Læknablaðið - 15.03.1983, Qupperneq 17

LÆK.NABLADIÐ
73
possibilities from which the actual information
has been chosen with certain probability. Or it
is the amount of yes-no questions, which is
needed to exciude the actuai information from
the group. A simple formula to measure the
information transmission capacity of a noise-
Iess channel in bits can be presented in
Shannon’s terms as follows
in which
C = information transmission capacity
W = the width of the information transmis-
sion passband
P = the average signalpower
T = time
The formula is valid for a noiseless channel.
Shannon also presented a more general for-
mula for channels with noise, but there the
formula was denominated an indefinite form,
adopting which necessitated a number of new
concepts and notable mathematical know-
ledge«.
Upplýsingayfirfærsla verður alltaf í raun
fyrir meiri eða minni truflunum (noise, Raus-
chen) 35, 36, 40, 41, 42). í upplýsingakerfi
röntgengreiningarinnar gætir þessara truflana
annars vegar á sviði hinna tæknilegu yfir-
færsluþátta a)- c) í skipuriti 4.5.1. og hinsvegar,
á c)-d) — stigum þeirrar ferilmyndar, þar sem
líkur á truflunum eru mun meiri en á öðrum
stigum, eins og augljóst er (sbr. einnig mynd
4.2.1.)
4.7. í grein um myndgæðaeftirlit í röntgen-
greiningu, o.fl. segir Rossman svo um þau efni,
sem rædd voru í köflum 4.1.- 4.6. (42):
»The task is compiicated by the fact that all
the operations constituting the radiologic
process are intimately interrelated in that the
output from one forms the input to the next.
Two important consequences follow from this
fact. First, the diagnostic information trans-
mitted by the radiologic process as a whole
cannot be greater than that transmitted by
any of the operations constituting the process.
In order to obtain maximum over-all trans-
mission of diagnostic information, or the most
reliable diagnosis, therefore, each operation
should be designed to deteriorate as little as
possible the diagnostically significant informa-
tion put into it. Second, none of the operations
can be optimized without sufficient knowledge
regarding at least those operations with
which it is interfaced. For example, it is
impossible to design the best imaging and
recording system without knowing what is to
be imaged and how the image is to be read
out. Once the input to the system is known, an
imaging system can be designed, at least in
principle, which converts the input into an
output that is compatible with the subsequent
operation.
This approach also takes into consideration
the constraints imposed by the radiologic
process as a whole and, thus, helps to avoid
overdesigning of one operation relative to the
limitation introduced by another. For example,
it is useless to design a high-resolution ima-
ging system when resolution is limited in the
input by geometrical unsharpness introduced
as a result of tube load limitation in the
exposing operation«.
Rossman ræðir í sömu grein (42) greiningar-
þátt röntgenferilsins og kemst svo að orði:
»The recognition operation takes place in the
brain of the radiologist and involves such
intangibles as comparison of a given radio-
graphic image with a remembered »normaI«
image established through education and ex-
perience and the pattern-recognition ability of
the radiologist. The sharp division between
visual detection and recognition postulated
here may appear to be rather artificial, since
these operations are closely interwoven in
practice. It is certainly true, however, that the
diagnostic significance of radiographic images
cannot be recognized by the radiologist unless
they are first recorded detectably. In this
context, it is meaningful to state that radio-
graphic image quality is that attribute of the
image which affects the certainty with which a
diagnostically important detail can be detec-
ted visually by the radiologist. A radiographic
image is of the highest quality if it does not
adversely affect the diagnosis, which is the
final output of the radiologic process or, in
other words, if it does not limit diagnostic
certainty«.
4.8. Þegar ég var að hugleiða gerð og innihald
skráningarkerfis fyrir röntgengreiningar,
sýndist mér eðlilegt að skoða þá þætti kerfis-
ins, sem þýðingu hafa, á þann hátt og með
þeim röksemdum, sem lýst er í köflum 4.1.-4.7.
Augljóst virtist, að lýsing röntgenrannsóknar