w7

Question 1

what is Health informatics

Answer 1

It deals with the resources, devices, and methods required to optimize the acquisition, storage, retrieval, and use of information in health and biomedicine

Question 2

what is the need for health informatics

Answer 2

Practitioners cannot rely on memory alone.
Provide access to knowledge bases.
Health information to be shared among authorized persons.
Continuity of care (Patient information should be available to any authorized healthcare professional)

Question 3

what are some Medical Data Management Systems

Answer 3

HIS (Hospital information system)
RIS (Radiology information system)
PACS (Picture archive and communication system)
Anaesthesia system
Order entry system
Pharmacy system
Surgery scheduling system

Question 4

PACS originated as an image management system for Radiology practice.

PACS has evolved into a health care enterprise-wide system that integrates information media in multiple forms, including voice, text, medical records, waveform images, and video recordings

To integrate these various data types, PACS requires the technology of multimedia. What are these multimedia ?

Answer 4

hardware platforms
information systems and databases
communication protocols
display technology
system interfacing and integration

Question 5

what kind of text data is collected for medicine

Answer 5

includes general patient data (name, birthday, address…), medical history, symptoms (type, duration, timing), past medical treatment, examination etc

Question 6

Why may valuable diagnostic and prognostic information contained in databases be unusable.

Answer 6

The majority of acquired medical images are currently stored with a limited text-based description of their content.

Question 7

What is the alternative to Content-Based Image Retrieval (CBIR)

Answer 7

Image retrieval based on the textual annotation of images, i.e., images were first manually annotated by keywords or descriptive text and organized by topical or semantic hierarchies in traditional DBMS to facilitate easy access based on standard Boolean queries.

Such purely text-based method poses significant limitations in image retrieval

Question 8

what is the benefit of CBIR compared to text based retrieval

Answer 8

support full retrieval by visual content / properties of images, i.e., retrieving image data at a perceptual level with objective and quantitative measurements of the
visual content and integration of image processing, pattern recognition and computer vision.

this is opposed to text-based annotations, which are ultimately subjective judgements and prone to inconsistancies.

Question 9

what are some common features used in CBIR, and why is it used

Answer 9

Color: most frequently used visual feature due to its invariance with respect to image scaling, translation and rotation, and to its three-color-component values which make its discrimination superior to grayscale images. robust to background complication.

Texture: widely used in pattern recognition and computer vision for identifying visual patterns that cannot result from the presence of only a single color or intensity.

Shape: used to identify an object or region as a meaningful geometric form. To humans, perceiving a shape is to capture prominent elements of the region. Normally represented after the image has been segmented into objects or regions.

Spatial relationships: between multiple objects or regions in an image. Usually capture the most
relevant and regulated part of the information in the image content.

Question 10

The Minkowski distance is a metric in a normed vector space which can be considered as a generalization of both the Euclidean distance and the Manhattan distance

When is Minkowski-Form (MF) Distance used

Answer 10

If each dimension of image feature vector is independent of each other and is of equal importance, the MF distance is appropriate for calculating the distance between two images.

Question 11

what is Histogram Intersection

Answer 11

if you think about overlaying two color intensity histograms, histogram intersection is the intersection of the two histograms

The Histogram Intersection can be taken as a special case of MF distance (when p=1) which is used to compute the similarity between color images.

Question 12

what are the four different categories of CBIR in Medical Domain (CBMIR)

Answer 12

physical visual features (color and texture)

geometric spatial features (shape, 3D volumetric features, spatial relationships)

combination of semantic and visual features (semantic pathology interpretation and generic models)

physiological functional features (dynamic activities)

Question 13

Color is the most extensively used low-level feature for the CBIR. However, since the majority of medical images are intensity-only-images, the color-based retrieval would only be applicable to medical images based on light photography.

what type of medical images fall under this category

Answer 13

Histological images: light microscopy, usually possess unique color signatures, including various subtle changes in color such as jaundice, congestion, changes in exudation and effusion, and pigmentation.

Dermatoscopic images: Skin color is produced by complex mechanisms and utilized to interpret the characteristics of a lesion and the depth in skin at which the lesion exists, for analysis of skin erythema, evaluation of wound status, and early detection of skin cancers.

Endoscopic images

Question 14

Retrieval by color similarity requires that models of color stimuli are used, such that distances in the color space correspond to human perceptual distances between colors.

what are some commonly used color descriptors

Answer 14

color moments

color histogram

color coherence vector (CCV)

color correlogram

Question 15

which color model is most suitable for image retrieval and why

Answer 15

HSV (or HSL, or HSB) space is widely used in computer graphics and is a more intuitive way of describing color, compared to RGB.

The hue is invariant to the changes in illumination and camera direction and hence more suited to object retrieval.

Question 16

what are color moments (Physical Visual Features)

Answer 16

measures that characterise color distribution in an image.

The first order moment: mean
The second order moment: variance
The third order moment: skewness

Question 17

what makes color moments a good first pass to narrow down the search space before other sophisticated color features are used for retrieval

Answer 17

Since only 9 (3 moments for each of the 3 color components) numbers are used to represent the color content of each image, color moments are very compact representations compared to other color features. Due to this compactness, they may also lower the discrimination power.

Question 18

why would you use a color histogram for image retrieval (Physical Visual Features)

Answer 18

The color histogram is easy to compute and effective in characterizing both the global and local distributions of colors in an image.

An effective representation of the color content of an image if the color pattern is unique compared with the rest of the data set.

It is also robust to translation and rotation about the viewing axis and changes only slowly with the scale and viewing angle.

Question 19

Most medical images acquired and displayed in gray scale are often highly textured, and consequently, examination of medical images usually requires interpretation of tissue appearance. It is becoming one of the most commonly used characteristics in medical image analysis, classification, and retrieval.

what texture properties may you want to observe in medical images

Answer 19

smoothness, coarseness, regularity, and homogeneity

Question 20

what are the two main texture descriptors (Physical Visual Features)

Answer 20

co-occurrence matrices approach

Gabor filters

Question 21

What is the co-occurrence matrices approach

Answer 21

based on the repeated occurrence of some gray-level configuration in the texture. This configuration varies
rapidly with distance in fine textures and slowly in coarse textures

step 1: To construct co-occurrence matrices for given directions and given distances.
step 2: To compute texture feature vectors for four directions, different values of d, and various statistic measurements.

Question 22

how can the computation of co-occurrence matrices be made more efficient

Answer 22

data binning of gray levels (don’t have to test all gray level combos).

Question 23

On their own, co-occurrence matrices do not provide any measure of texture that can easily be used as texture descriptors. The information in the matrices needs to be further extracted as a set of feature values.

what statistical information is extracted from co-occurrence matrices?

Answer 23

Haralick texture features - set of 14 statistical metrics. Most frequently used: Energy, Entropy, Contrast, Inverse Difference Moment, Correlation, and Variance.

can be global (features computed using all distances and directions - then averaged) or local

Question 24

Unlike another well-known Tamura features method which only provides visually meaningful texture properties (such as coarseness, contrast, directionality, linelikeness, regularity and roughness), what do co-occurrence matrices allow you to do

Answer 24

allows detecting some abnormalities in medical images that are beyond human limited appreciation of complexity and otherwise difficult to determine by other texture extraction methods, and provides valuable information about the medical images that may not be visible to the human eye.

Question 25

Compared with color and texture features, shape features are usually described after images have been segmented into regions or objects.

Since robust and accurate image segmentation is difficult to achieve, the use of shape features for image retrieval has been limited to applications where objects or regions are readily available.

what are some commonly used shape features in CBMIR

Answer 25

chain codes

fitting line segments

Fourier Descriptors (FDs)

invariant moments

Question 26

what are Chain Codes (Geometric Spatial Features)

Answer 26

represents a boundary using a connected sequence of straight-line segments of specified length and direction

chain code generated by following a boundary in a clockwise direction and assigning a direction to the segments connecting every pair of pixels.

Typically based on 4- or 8-connectivity of the segments. The direction of each segment is coded by using a numbering scheme.

Question 27

what is the limitation of digital chain codes

Answer 27

Any small disturbances along the boundary due to noise or imperfect segmentation cause changes in the code that may not be related to the shape of the boundary.

Question 28

how can you make digital chain codes more resistant against small noise ????????????

Answer 28

Resample the boundary by selecting a larger grid spacing. (e.g. 4-directional or 8-directional)

As the boundary is traversed, a boundary point is assigned to each NODE of the large grid, depending on the proximity of the original boundary to that node.

Question 29

Straight-line segments can give simple approximation of curve boundaries.

describe the process for fitting line segments (Geometric Spatial Features)

Answer 29

(1) Approximate the curve by drawing a line segment joining its end points (A, B);
(2) If the distance from the farthest curve point (C) to the segment is greater than a predetermined quantity, join AC and BC;
(3) Repeat the procedure for new segments AC and BC, and continue until the desired accuracy is reached.

Question 30

can CBMIR be performed on 3D images?

Answer 30

no, feature extraction and retrieval of 3D medical images in most of the existing CBMIR systems so far are still performed based on 2D slices.

although 3d retrieval is desired, as it is believed that it holds more useful information than 2d.

Question 31

what are some CBMIR techniques for 3D medical imaging data

Answer 31

3D iMSP (ideal midsagittal plane)

3D concentric sphere

VOI-FIRS (functional image retrieval system based on volume of interest)

Question 32

describe the steps for VOI-FIRS

Answer 32

The center of mass V of the VOI is computed.

Using V as the center, a series of 1 … k concentric spheres in 3D domain are constructed with regular increments of their radius.

For each sphere, the k dimensional feature vectors Fs and Fr is constructed, representing the fraction of the sphere occupied by the VOI and the fraction of the VOI occupied by the sphere, respectively. The sphere feature vectors map the entire VOI to a specific point in the k-dimensional space.

Question 33

Spatial content in surgical or radiation therapy of brain tumours is very decisive because of the location of adjacent structures will have profound implications on the therapeutic decision. Low-level features cannot always capture or describe these complex scenarios.

What are some methods of representing spatial relationships

Answer 33

Attributed Relational Graphs (ARGs)

region relationship matrix

Question 34

what are Attributed Relational Graphs (ARGs)

Answer 34

objects are represented by graph nodes, and the relationships between objects are constructed by arcs (edges) between such nodes.

Nodes have a value p, while edges have values rd and rp.

p=perimeter – calculated as the length of its bounding contour;
rd=relative distance – calculated as the minimum distance between their surrounding contours;
rp=relative position – inside or outside.

Question 35

how does a region relationship matrix work

Answer 35

The spatial relationships between all region pairs in a tissue image are represented by a n×n region
relationship matrix.

for each region pair, use:
the ratio of the common perimeter to the perimeter of the first region,
the distance between the centroids,
the angle between the horizontal axis and the line joining the centroids,
each region pair is assigned a degree of their spatial relationship using fuzzy class membership functions