image retrieval. ‘Content based’ means that the search analyses the features of the image rather than its associated metadata (keywords, tags, descriptions).

How well did you know this?

Not at all

Perfectly

Why is CBIR better than other methods?

Searches that rely purely on metadata are dependent on annotation quality and completeness. Also manually entering metadata is time consuming.

How well did you know this?

Not at all

Perfectly

What are the two broad categories of feature extraction?

General visual features

Domain specific semantic features

How well did you know this?

Not at all

Perfectly

What are the 5 types of general visual features?

Colour
Texture
Shape
Spatial relationships
3D volumetric features

Why is colour the most frequently used visual feature for CBIR?

Invariant to image scaling, translation and rotation. That is, can be completed without regard to image size or orientation.

When is colour based retrieval used? (4)

Histological images
Dermatoscopic images
Endoscopic images
Tongue images

What does HSV stand for?

Hue = colour type
Saturation = vibrancy of colour
Value = brightness of colour

What is HSV?

A cylindrical coordinate representation of the RGB colour model in an attempt to be more intuitive.

Why are colour moments compact?

Since only 9 numbers are used to represent the colour content of each image. Normally used to narrow down the search space before other better methods are used for retrieval.

When is a colour histogram a good option for retrieval?

If the colour pattern is unique compared with the rest of the data set.

What is Meg’s middle name?

Erin

How do you identify specific textures in an image?

By modelling texture as a 2D grey level image, then comparing relative brightness of pairs of pixels.

Name 3 of Tamura's features of texture.

Coarseness, contrast, directionality, roughness, regularity, likeliness.

What are the two mainly adopted methods for textural retrieval?

Co-occurence matrices approach | Gabor Filters

What does shape mean in terms of image retrieval?

The shape of a particular region that is being sought out.

Why is shape information important?

In characterising pathologies - disease processes can affect the structure of organs.

What are the 3 steps involved in 3D concentric sphere features extraction?

1. Center of mass computed 2. Series of concentric spheres constructed with regular increments. 3. Vectors are constructed for each sphere to map to 3D space.

Why would using 3D volumetric features be useful for CBIR?

More accurate CBIR with more discriminating power can be achieved if we take full advantage of the information available in the 3D spatial domain.

Why are spatial relationships important for CBIR?

Given that abnormalities are defined as deviations from normal anatomy, spatial relationships between structures are critical to diagnosis. E.g. knowing where a brain tumour is will affect surgical therapy method

What is a ARG?

Attributed relational graph. Objects are represented by graph nodes and the relationships between objects are constructed by arcs between such nodes.

What is a region relationship matrix?

nxn matrix that represents spatial relationships between all region pairs.

What are the three categories of class relationship?

1. Perimeter - disjoined, bordering, invaded, surrounded 2. Distance - near, far 3. Orientation - right, left, above, below

Will Meg get with Jess this Friday night?

probably yes GTG

What is the key to success for CBIR?

Relating high level semantic knowledge with low level visual data to find medically meaningful similar cases.

What is an approach for semantic pathology interpretation CBIR?

A physician in the loop approach. Before archiving in the database, a physician spends a few moments delineating the pathology bearing regions and any relevant anatomical landmarks.

What are some key challenges CBIR needs to overcome? (4)

Majority of medical data is grey level - no colour retrieval. Usually low res and high noise - difficult to consistently automate retrieval. Diseased tissue often does not have a clear shape or boundary - also makes it difficult to automate retrieval. Most CBIR are task-specific (only limited to a particular modality or body part)

What is the main advantage of using CBIR?

By observing an abnormality in a diagnostic image, the physician can query a database of known cases to retrieve images that contain regions with features similar to what is observed in the image of interest.