Digital
Computer
Analysis and Display
of the
Radionuclide
Scan2
Donald W. Brown, M.D.1
Denver, Colorado
Many series correlating radionuclide scans with the clinical diagnosis have
been reported recently. Accuracy in these tests is generally 75 to 85 per cent.
The relatively high incidence of error is due in part to the subjective methods
of interpretation in use. Some method of bringing more objectivity to the evalu
ation of scans must be evolved, and we feel the appropriate application of modern
statistical methods should help to accomplish this end. Since the amount of data
involved in radionuclide scanning is so large, this application of statistical pro
cessing requires the use of a digital computer.
Reports of these applications are now becoming available in the literature.
Kawin (1, 2) and Winkler (3) were the first to describe the mechanics of systems
designed to perform this type of study. Winkler's publication is particularly note
worthy, because of his interest in applying statistical testing to the evaluation
of variations in intensity of counts between neighboring areas. Apparently, the
first successful application of these techniques to patients was published by
Brown in 1964 (4, 5). Tauxe (6) presented findings using a similar method to
that of Winkler and Brown in 1965. He showed convincingly the potential value
of the digital computer in performing contrast enhancement and background
erase. Gorton (7) is presently using a computer in the analysis of scans quanti
tating experimental myocardial infarctions in isolated animal hearts. Many other
investigators are becoming interested in the application of computers to scan
ning (8).
1From the Department of Radiology, University of Colorado Medical Center, Denver,
Colorado 80220.
2This work was supported in part by U. S. Atomic Energy Commission Contract
#AT-( 11-1)-1472.
DIGITAL COMPUTER ANALYSIS AND DISPLAY OF RADIONUCLIDE SCAN 741
The following is a description of the method and equipment presently in use
at the University of Colorado Medical Center for digital computer analysis of
the radionuclide scan.
METhODS
A modified Picker Magnascanner with a three-inch sodium iodide crystal is
used. Pulses from the pulse height analyzer of the scanner are received by an
events-per-unit time (EPUT) instrument.' This instrument receives the train of
random pulses coming out of the analyzer and sums them for a preset amount of
time. It then discharges the resulting count into a paper tape perforator, which
punches the count in code compatible with the paper tape reader of the computer.
The EPUT device has time intervals which are variable from 0.4 to 1.0 seconds
in increments of 0.1 seconds. In order to avoid significant amounts of dead time
while data is being transferred to the tape perforator, the instrument incorporates
two scalers. When the first is finished with a count, control is passed to the
second which begins a new sum while the first discharges its data into the paper
tape perforator, after which it is set to zero and remains in wait until the second
scaler has completed its count. Dead time between the two counters achieved
by this method is on the order of a few microseconds. Alternating in this way,
the scalers act as buffers for each other.
In order to record all the data sequentially on the punched paper tape, it
is necessary to add signals which the computer can recognize as the ends of
scanning lines and the end of a scan. The method in use presently is to punch
four characters for each count. The first digit is used to show the direction in
which the scanner is moving, while the last three digits are used to record the
count, decimally. If the crystal is moving away from the scanner, the first digit
is a one, and if the crystal is moving toward the scanner, this is a zero. The
EPUT device also encodes an end of record mark after every tenth four digit
‘C-Deck, Colorado Instrument Co., Broomfield, Colorado.
FABLE I
ORGANS STUDIED IN SERIES OF 100 COMPUTED SCANS
Photoscans (Per Cent)Computed Scans (Per Cent) 7Technically Superior5Diagnostically Superior213Agreement
with Clinical Findings9292
number and a ciollar sign at the end of a scan to transfer control back to cards.
Data such as the patient's name, date, dose, and form of isotope, background,
scanning speed, and other pertinent information are punched on cards. These
cards control the reading of the punched paper tape. The computer program
presently in use has been modified extensively from that originally developed,
so that it is now both efficient and flexible.'
The punched paper tape and data cards are processed by an IBM 7044 com
puter at the Graduate School Computer Center of the University of Colorado.
Processing takes less than a minute for most scans and proceeds as follows.
Once the data is read by the computer, the number of data points in each
scan line are counted and the total number of lines in the scan determined. Since
the scanner traverses in one direction, then spaces and traverses back in the
other direction, the next step necessary is to reverse every other line. In doing
this, the entire array of data is converted from one to two dimensions. Next, each
data point is replaced by the sum of its eight surrounding neighbors and two
times the actual data point being replaced, thus averaging out irregularities in
the scan, but not accomplishing a statistical processing of the data noted above
to be the long term goal of this project. Intensive studies in this direction are
under way, but as yet have not resulted in the development of a more satis
factory method.
The computer next determines the maximum of all of the averaged values
in the array and on this basis, weights all of the numbers so that they now
range from one to ten. Now, using the printer of the computer, eight pictures
of the organ under study are printed out. Seven of these are printed with increas
ing background erase. This is achieved in the following manner. Supposing a
30% cutoff is desired, then all numbers in the array with values from one to three
are replaced with blanks and higher numbers with characters of increasing op
tical density. A 40% cutoff would be achieved by replacing the numbers from one
to four in the array by blanks. The eighth picture has a 40% cutoff also display
ing the patient's name, and other clinical data, making it convenient to send this
‘Acopy of the Fortran IV program will be provided upon request.
TABLE II
743 4 1_@*$ 4 — .4 +4+44+44 4+4+ ++ ,,+,,,•+••++ + 4+4, + + +4 6 + +4 4 + 4+ + •+.+. @xXxxxx. ... 4+++• .XXXXXXXXXXXA'+.+ •,XXXXXXXXXXXX(A+.4+#+ •#+6•+XXXX5@XXXXLX ‘.XX.. +,++AXXXL@**5iI5P((XX4+'4G' +#..xxAXXXt'I4SII4@X+@XX.+'4 ++,.XXXXXXI@5'5IPI51i5jXXLL4..#4' ,++.XXXXXXXXX(5i@@f.kX4XX++ •#+.XXX(X(XXIS'@I@IB@XXXX.+.# ....XXXXXXX(X@XX1455I5XLXX.+4# 4+. + X XX XX LXXX (XXX XXX (4 + 4+44 + ,, + xxx xg,g.x xx xxxx xXxx L+4' 4 ,,++XXXXXXXXXXXXXXXX/(++4'46 + +XXXXXXAXX (XXXXXXX (‘+4' ++.+X*XX+XXXxAXXXX(XXA++' +4+ ,,XXXXXXXXAXXXXXXXXXXXXX++4' 4+ , ix XX (X XX XX XX XXXX XXX' ‘+4+ +4 ‘4 ++X XX XX*X XmSWISX XXXX XX XX (44 4+ #4 XXXXXXXX IiI,I*II'BISXX XXX4 ‘ii 4+ XX' +XXZX1ISa5ftI44a$@xXX( 4 .+ ++ ++++XXXX@@*@*I$*5I5@b5I5X4*X4 +44 +,++XXXXXX5S5@S@I$II45SNU XX ‘@6 @ + + ##+.XXXXXXXXXX*.HIRS*XX(4++6+++ 4, ++ ix XX XX XSSlIiØ@PXX XX LX'' 4 ....+.XXXXXXXXXXXXXX•―'''+' + # + + X XX XXX X XXX XX (X4 +4' + +++4++++++++#, +4464+464 +,,+++#,,++ .4,,,,, +44+44+, +4 Fig. lB +4+4+4-44+4664 44 4+, ii +++4+ +++@ (LXXX. + 4 64+ 4+++X(XXXXXXXXX*+++##+4+ ++ . iL XXXX XL XX XXLX XXX# 4 + 4 4 4 +XXX XX X*ttS$X AXXXXX XX (4+ 6+ .+XXXXXX(IISISIISX/.XXXXXX#++# ++• +XXXXXA*@J5I54VJX(X(4++' .4+4 XXXX@W* I'5IS4*$@*@XL*XXX' + •,,+XXXXSI**SiId,SSIXXXX++++ +.,+XXXXI45@SXXXXXXXX46 ,.+,(XXXXS*iIS6I4IIIIXXXX+#++#' + +4+4 ,XXXX*@*eai@B@,5XXXX++'+ +4,. (XXXXX@54&4@WXXXXXX4++6 ... ,xxxx@5I@bft@as@*a(XXxXX+. #4#,XXXXLIWSIISI4SXXXXXX+444++ ,+#ixxXx5@I@5I1,55XXXLXX+44+6' 4+ 6,XXXXAX*I4ii5l@*id(XXX6+++ •+.+XXW@XX@5*,1IkXXXXXX464+ •4++++XX(XXX4/A5455I5@X(XXXXXA++++ 6 +4+ 4 4 +X (X(XII**lI,I6I,B@XLXXXX++4. 64 ,,,+,+++PXXX*$*,U151t135$XXXXXX+6+6 4,,+,+XXXXXXXX@4jII'IX(XXXX4+#++4ie ,+6+X(XXxXXX(XX((XX'.4+ 4+44 4++'46XLXXXxXXXXXXXXXX+6++4+ ++++++XAXXXXXXA((LX(++4+++ 6+6++4'AXXXXLXXX+44+6#@@6' + +4+4 + 4+4+44 XAXXX(4 64+4+ ,.4,,,XXXX*XXXXX#+4@+# 4 4, +4 6 XXII XXXXXXX+ #6+ +4+ + 44 Xx(XXXXX.,+#.' +e++++XLXXXXXX++++46 4,,,,, XXXX#+ +‘#+. +66+ XX (XXX +4+6+6 +64 #++#+#+++ +6,4+, ++++++
DIGITAL COMPUTER ANALYSIS AND DISPLAY OF RADIONUCLIDE SCAN
Fig. 1A
Fig. 1 (A and B). A lung scan performed with ‘311-macroaggregated human serum
albumin. The computed scan (B) shows more clearly than the photoscan (A) the area of
significantly reduced pulmonary artery perfusion corresponding to the patient's pulmonary
copy to the patient's ward for filing in his clinical record. Ideally, one would like
to have at least ten gradations of density in the characters used by the computer's
printer. Presently, however, we use only four gradations—a blank for the least
intensity of counts, next a “+“,then an “X―and fourth, an “H―with an “I―over
printed. This approach resembles the technique used by Ledley and Perry (9, 10).
Attempts to introduce more gradations of density using the printer of the com
puter have resulted in loss of clarity of presentation and usually present a more
confusing picture.
Tauxe (6), Winkler (3), and Kawin (2), all used X-Y plotters for the display
of their computed scans. We have experimented with this technique, but find
the amount of computer time necessary to draw these rather complicated plots
is much greater than that necessary to print the eight copies of the scan which
we presently use, and we prefer our pictures for clinical interpretation. Improved
methods of displaying the computer-processed radioisotope scan are under
development.
Several options, such as summing and subtracting areas, are available in
the program and are discussed below.
++++# #444 + •4++ 4+4+4+4+4+4+4 +44+++++••+ +4+ 4+44+44+ + 4+44+ 4+ +4,,,,,, 4+64+44 4@46+4+4 4+++ 4@+4+++ +++4,,, 4@++ 44 4++4+ 4+ 44 •••,.xx + +. x (*4 +++X*X XXXXXXX *XXXXXX ((xx' ‘4+ ‘+++++ XX #,,+XXXX(XXXXXXX(XXXXS@U(HIIIIP@bSXA(X(X6XXXX(XXXXXX*X++'+
Fig. 2. A 99mTc AP brain scan displayed in such a way that the computer shows there
DIGITAL COMPUTER ANALYSIS AND DISPLAY OF RADIONUCLIDE SCAN 745
RESULTS
One hundred consecutive computed scans performed on patients whose
radionuclide procedures were ordered on a routine clinical basis have been an
alyzed. These computed scans were compared to the photoscans which would
otherwise be used as the basis for interpreting the test. Both scans were also
correlated with the patient's clinical findings and other diagnostic procedures
(Tables I and II).
In two patients, the photoscan was judged to have more diagnostic informa
tion than the computed scan. There were five instances in which the photoscan
was technically superior to the computed scans. In four of these instances, this
superiority was because the computer had been instructed to process the scans
at the wrong speed and in one instance, a punching error had been made by the
tape perforator. Thirteen computed scans were superior in their diagnostic con
tent to the photoscans run simultaneously and six computed scans were superior
on technical grounds. It should be strongly emphasized, however, that the dif
ferences are generally slight. In no instance was there diagnostic information in
either the computed scan or the photoscan which was different enough to have
real clinical meaning and in which it could be said that clinical management
of the patient was altered. These results show clearly, however, that the com
puted scans have already achieved a diagnostic value equal to the photoscans
ordinarily used for interpretation of scanning procedures.
Fig. 3. Six of the eight pictures of a normal liver routinely displayed by the computer's
prmter with increasing cutoff and contrast enhancement. Note the diffuse homogeneity of
Perhaps the real goal of computer analysis of scans can be illustrated by
Figure 1, which shows studies from a patient with a right pulmonary embolus.
In interpreting the photoscan, the physician must guess whether the decrease
in radioactivity over the middle portion of the right lung is significant. However,
the computer has processed the data in such a way that it clearly shows a highly
significant decrease in intensity, thus helping remove subjectivity from the inter
pretation and decreasing the likelihood of error ( 11 ). To date, the organ most
successfully studied with computed scanning has been the liver. Figure three
shows the smooth homogeneous density throughout the body of a normal liver
printed out by the scanner. By contrast, Figure 4 shows a distinctly abnormal
patchiness in the center of a liver infiltrated with lymphoma.
One of the most promising aspects of digital computer analysis of radio
nuclide scans is the ability to quantitate the results of the test. It is a simple
matter, for instance, to instruct the computer to sum the total number of counts
over each lobe of the thyroid gland, over each kidney, or over one lung and
compare it to the other.
An application of this quantification of the data developed by Dr. Phillip
L. Dern of our Department of Medicine is presented in Figures 5 and 6. Scans
were obtained over the kidneys during constant infusion of 131I-ortho-iodohip
purate before and after urea washout. Normally, a significant dilution of the
counts occurs after the infusion of urea. However, in five patients with proven
•$•4+ 4+
-446+4+4+444+6+46
#4+ 44+0 4+ 4#+4++ 04+ 04 •4'++++$4'%.X##4#XX44+4++,++4 +6 +4 ++#+++XXXE(XXXX(XX++XX+++ 60 6+ •#4 + + ix xx * XXxX (LXXX XLXXX.( (XX' 4+ + 0'#+0+'.XxAXXX+4t@XxXX*((XL(XXXX+#.. 6+ +44+''XXXXXLX(I@5@l@$'jI5'XXXXXX/XLXXXXXt4+, 4+44+4 •.XX(x(LXXL*LL((XXXX((X@XXX@($-UXXXXXXXX(.,,,+6,+#, ‘4'+4L(X@AXfl@b4b-X((XXXXX(XX(LXXxXXLx(X,4,+64+#,t64 44 6+x(XXLXLXXX5X(Ii$@d$-5XXXX*4I4X*X(d@kJ(XXXXX(X(X6,f64+P+++4+##$.+6, @ #4++•++'++t+••# #+XXXX(XXL(X*th@k@XLL@'UHi@IBi04$f$54II1iXXXxXXL(%X/LXX4,,f+,++ 4+++X((XXXi45―fIthi4.'X,fb'+i$i$4IXX$I4bXXXXXXxXXXXX.#,$ ++ •.Xxxxxx(XxI@IiIxXxxa@o4s@o*nxAxxxXx%XX@.(xX.+4#.44++#o# XL##XXXX1fI@b+XXXxAX@X@bX*XXXX4+.,#+.$++f$++ 4$ +,XX++XXXXx(XX@XXXXAAXXXXXAXX*++++ ++f4XXXXX@(XLXXXXA++40XX'#++ + + .4 LXXXXxLX+4 (X4++ +4+4 + ••XXXXXXAX++I+XXXX+*++ •+XXXX.+t4+''@LXXXA 4+4++0 •$@ +4 44+4 +0Fig. 4. The computed scan of a liver diffusely infiltrated with lymphoma. Contrast the
patchy interior of this picture to the normal liver scans displayed in Figure 3. This ab
747
‘I‘4'XXSXX+ 4+4+ ,+,‘+,I.XxxXX+++ ‘IIlllulIIlIl,',+• I ++..XXSSX*XI++ I •.iXXXX,,1 1 ,+.+..•+ 1 +4 I +#X+@.+@ 1 0+ +4+ 1 “XXX....' 1 • ii +‘X*XX++++'14 4, 1.+..+.XXX... 1+4 4,, 14+4..+.++ 1 4' uiiliiiiiiiil 22222222222222+ + •2XXS$SX*X*+++2 4' ++.‘+2XSISSSXXXXX*2 +..“+X2XXOSX*XXXXXX2 •‘++“2XXXXXXXX*$XX2+++# +.++‘2++++XXXXXXX*2'+++ +4 2 +++XXXX*XX2+++++ 2 .,...XXX..2,...,, 2 •.+.+,,X..,2,.,,, 2 +4+4 4+4+624 22222222222222 44@ 4+4 64*44444' 44•+•4,4• 644 4*XJ@ •,,‘,. * *,,•,,,+,.4 464 •@ .4―. 22222222222222'―' 4' 2.**.―4....2, ••411.46' 4,.,,,,. 211 @ 24@•,4444@@, 2*******'X*―?―X *44416446' 6•@•@@ 2*****Z******26**―.―'4• ..‘..‘.‘..2SES*$*X**X**2**X'― 4. ‘X***2USSSSI****12'... “.6. •1S1S2**ESBSS*****2 “*6446 ‘**12**SI***$E**12'4*X4'4―4 644 ‘4•.**2**Ss•X*****12*XxXX•,., ‘ ‘4, 6.4'.'. 22222222222222*1 4* **1***.@46•'444 ,•.,,,,*X*.•.z.*1*****•,,,, •6 “4,,., ****X*•Z* *4444644 • +4444@41* @ **•+• ••4 4 444444444*14.. 4+44 AREASTOTALEDIN BLOCKS 0+ + + +4+ 4+ 4+4 ii44―+•+' ,,,.,,,.,,X,, +4@ @4X*44 + 4+0 444*i++.44..X... .4, 4, •.‘+ ‘111111111111114.4.. .4+ •@4+ 1.,.XXXX**.,,1,X,X*. 4*•4++ 1+.+**.'*XxX*I.*X...'. 4., 4'.,.. 1...x**'XX***i*SSX..Xx*•... 1..4X**X**XX.1,XX. .*S*,..*,. + + 4 ‘ ‘14' •XXXXXX*Xi*X1. ,*X*,44*4 • , @ +4 4+4444.41' •+X4.*S,SXII**―4..+ 44 ••“440••1+4.XX..ZSSXXI*X *4++•@@+ 44 4 •4+'*.*.' 1 *EX.*1X*X'―4―4 ++++‘.‘,.,,,+,,xi1ui1iIIi1111,*,,.,•,•' 4 +4 ,444•4•••X•'*X**SSS 4,, 4. ,,,••***z*S.••SXX “4,,., ***14*SSS*X***Z*4.4.―4'4 •, , *.‘4.XXNE$S*X X4'@4•4@44 + •@4•@•4@ 44*4+ •••*SUBXX*+' 4., +4 •4+444*4••4+.+4..++XX *44••4 *4 + 4, •44•+444•+4* 144 .4444*444. 4+4*4 4+ @+* 6• XI @4DIGITAL COMPUTER ANALYSIS AND DISPLAY OF RADIONUCLIDE SCAN
AREAS TOTALED IN BLOCKS
100 NICROCURIES OF 1—131+$IPPURANBEFORE UREA
HOSPITAL NC. 222111 10—19—65
AVERAGEBACKGROUND•1.6600
SPEED•45.00 CP.F$IN. SPACING•0.42 CM.
TIME INTERVAL •0.50 SEC.
THE TOTAL COUNTS PER SECONDABOVE BACKGROUNDBETWEENC+OARACTERS2?AND 40 AND LINES 12 ANO 22 IS——143$
THE TOTAL COUNTS PER SECOND ABOVE BACKGROUND BETWEEN CHARACTERSAO AND 93 AND LINES 20 AND 30 IS—— 176$
Fig. 5A
@N100 NICROCURIESOF 1—1)1HIPPURANAFTERUREA
HOSPITAl.NC. 222111 10—19—65
AVERAGE BACKGROUND •1.6600
SPIED•45.00CR.IMIN. SPACING•0.42CM. TIME iNTERVAL•0.30SEC.
THE TOTAl.COUNTSPER SECONDABOVEBACKGROUNDBETWEENCNARACTERS23ANO 36 AND LINES 9 AND 19 IS— 630 THE TOTAl. COUNTS PER SECOND ABOVE BACKGROUND BETWEEN CHARACTERS7A AND $9 AND LINES 17 AND 27 IS— 774
Fig. 5 (A and B). Renal scans performed during intravenous administration of l3l@tho.
iodohippurate before (A) and during (B) urea induced diuresis in a hypertensive patient
without renal disease. The radioactivity within the outlined rectangles has been quantitated,
renal artery stenosis, hyperconcentration of hippurate was demonstrated in the
involved kidney. That this hyperconcentration of hippuran persisted during urea
washout was vividly shown by the quantitation of counts per minute per square
centimeter over the kidneys.
Winkler (3) published studies in which he scanned a phantom liver and
pancreas gland and subtracted the liver scan from the pancreas scan, accentu
ating the localization of selenomethionine within the pancreas. We have had
the opportunity to perform this maneuver in several patients. These patients
were given 200 @xCof T5Se-selenomethionine and an area scan of the bed of the
pancreas performed. Following this procedure, 100 uC of 198Au colloid was given
intravenously and the same area rescanned with the pulse height and analyzer set
above the 280 keV peak of selenium-75. The two scans are then processed by the
computer in such a way that the liver scan is subtracted from the pancreas scan
after appropriate weighting. Despite the theoretical advantages of this technique,
display of the pancreas gland is still not adequate (Fig. 7). A great deal of ac
tivity always appears in the duodenum and jejunum. This problem is so promi
nent that it has not been possible to differentiate the pancreas from the intestinal
secretions. The method is still under evaluation, but results to date have made us
skeptical of the reported reliability of T5Se-selenomethionine as a scanning agent
for the pancreas gland (12, 13, 14).
Undoubtedly, many new uses for this quantitation of data from the scanner
will soon be forthcoming.
• IILIIILIUUU 4 4 3 •‘‘p I 4 3 3 •...x••....•.1. . 3 3 •1••,,,.,.,. *6@1.' • • 4 , 4 3 3 ..1'..'.''..*l.l' 4,',•,•4. * 3 Z... 1.... 3 3 ... @.•.. 3 3 ..i 3)333333333333) .IUL)3)LL)l1)3L .222222222222222 •. 2 **K2 •.‘‘.,2P @..*.*•1. *X*2 . . ..? £*@@Z.2 2..44XX4 S$SS•X2W.. BIBS SS.IS5Z,•.. 2.. :227222?222222?...
DRSNOGBAP SIFOWE USEA
HOSPITAL NC. 244441 9-22-65
AVERAGE WACKGHOUP.D•1.6600
SPEED •41.00 CP./MIN. SPACING . 0.4? CM. TIME INTERVAL - 0.3* SEC.
THE TOTAL COUNTS PER SECOND AEJVE NAC6000UND BETWEEN CHARACTERS)) AND 47 AND LINES 6 AND 15 15—- 770
THETOTAL COUNTS POW SECOND AB.JV, SACEGNO(JID BETWEENCH*TACTERSTI 6N13 03 AND LINES 13 AN')26 IS—- 1236 THE TOTALCOUNTS PEW SOLON') ABOVE BACE060UNO DETWEENCHAH4CTEHS 3 AND 1? 6'AO LISTS 665!) 13 IS—- 14
THE TOTAL COUNTS PES SECOND ABOVE BACE610UNO BUTWEON LOAVACTENSTI AN') SN AND LINES 2ANT 11 IS-- 33.)
749 44(64464446444', 4 4 4 4 4 4 4 4 4 +6 4 4 4 4 4 4 4 444444444444444 6*1466 •+64 222222722222222' 2. *Z****U.+.42. 646 ,.2,,****SIX*.++24.64 +26 ,+*S$SXIIIX2O+―+@46 •6 • ?,+,*IIIXISX*X2X*.X#64' .,,,,2**X*IISIT*XX12*'.6+' 4 ,,..2***XIIII*****24―4' .,**7IT*XXBIIXX**D2++6' ,..222Z22222222272'@4 ,,,,*,+*D*SI*X.***4+ •11I11111I111111'6 445 @+*+•46+V4U@sl,+* ‘+l+•64@46U64U6@L +66 1.D*.+O4.+.**41 •@+ I 4*VU*0,4.S***1•+#6W644 @+U40**+@@+,,4 14 ,.. .4.4 ,,..l..U.*A'UU'44'l U.. •4 4. I4@4U.606@U+@1U •4 1111111111111110――' .,**1•..•,•.,.,,,. 4.01*, •444A44U6•0 1676 2291 .4 636 Fig. 6C
DIGITAL COMPUTER ANALYSIS AND DISPLAY OF RADIONUCLIDE SCAN
El Fig. OB 0% $ S. a a I It 333333333333333 3 3 3 3 3 3 a 3 3 3 3 3 3 3 3 3 333333333333333
OERNOGR4P AFTEO UREA
HOSPITAL NC. 243541 9—22—65
AVERAGE OACKGMOUE.D •1.6600
SPEED • 45.0C CP./MI9. SPACING •0.42 CV.
TIME INTLRVAL - 0.30 SEC.
THE TOTAL COUNTS PER SECOND ABOVE BACKGROUND BETWEEN CHARACTERS31 AND 45 AND LINES GE AND 27 IS——
tHE TOTAL COUNTS PEE SECOND ABOVE BACKGROUND BETWEEN CHARACTERS72 AND 86 AND LINES 23 AND 32 IS——
THE TOTAL COUNTS PER SECONDABOVEBACKGROUNDBETWEENCHARACTERS3 AND 17 AND LINES 1$ AND 27 IS——
I I I I. S I I III I S 5 I I I I I 1 I a.J% I I II a ‘I? S I I I S a s P I S I I 5 I S I SI I I I I I a I I I I I S S I I S I I I 5 SI Fig. 6D
Fig. 6 (A, B, C, and D). Renal scans performed identically to those in Figure 5, but in
a patient with bilateral renal artery stenosis. This time quantitation reveals a significant hy
perconcentration in the more severely involved right kidney (A and B) and a failure of the
normal “washout―to occur with urea administration, bilaterally (C and D).
DISCUSSION
The IBM 7044 digital computer used in this project has a memory of 32,000
words. We have found that this size memory can be programmed to hold a grid
which is 133 characters wide and 100 lines long, the same size as that encom
passed by the 14 x 17 inch x-ray films generally used by scanners. Using this
size grid, the program with its options and an adequate intake-output buffer uses
all of the 32,000 word memory, except for 21 locations. The grid provides a reso
lution slightly less than half a centimeter which is in keeping with the best resolu
tion which can be obtained with crystals and collimators presently in use. The
program is designed to allow the right margin of the scan to be varied at will,
but because of technical considerations, the left margin is kept fixed.
By running the scanner at 30.5 cm per minute and the time interval of the
counters at 0.5 seconds with a spacing of 0.4 cm, computed scans are produced
which have the same size as the organ of the patient under study. We run our
thyroid scans at 15.25 cm per minute with a line spacing of 0.2 cm. The result
is an exact doubling of the size of the picture of the thyroid gland, which we
have found to be very convenient for study of this organ. With this technique,
the resolution is less than 0.25 cm, again better than that of the collimator. The
0 13
I I
I ISI
751
DIGITAL COMPUTER ANALYSIS AND DISPLAY OF RADIONUCLIDE SCAN
ME
Fig. 7A
LUI ‘lEE•No PE'LENO
..,.* .5. tI•A@ 55 I..SRUB. 2. 55156551 UU 55 VNS*S+V ,.S, •5 BBSa', NH.. U. VU SI S. .•.*I I B Fig. 7B a
Fig. 7 (A and B). A 73Se-selenomethionine pancreas scan (A) was performed followed
by a 198Au colloid scan. The computer then subtracted the liver from the pancreas scan, re
vealing intestinal rather than pancreatic concentration of the selenium (B).
program also allows scanning at 45 and 60 cm per minute when faster speeds
are needed. When running at 45 cm per minute, the appropriately sized organ is
pictured by averaging each consecutive pair of numbers and inserting the average
as a new data point between them. When running at 60 cm per minute each data
point is duplicated, thus producing double the number of data points actually
recorded. At the present time we are installing new equipment to allow recording
at speeds of 200 cm/min with a five-inch area scanner without necessitating this
duplication of data and avoiding the resulting loss of resolution.
The importance of correct X-Y orientation of the recorded data points with
respect to the patient is obvious. Our present equipment depends upon a con
stant speed of the scanner's motor to achieve this goal, but we have found that
our speeds vary as much as five per cent from line to line. This variation has not
produced serious distortions in the display of the organ under study, but we feel
that the use of a fixed time interval to control gating of the counter is probably
not the best method available. In our new equipment, gating will be controlled
by the distance traveled by the scanner's crystal instead. Variation in the scanner
motor speed will then result in a decrease or increase in counting rate which can
be tolerated better than a distortion in X-Y orientation. This system will also allow
a continuously variable instead of incremental setting of the scanning speed.
One of the great advantages of digital computer analysis of the radionuclide
scan is that most of the complicated electronic settings on the scanner are com
pletely bypassed, resulting in a great reduction in scans which must be considered
technically unsatisfactory. The ratemeter, cathode ray tube voltage, and count
rate differential settings are no longer necessary. Although a dot scan would
continue to be desirable for monitoring the procedure, it would be quite feasible
to build a scanner without installing any of these devices.
SUMMARY
In order to obtain greater objectivity in the interpretation of radionuclide
scans through the application of statistical testing, a new method of analysis
and display of the data available from the scanner using digital computer tech
niques is under development. Pulses from the pulse height analyzer of the
scanner are recorded on punched paper tape and are then processed by a digital
computer. Pictures of the organ under study are printed out on the computer's
printer. One hundred routine clinical scans studied in this way were correlated
with simultaneous photoscans and the patient's clinical findings. The results of
the computed scans were at least as good as those obtained with photoscanning.
A great advantage of computer processing of scans is the ability to quantitate
the results of scanning and thus provide information not previously available.
These developments are in their infancy and it is felt that their future is bright.
REFERENCES
1. KAWIN, B., HUSTON, F. V.: Dataphone/Computer System for Radioisotope Scan Dis
play. Proceedings of the 16th Annual Conference on Engineering in Medicine and Biology,
DIGITAL COMPUTER ANALYSIS AND DISPLAY OF RADIONUCLIDE SCAN 753
2. KAWIN, B., HUSTON, F. V., AND COPE, C. B. : Digital Processing/Display System for
Radioisotope Scanning, I. Nuci. Med. 5:500-514, 1964.
3. SCHEPARS,H., AND WINKLER, C. : An Automatic Scanning-System Using a Tape Per
forator and Con3puter Techniques, Proceedings of the Symposium on Medical Radioisotope
Scanning, Athens, Greece, April 20, 1964, Vol. 1, pp. 321-330.
4. BROWN, D. W., Digital Computer Analysis and Display of the Radioisotope Scan,
A scientific exhibit presented at the 11th Annual Meeting of the Society of Nuclear Medicine,
Berkeley, California, June 17-20, 1964. 5. Ibid, J. Nuci.Med.,5:802-806,1964.
6. TAUXE, W. N., CHAAPEL, D. W., Contrast-Enhancement of Scanning Procedures by
High-Speed Computer, I. Nuci. Med. 6:326, 1965, Abstract of a paper presented at the 12th
Annual Meeting of the Society of Nuclear Medicine, Bal Harbour, Fla., June 17-19, 1965.
7. GORTON,R. J., V. A. Hospital, Durhan3, N. C., Personal Communication.
8. Symposium entitled “Computers and Scanning―, New Orleans, La., Dec. 16-17, 1965.
9. LEDLEY, R. S., RUDDLE, F., MIGIC0N, B.: Computer Analysis of Chromosomes. Proc.
of the 16th Annual Conference on Engineering in Medicine and Biology, Baltimore, Md.
pp. 166-167, Nov. 18-20, 1963.
10. PERRY, B., MENDELSOHN, M. L., Picture Generation with a Standard Line Printer,
Comm. A.C.M., 7:311-313, 1964.
11. SUTHERLAND, J. L., DENARDO, G. L., BROWN, D. W.: 153 Lung Scans Performed
with 131J Labelled Macro-Aggregated Albumin, In press—Am. I. Roent.
12. SODEE,D. B.: Radioisotope Scanning of the Pancreas with Selenomethionine (Se-75)
Radiol. 83:910-916, 1964.
13. Bu@nmNE, J. A., HAYNIE, T. P.: Diagnosis of Pancreatic Carcinoma by Photoscanning.
J.A.M.A. 194:979-983, 1965.
14. TABERN, D. L., KEARNEY, J., DOLBOW, A.: The Use of Intravenous Amino Acids in