EANM practice guideline for quantitative SPECT-CT

h ps://doi.o g/10.1007/s00259-022-06028-9
GUIDELINES
EANM p ac ice guideline o quan i a i e SPECT‑CT
JohnC.Dickson1· IanS.A ms ong2· PabloMinguezGabiña3,4· AnaM.Denis‑Bacela 5· A onK.K izsan6·
Jona hanM.Gea 7· TimVandenWyngae 8,9· Lioe‑FeedeGeus‑Oei10,11· KenHe mann12
Recei ed: 19 Augus 2022 / Accep ed: 30 Oc obe 2022
© The Au ho (s) 2022
Abs ac 
Pu pose Quan i a i e SPECT-CT is a modali y o g owing impo ance wi h ini ial de elopmen s in pos adionuclide he apy
dosime y, and mo e ecen expansion in o bone, ca diac and b ain imaging oge he wi h he concep o he anos ics mo e
gene ally. The aim o his documen is o p o ide guidelines o nuclea medicine depa men s se ing up and de eloping
hei quan i a i e SPECT-CT se ice wi h guidance on p o ocols, ha monisa ion and clinical use cases.
Me hods These p ac ice guidelines we e w i en by membe s o he Eu opean Associa ion o Nuclea Medicine Physics,
Dosime y, Oncology and Bone commi ees ep esen ing he cu en majo s akeholde s in Quan i a i e SPECT-CT. The
guidelines ha e also been e iewed and app o ed by all EANM commi ees and ha e been endo sed by he Eu opean Asso-
cia ion o Nuclea Medicine.
Conclusion The p esen p ac ice guidelines will help p ac i ione s, scien is s and esea che s pe o m high-quali y quan i a-
i e SPECT-CT and will p o ide a amewo k o he con inuing de elopmen o quan i a i e SPECT-CT as an es ablished
modali y.
Keywo ds SPECT-CT· Quan i ica ion· Dosime y· Bone· Neu ology· Ca diology
P eamble
The Eu opean Associa ion o Nuclea Medicine (EANM) is
a p o essional non-p o i medical associa ion ha acili a es
communica ion wo ldwide among indi iduals pu suing clin-
ical and esea ch excellence in nuclea medicine. The EANM
was ounded in 1985. These guidelines a e in ended o assis
p ac i ione s in p o iding app op ia e nuclea medicine ca e
o pa ien s. They a e no in lexible ules o equi emen s
o p ac ice and a e no in ended, no should hey be used,
o es ablish a legal s anda d o ca e. The ul ima e judgmen
ega ding he p op ie y o any speci ic p ocedu e o cou se
o ac ion mus be made by medical p o essionals aking in o
accoun he unique ci cums ances o each case. Thus, he e
is no implica ion ha an app oach di e ing om he guide-
lines, s anding alone, is below he s anda d o ca e. On he
This a icle is pa o he Topical Collec ion on Miscellanea
* Ken He mann
Ken.He [email protected]
1 Ins i u e o Nuclea Medicine, Uni e si y College London
Hospi als Founda ion T us , London, UK
2 Nuclea Medicine, Manches e Uni e si y NHS Founda ion
T us , Manches e , UK
3 Depa men o Medical Physics andRadia ion P o ec ion,
Gu u ze a-C uces Uni e si y Hospi al/Bioc uces Heal h
Resea ch Ins i u e, Ba akaldo, Spain
4 Depa men o Applied Physics, Facul y o Enginee ing,
UPV/EHU, Bilbao, Spain
5 Na ional Physical Labo a o y, Tedding on, UK
6 ScanoMed Nuclea Medicine Cen e s, Deb ecen, Hunga y
7 Join Depa men o Physics Ins i u e o Cance Resea ch
andRoyal Ma sden, NHS Founda ion T us , Su on, Su ey,
UK
8 Depa men o Nuclea Medicine, An we p Uni e si y
Hospi al, Edegem, Belgium
9 Facul y o Medicine andHeal h Sciences (MICA – IPPON), ,
Uni e si y o An we p, Wil ijk, Belgium
10 Depa men o Radiology, Sec ion o Nuclea Medicine,
Leiden Uni e si y Medical Cen e , Leiden, TheNe he lands
11 Biomedical Pho onic Imaging G oup, Uni e si y o Twen e,
Enschede, TheNe he lands
12 Depa men o Nuclea Medicine, Uni e si y
o Duisbu g-Essen, andGe man Cance Conso ium
(DKTK)-Uni e si y Hospi al Essen, Essen, Ge many
/ Published online: 5 Decembe 2022
Eu opean Jou nal o Nuclea Medicine and Molecula Imaging (2023) 50:980–995
1 3
con a y, a conscien ious p ac i ione may esponsibly adop
a cou se o ac ion di e en om ha se ou in he guidelines
when, in he easonable judgmen o he p ac i ione , such
cou se o ac ion is indica ed by he condi ion o he pa ien ,
limi a ions o a ailable esou ces o ad ances in knowledge
o echnology subsequen o publica ion o he guidelines.
The p ac ice o medicine in ol es no only he science bu
also he a o dealing wi h he p e en ion, diagnosis, alle-
ia ion and ea men o disease.
The a ie y and complexi y o human condi ions make i
impossible o always each he mos app op ia e diagnosis
o o p edic wi h ce ain y a pa icula esponse o ea -
men . The e o e, i should be ecognised ha adhe ence o
hese guidelines will no ensu e an accu a e diagnosis o a
success ul ou come. All ha should be expec ed is ha he
p ac i ione will ollow a easonable cou se o ac ion based
on cu en knowledge, a ailable esou ces and he needs o
he pa ien o deli e e ec i e and sa e medical ca e. The
sole pu pose o hese guidelines is o assis p ac i ione s in
achie ing his objec i e.
In oduc ion
In nuclea medicine, gamma came as and SPECT-CT sys-
ems a e ou inely used o quan i a i e imaging. F om de e -
mining ela i e kidney pe o mance, o binding a ios in he
b ain, one o he s eng hs o gamma came a imaging is i s
abili y o quan i y in- i o physiology o a wide ange o
condi ions and applica ions. While adi ional gamma cam-
e a imaging has ocused mos ly on ela i e up ake o he
adiopha maceu ical, o example o he opposing kidney,
o speci ic s uc u es in he b ain, PET imaging has ocused
on absolu e quan i ica ion in o gans o ea u es, p oducing
images wi h quan i a i e uni s such as kBq/mL o s anda d-
ised up ake alue (SUV). This has b ough ad an ages. The
abili y o pe o m absolu e quan i ica ion, wi h 18F-FDG o
example, p o ides he oppo uni y o assess he me abolic
s a us o he disease o diagnosis, s aging, ea men moni-
o ing and disease p og ession.
Fo many yea s, SPECT-CT imaging has been pe cei ed
o be he poo ela i e o PET-CT o quan i a i e imaging.
Howe e , equipmen and so wa e de elopmen s including
he inco po a ion o measu ed CT-based a enua ion co -
ec ion, sca e co ec ion and co ec ion o pa ial olume
e ec s, ha e made signi ican p og ess in imp o ing SPECT-
CT quan i ica ion [1]. Using mode n echniques, quan i ica-
ion o SPECT-CT da a is now possible in a simila way as i
is in PET-CT. Howe e , quan i a i e SPECT-CT o e s many
ad an ages o e PET-CT, which has he po en ial o lead o
a wide ange o applica ions. SPECT uses longe physical
hal -li e adiopha maceu icals ha can be e ma ch biologi-
cal p ocesses. While longe hal -li e adionuclides such as
64Cu, o 124I a e a ailable in PET, hese ha e in e io imaging
cha ac e is ics in e ms o hei posi on emission p obabili y
and high e ec i e dose [2, 3], which diminish some o he
ad an ages o PET imaging. A u he ad an age o SPECT
is ha i s adiopha maceu icals can be labelled wi h di e -
en adionuclides, o e ing he possibili y o image mul iple
physiological p ocesses a he same ime. S udies wi h CZT
SPECT sys ems a e showing he abili y o pe o m simul a-
neous 99mTc/123I imaging [4], which, o example, could be
applied in simul aneous pe usion/inne a ion s udies in he
hea , and pe usion/DaT (dopamine anspo e ) a ailabili y
in he b ain. Bu i is he widesp ead a ailabili y o SPECT-
CT and he wide ange o accompanying adiopha maceu-
icals which p o ides he g ea es ad an age o quan i a i e
SPECT-CT o e PET-CT, wi h much wide global ma ke
pene a ion due o he lowe cos o imaging equipmen , and
he s aigh o wa d adiopha maceu ical p oduc ion.
Al hough a an ea ly s age, he applica ions o quan i a i e
SPECT-CT a e becoming clea e . The d i ing o ce in he
inc easing in e es in quan i a i e SPECT-CT comes om
he g ow h in adionuclide he apy and he anos ics, and
he associa ed g ow h in pe sonalised ea men planning.
The unique ad an age o many nuclea medicine he apies
employing gamma-emi ing he apeu ic adionuclides is ha
hey allow us o image and quan i y he adiopha maceu ical
bio-dis ibu ion a e adminis a ion by SPECT-CT imag-
ing. Because o his, in e nal abso bed dose calcula ions o
adionuclide he apies using cu en commonly used he a-
peu ic adionuclides such as 177Lu and 131I [5, 6] p esen he
oppo uni y o e i y ea men deli e y and/o pe sonalise
he ea men o e he cou se o mul i-cycle adminis a ions.
Fu he mo e, imaging su oga es used wi h quan i a i e
SPECT-CT can be used o diagnosis and pa ien selec ion
p io o adionuclide he apy, in addi ion o assessmen o
disease p og ession [7, 8] o ea men esponse. Applica-
ions in he quan i ica ion o bone ace s o o hopaedic
[9, 10] and cance applica ions [11] a e de eloping, while
oppo uni ies in ca diac [12] and neu ological [13] imaging
also exis .
Gi en hese ea ly, bu apidly de eloping applica ions,
i is i al ha in eg a ion o such echnology in o clinical
p ac ice is pe o med co ec ly. This is necessa y in e ms
o co ec image in e p e a ion and o acili a e he pool-
ing o da a o imp o e scien i ic knowledge. Fu he mo e,
o success ul adop ion, i is necessa y ha echniques a e
obus , consis en and co ec ly unde s ood by he use o
hei applica ion. The e mus also be clea applica ions o
he echnology. Fo example, quan i a i e SPECT-CT may
enable a be e unde s anding o why a ce ain subg oup o
pa ien s does no bene i om adionuclide he apy despi e
adequa ely exp essing he a ge . And in he de elopmen
o new adionuclide he apies wi h a na owe he apeu ic
index, quan i a i e SPECT-CT has he po en ial o suppo
981Eu opean Jou nal o Nuclea Medicine and Molecula Imaging (2023) 50:980–995
1 3
clinical de elopmen in iden i ying he op imal ac i i y o
be adminis e ed o each pa ien . The inc emen al alue o
using quan i a i e SPECT-CT o clinical ca e needs, how-
e e , s ill needs o be demons a ed. A de ini ion o how o
app op ia ely pe o m quan i a i e SPECT-CT is he e o e
essen ial o p o ide a s able pla o m o he unde s anding
and applica ion o his echnology.
These guidelines ha e been w i en wi h he objec i e o
de ining s anda ds o quan i a i e SPECT-CT. Re iewing
he cu en s a us o quan i a i e imaging in SPECT-CT and
highligh ing limi a ions and a eas in need o de elopmen ,
we demons a e whe e quan i a i e imaging is po en ially
mos bene icial. In hese guidelines, we desc ibe he p oce-
du al conside a ions ha mus be me in bo h achie ing and
implemen ing quan i a i e measu emen s o SPECT imag-
ing and p o ide clinical applica ions o which i s implemen-
a ion has al eady been success ul.
Imaging p o ocols o quan i a i e SPECT‑CT
O e iew
The scope o his sec ion is o p o ide echnical ecommen-
da ions on image acquisi ion and econs uc ion o achie e
eliable quan i a i e SPECT-CT images. They a e in ended
o be used wi h gamma came as wi h con en ional plana
de ec o s o a ing a ound he pa ien du ing he SPECT
acquisi ion. O he gamma came a designs such as hose wi h
solid s a e de ec o s o no el non-pa allel collima ion a e
no co e ed wi hin hese guidelines, al hough some o he
concep s desc ibed a e s ill ele an .
The p ima y aim o quan i a i e SPECT-CT is o p oduce
a omog aphic image wi h oxel alues ep esen ing ac i i y
concen a ion. The u ilisa ion o hese oxel alues will hen
be dependen on he clinical applica ion. Fo adionuclide
dosime y ( o diagnos ic and he apeu ic applica ions), he
absolu e ac i i y in Becque els (Bq) wi hin a delinea ed
olume o o gan is ex ac ed om he images, and hence
accu acy is pa amoun . Fo o he diagnos ic applica ions,
oxel alues may be con e ed o s anda dised up ake alues
(SUV). In his la e example, he SUV should ansla e o a
clinically ele an bioma ke and a guably he ep oducibil-
i y o his me ic is mo e ele an han i s absolu e accu acy.
This is pa icula ly pe inen o ollow-up s udies moni o ing
disease esponse o ea men s.
While ac i i y concen a ion o dosime y is almos
always de ined in a olume ep esen ing an o gan o can-
ce ous deposi , SUV can be measu ed using se e al me ics.
Like in dosime y, he mean alue o SUV (SUVmean) in
a ea u e can be de ined. Howe e , SUVmax is mo e com-
monly used as a be e ep esen a ion o he in ensi y o
up ake in a ea u e, e en wi h i s limi a ion o being elian
on a single pixel alue and he e o e being mo e suscep i-
ble o image noise a ec ing bo h he bias and p ecision o
he measu e. SUVpeak is an al e na i e measu e o quan i-
a i e SPECT, cap u ing he s a us o me abolically ac i e
ea u es while mi iga ing image noise [14, 15]. Gi en ha
i can mi iga e o di e ences in image quali y and con as
eco e y be ween imaging sys ems i is also well-sui ed o
mul i-cen e s udies [16]. All h ee SUV me ics a e used in
clinical p ac ice depending on he objec i e o he measu e-
men , bu i is impo an o unde s and he ad an ages and
he limi a ions o each SUV me ic.
The e a e many image- ela ed ac o s, bo h con ollable
and uncon ollable, ha mus be conside ed when pe o m-
ing op imisa ion o quan i a i e SPECT-CT. Mos o hese
ac o s will lead o nega i e bias (an unde es ima ion o he
ac ual ac i i y concen a ion) in he quan i a i e alues while
o he s will lead o posi i e bias (an o e es ima ion o ac i i y
concen a ion). Assuming ha econs uc ion co ec ions,
such as measu ed a enua ion and sca e , and a obus c oss-
calib a ion be ween he adionuclide calib a o and gamma
came a a e pe o med co ec ly, nega i e bias is la gely due
o pa ial olume e ec s om limi a ions in spa ial esolu-
ion. Posi i e bias is mainly a ibu ed o Poisson noise, a is-
ing om ela i ely low pixel alues due o he cons ain s o
scan ime and pa ien adminis e ed ac i i y, bu may also be
associa ed wi h econs uc ion a e ac s in some si ua ions.
Some echnical ac o s ha will in luence hese nega i e and
posi i e bias ac o s a e gi en in Table1.
The op imisa ion p ocess o quan i a i e SPECT-CT mus
e alua e he in luence o he con ollable a iables on he
accu acy and ep oducibili y o image-de i ed measu e-
men s. Fu he mo e, he pa ame e s should be chosen such
ha hey minimise, as a as possible, he in luence o he
non-con ollable ac o s. I is essen ial ha his is cha ac e -
ised using sui able phan om da a, which will allow e o s in
he measu emen o be de e mined.
This sec ion desc ibes he conside a ions needed om
bo h he image acquisi ion and econs uc ion p ocesses in
achie ing eliable quan i a i e da a om SPECT-CT images.
I should be highligh ed ha cen es in ending o pe o m
quan i a i e SPECT-CT should ha e sui able equipmen
o assess sys em pe o mance wi h a minimum o uni o m
phan om o assess calib a ion and an IEC image quali y
(IQ) Phan om, a Jaszczak Phan om o simila o cha ac e -
ise ac i i y concen a ion eco e y coe icien s (ACR) o
a ious objec sizes.
Scanne calib a ion andcha ac e isa ion
The di e en ia ing ac o be ween adi ional SPECT-CT
and quan i a i e SPECT-CT is he sensi i i y calib a ion
ac o . Hence, he i s s ep is o de e mine a eliable c oss-
calib a ion ac o be ween a adionuclide calib a o and he
982 Eu opean Jou nal o Nuclea Medicine and Molecula Imaging (2023) 50:980–995
1 3
gamma came a. The exac echnique may di e depending
on he sys em manu ac u e and he adionuclide being
used.
A p e equisi e o scanne calib a ion is a obus meas-
u emen o ac i i y in a adionuclide calib a o . Ac i i ies
mus be aceable o na ional and in e na ional s anda ds.
In e na ional guidelines ecommend accu acies o 5–10%
o diagnos ic, and 5% o he apeu ic adionuclides [17,
18]. The use o ac i i y measu emen s aceable o p ima y
s anda ds is no common p ac ice in all coun ies [19], e en
hough la ge a iabili ies ha e been obse ed in in e na-
ional compa ison exe cises [20]. The e o e, adionuclide
calib a o s should be egula ly main ained and calib a ed
agains a p ima y s anda d o he adionuclide and measu e-
men geome y o in e es ollowing a ailable good p ac ice
guidelines [21, 22].
The e a e wo dis inc conside a ions ha will in luence
he accu acy o quan i a i e da a ex ac ed om econ-
s uc ed SPECT-CT images. These a e he ac i i y calib a-
ion and cha ac e isa ion o he sys em. I is impo an o
app ecia e he di e ence in hese concep s. Calib a ion o a
sys em is he ac o including he ac i i y calib a ion ac o
o he sys em o p oduce ac i i y concen a ion measu e-
men s in Bq/mL. Cha ac e isa ion will desc ibe he pe -
o mance o a gi en measu emen echnique, o example
SUVmax, de i ed om images ob ained om a speci ic
combina ion o acquisi ion and econs uc ion pa ame e s.
Bo h will be impac ed by he acquisi ion and econs uc ion
pa ame e s chosen as has been desc ibed in Table1.
Clea ly, a key componen o quan i a i e SPECT-CT is
he accu acy o he calib a ion ac o ela ing econs uc ed
coun s om he scanne o ac i i y measu ed in a adionu-
clide calib a o whose measu emen s should be aceable o
a p ima y s anda d. E o s in his calib a ion ac o will lead
o sys ema ic bias in quan i a i e alues de i ed om he
images. I is impo an o app ecia e ha his is a wo-s age
p ocess: one o calib a ion, and ano he o e i ica ion o
quan i a i e accu acy ollowing he calib a ion p ocess. The
calib a ion p ocess i sel will a y acco ding o he manu-
ac u e ’s equi emen s, which may ange om a pe i-dish
sou ce plana measu emen ; a long-li ed sealed poin sou ce
plana measu emen ; o olume ic measu emen s om a
la ge uni o mly illed phan om acqui ed, econs uc ed and
co ec ed using he same pa ame e s used o pa ien imag-
ing. Rega dless o he me hod, i is ecommended o pe o m
his calib a ion using he me hod and wi h he equency
sugges ed by you sys em manu ac u e . I no ecommenda-
ions a e gi en, he calib a ion should be pe o med a leas
annually, o a e any majo changes o ha dwa e o so wa e
using a olume ic app oach gi en ha i be e ep esen s
imaging condi ions [23, 24]
Once calib a ion has been pe o med, i mus be e i-
ied be o e clinical use. This should be pe o med using a
la ge uni o m olume ic phan om illed wi h he app op ia e
Table 1 Con ollable and uncon ollable ac o s ha in luence he quan i a i e accu acy o ac i i y concen a ion measu emen s ob ained om
SPECT-CT images
Fac o Con ollable? Impac o quan i a i e accu acy
Adminis e ed ac i i y Some imes Fo SUVmax and o he egions based upon h eshold o SPECT oxel alues, posi i e
bias can occu due o changes in noise le el
No e ha geome ic egions, manually delinea ed on whole o gans o lesions whe e
all encompassed SPECT oxel alues a e a e aged, a e less suscep ible o a ia-
ions in image noise le els
I should be no ed ha o diagnos ic applica ions, he adminis e ed ac i i y is de ined
locally while, o he apeu ic applica ions, i is likely o be ei he a ixed ac i i y o
de ined by he planned he apy abso bed dose o he pa ien
Acquisi ion ime Yes Posi i e bias om image noise—see commen s abo e ega ding adminis e ed ac i i y
Collima o Yes Nega i e bias due o deg ading spa ial esolu ion
Ma ix size Yes Nega i e bias due o changes in spa ial sampling
Posi i e bias due o changes in noise
SPECT o bi adius Yes Nega i e bias due o deg ading spa ial esolu ion o la ge adii
Numbe o upda es (p oduc o i e a ions
and subse s) o i e a i e econs uc ion
Yes Nega i e bias due o unde -con e ged image
Posi i e bias due o image noise
Pos - econs uc ion smoo hing il e Yes Nega i e bias due o addi ional image blu ing
Posi i e bias due o con ol o image noise
Dead- ime No Nega i e bias due o dead- ime e ec s a e y high coun - a e le els
Lesion/o gan size and shape No Nega i e bias due o deg ading spa ial esolu ion o small olumes
O gan- o-backg ound con as No Va iable due o changes in spill-o e om su ounding ac i i y
O gan loca ion No Nega i e bias due o inc easing dis ance om he de ec o
Pa ien mo emen No Nega i e bias due o inc easing image blu ing
983Eu opean Jou nal o Nuclea Medicine and Molecula Imaging (2023) 50:980–995
1 3
adionuclide and acqui ed on he SPECT-CT wi h a clini-
cal imaging p o ocol. I he ini ial calib a ion is pe o med
using a olume ic me hod, he e i ica ion scan should no
be pe o med immedia ely a e he calib a ion using he
same phan om as any e o s in he ac i i y measu emen
o phan om illing will simply ans e ac oss o he e i-
ica ion scan. The e i ica ion should be pe o med wi h a
eshly illed phan om o allow a es o he en i e p ocess o
ac i i y measu emen , phan om illing and image acquisi-
ion. The impo ance o es ablishing a good echnique o
measu ing ac i i y in he adionuclide calib a o should no
be neglec ed and should mi o he measu emen app oach
used o pa ien s. I pa ien injec ions a e pe o med in spe-
ci ic sy inge sizes and dilu ed o a gi en olume, his should
be eplica ed when measu ing he ac i i y o he calib a ion
and e i ica ion phan om scans. The ou come o e i ica ion
is o ensu e ha he measu ed ac i i y concen a ion ag ees
wi h he ue ac i i y concen a ion.
Once a sys em has been calib a ed, i is essen ial o cha -
ac e ise he ela ionship be ween objec size and measu e-
men accu acy. This ela ionship is commonly demons a ed
in he o m o ac i i y concen a ion eco e y (ACR) cu es,
which a e commonly de i ed in PET-CT [25]. These meas-
u emen s a e pe o med using phan oms, such as he NEMA
IEC IQ o Jaszczak phan oms, wi h a ep esen a i e ange o
illable olumes. Reco e y cu es a e speci ic o he con as
and measu emen o ac i i y concen a ion ha has been pe -
o mede.g., SUVmax, SUVmean and hence ca e should be
aken o ensu e clinically ele an measu emen s a e aken
when cha ac e ising he sys em. Figu e1 shows an example
o an ACR cu e.
Acquisi ion
Acquisi ion pa ame e s can a y depending on he pa ien
in es iga ion being pe o med and, on he cha ac e is ics o
he came a. Each cen e should he e o e op imise i s own
acquisi ion pa ame e s. Some s eps ha should be ollowed
in he acquisi ion p ocess a e:
• S eps should be aken o limi he possibili y o pa ien
mo ion. I is impo an ha he pa ien emains in he
same posi ion du ing bo h he CT and SPECT acquisi-
ion o ensu e good image egis a ion and accu a e CT
a enua ion co ec ion.
• The op imal collima o will depend on he adionuclide
being imaged. Rele an imaging guidelines should be
ollowed when choosing an app op ia e collima o .
• S ep and shoo o con inuous acquisi ion mode o acqui-
si ion can be used. The la e can o e a 1–2min sa ing
on scanning ime o e 60 o a ion angles.
• De ec o au o-con ou ing is ad ised o minimise he
dis ance be ween he de ec o s and pa ien o p o ide
op imal spa ial esolu ion. Howe e , o some applica-
ions de ec o s can be kep a a ixed bu close dis ance.
• Acquisi ion should ypically be pe o med wi h oppos-
ing de ec o s a 180° om one ano he , bu o ca diac
applica ions a 90° con igu a ion may be used.
• A pixel size smalle han hal he ull wid h a hal
maximum (FWHM) spa ial esolu ion o he sys em
o he adionuclide used is ecommended o ensu e
app op ia e spa ial sampling. Commonly, a ma ix size
o 128 × 128 is used. I should be no ed ha dec easing
he pixel size esul s in a noisie image.
• The numbe o p ojec ions is ecommended o be
simila o he ma ix size (e.g. 120–128 p ojec ions
o a 128 × 128 ma ix) o ensu e app op ia e angula
sampling. Ca diac applica ions using a 90° de ec o
con igu a ion may use a educed a c and numbe o
p ojec ions, al hough dis o ion and inaccu a e quan i i-
ca ion will occu when he e a e insu icien p ojec ion
da a— ypically dis an om he hea .
• The ime pe p ojec ion will depend on he amoun
o adioac i i y in he pa ien . As noise in he p ojec-
ion da a ollows a Poisson dis ibu ion, and in econ-
s uc ed da a is much wo se [26], imaging ime mus be
high enough o educe image noise as much as possible.
I mul iple ields o iew (FOVs) a e acqui ed, he ime
pe p ojec ion may ha e o be dec eased o pa ien
com o .
Fig. 1 Example o an ac i i y concen a ion eco e y (ACR) cu e
ob ained om he six ho sphe es o a NEMA IEC image quali y
phan om illed wi h 10:1 con as o backg ound. In his example,
ac i i y concen a ion was measu ed by he maximum oxel alue,
which is akin o SUVmax
984 Eu opean Jou nal o Nuclea Medicine and Molecula Imaging (2023) 50:980–995
1 3

Recons uc ion
I e a i e me hods a e ecommended o econs uc he
acqui ed SPECT p ojec ions. No mally, he algo i hm
used will be ha included in he so wa e p o ided by he
endo o he gamma came a; howe e , hi d-pa y algo-
i hms a e also a ailable. Fo quan i a i e pu poses, he
numbe o upda es, de ined by he p oduc o he numbe
o i e a ions and subse s, may be g ea e han o econ-
s uc ions wi h quali a i e pu poses [27]. P e e ably, ha
numbe should be ob ained om phan om measu emen s
in which he con e gence o ac i i y concen a ion is s ud-
ied, paying a en ion no o each excessi e noise le els o
in oduce image a e ac s, such as Gibbs a e ac s om
he use o esolu ion modelling. Ideally, such op imisa ion
should be pe o med using an h opomo phic phan oms
which mimic he clinical si ua ion, bu such phan oms
a e no always a ailable. I NEMA IEC IQ o Jaszczak
phan oms a e used as pa o he op imisa ion p ocess, i
should be no ed ha SPECT eco e y cu es a e a mo e
dependen on sphe e posi ion han PET. This is likely due
o he a iable de ec o adius acquisi ion esul ing om
au o-con ou ing [28] and so i is ecommended ha mul-
iple sphe e con igu a ions a e e alua ed. Figu e2 shows
an example o how he a iabili y o ACR can be aligned
ac oss di e en sphe e con igu a ions when su icien
upda es a e pe o med.
Co ec ions
A enua ionco ec ion
A enua ion co ec ion based on CT da a should be used
o quan i a i e SPECT-CT. The CT images should be con-
e ed o a map o a enua ion coe icien s o he app o-
p ia e adionuclide ene gy and inco po a ed in o he i e a-
i e econs uc ion p ocess [29]. I mus also be possible
o measu e he ull pa h leng h o he gamma- ay om he
Fig. 2 NEMA IEC image qual-
i y phan om illed wi h 99mTc a
10:1 con as in h ee di e en
sphe e con igu a ions econ-
s uc ed wi h 3 i e a ions and
6 subse s (uppe se o images)
and 20 i e a ions and 6 subse s
(lowe se o images). The co -
esponding eco e y cu es a e
gi en below o maximum oxel
ACR. No e ha all phan om
images shown ha e co ec ions
o a enua ion, sca e and col-
lima o esponse included and a
10mm Gaussian il e applied
985Eu opean Jou nal o Nuclea Medicine and Molecula Imaging (2023) 50:980–995
1 3
poin o emission o he pa ien bounda y on he a enua ion
map. I he a enua ion map is unca ed, ei he by way o
la ge pa ien habi us o inadequa e CT ield o iew, hen he
a enua ion co ec ion will no be accu a e and quan i a i e
e o s will occu . Hence, ca e should be aken when posi-
ioning he pa ien o minimise he likelihood o unca ion.
Whe e app op ia e, and when a ailable, i is ecommended
o use me al a e ac co ec ion algo i hms [30] o p oduce
mo e sui able a enua ion maps [31].
Sca e co ec ion
To co ec o sca e ed gamma- ays p esen wi hin he pho-
opeak window, mul iple ene gy window sca e co ec ion
me hods a e ypically applied, al hough model-based sca e
co ec ion can also be used i a ailable [32]. Dual-ene gy
window me hods can be applied i he e a e no emissions
abo e he pho opeak o he adionuclide used; o he wise,
a iple-ene gy window me hod should be implemen ed.
Smoo hing o he sca e window image may also be bene i-
cial o educe p opaga ion o image noise om he co ec-
ion o he econs uc ed image. I is impo an o alida e
sca e co ec ion echniques using app op ia e phan oms
con aining a eas o no ac i i y, su ounded by uni o m ac i -
i y, o demons a e ha he algo i hms do no o e -co ec
he inal images.
Resolu ion modelling
Resolu ion modelling is a ailable in mos mode n econ-
s uc ion so wa e and pa ially compensa es o he limi ed
spa ial esolu ion due o he collima o and he de ec o by
inco po a ing a dep h-dependen collima o esponse model
in he p ojec ion ope a ion o he i e a i e econs uc ion
me hod [33]. I is ecommended ha esolu ion modelling
echniques a e used i a ailable o imp o e quan i ica ion.
Decay co ec ion
Unde s anding how and when decay co ec ion is applied is
impo an in quan i a i e SPECT-CT. Gi en he ela i ely
long physical hal -li e o mos SPECT adionuclides, i s
applica ion o ensu e di e ences a e accoun ed o in he
acquisi ion o he i s and las p ojec ion a e ela i ely
mino . One excep ion being 81mK SPECT-CT whe e di e -
ences in acqui ed p ojec ion coun s will be la ge and appli-
ca ion o such co ec ions essen ial. In mul iple SPECT ield
o iew s udies whe e he s udy may ake up o 1h, decay
co ec ion should also be pe o med o ensu e consis ency o
ela i e pixel alues ac oss all acqui ed p ojec ions.
A second a ea whe e decay co ec ion is impo an is in
he measu emen o SUV me ics. Wi h SUV no malising
up ake o injec ed ac i i y, i is impo an ha SPECT da a
acqui ed se e al hou s o e en days la e accoun s o he
decay (especially o adionuclides such as 177Lu). Ensu ing
clocks on scanne s and in injec ion ooms a e consis en is
an impo an elemen o ensu ing his is done app op ia ely.
Pa ial olume (sphe e andnon‑sphe e)
Resolu ion modelling will a ely ully compensa e o he
limi ed spa ial esolu ion o he gamma came a. Hence, o
quan i y he ac i i y in he olumes o in e es (VOI), pos -
econs uc ion co ec ions can be applied. These eco e y
coe icien s a e commonly ob ained om phan oms wi h
sphe ical inse s o known olume ha a e illed wi h known
ac i i y concen a ions [23]. No e ha he applicabili y o
his me hod has limi a ions o non-sphe ical olumes. Fo
some ana omical egions (e.g. b ain), me hods based on ana-
omical in o ma ion such as he geome ic- ans e ma ix
me hod can be applied [34]. While pa ial olume co ec-
ions a e no gene ally a ailable wi h comme cial scanne
so wa e and he e o e no clinically a ailable, ‘homemade’
si e applied co ec ions can be use ul o dosime y, and
some esea ch applica ions. No e ha o clinical use, any
si e w i en so wa e should be in alignmen wi h app op ia e
medical de ice legisla ion [35].
Dead‑ ime
Fo acquisi ions pe o med soon a e adminis e ing he a-
peu ic amoun s o adiopha maceu icals, e.g. ea men o
neu oblas oma wi h 131I-mIBG o 177Lu e ium pep ide he -
apy o neu oendoc ine umou s, dead- ime co ec ion may
need o be applied [36, 37]. The non-linea i y o he coun
a e will no be he same o e e y p ojec ion angle. The e-
o e, he dead- ime co ec ion should p e e ably be applied
on each p ojec ion acqui ed [38]. Howe e , o simplici y,
an a e age co ec ion could be applied based on he coun
a e a e aged o e all p ojec ions.
Recons uc ion pos ‑ il e ing
The applica ion o a pos - econs uc ion smoo hing il e
will ine i ably deg ade he spa ial esolu ion and exagge -
a e pa ial olume e ec s. I is he e o e o en assumed ha
il e ing should no be applied o images o quan i a i e
applica ions. Howe e , he applica ion o a pos - il e is
in ended o con ol he deg ee o noise in he image and
hence he po en ial posi i e bias ha may a ise om noise.
The applica ion and choice o a pos - econs uc ion il e is
dependen on he desi ed ou come om he image and na u e
o he measu emen . Insi ua ions whe e la ge egions a e
d awn, such as o gan delinea ion, and whe e mean ac i -
i y concen a ion measu emen s a e de i ed om all ox-
els, hen a pos - il e is unlikely o be bene icial. Howe e ,
986 Eu opean Jou nal o Nuclea Medicine and Molecula Imaging (2023) 50:980–995
1 3
insi ua ions whe e images a e d awn on small objec s and
maximum oxel alues a e ex ac ed, i.e. whe e noise is a
mo e signi ican in luence, hen a pos - il e can be help ul
in ensu ing a g ea e le el o consis ency. Figu e3 dem-
ons a es how he applica ion o a 10mm Gaussian pos -
il e can subs an ially imp o e he consis ency o measu e-
men s o ac i i y concen a ion. The il e has been applied
o images econs uc ed wi h he same numbe o i e a i e
upda es. I is shown ha he pos - il e inc eases he deg ee
o nega i e bias in he smalle objec s, bu emo es he posi-
i e bias, due o noise, p esen in he la ge sphe es o he
un il e ed images.
Fo clinical epo ing, many nuclea medicine images a e
p oduced acco ding o a isual p e e ence o he clinician,
and hence i may be app op ia e o c ea e a second econ-
s uc ion ha is op imised o quan i ica ion.
Quali y con ol andha monisa ion
Quali y con ol equi emen s
Accep ance es ing and quali y con ol o SPECT-CT sys-
ems o SPECT and CT componen s should ollow in e na-
ional guidelines [21, 39]. Gi en he impo ance o a enua-
ion co ec ion o quan i a i e accu acy, ocus should also
be gi en o he alignmen be ween SPECT and CT, which
mus be checked pe iodically. The sensi i i y calib a ion
wo k low can di e signi ican ly be ween endo s. I is
he e o e ecommended ha he local physics eam should
de e mine he app op ia e equency in which o pe o m
e i ica ion scans o sensi i i y, based on manu ac u e ec-
ommenda ions. The sensi i i y calib a ion mus be epea ed
wi h majo changes in so wa e o ha dwa e.
Fig. 3 NEMA IEC image
quali y phan om illed wi h
99mTc a 10:1 con as in h ee
di e en sphe e con igu a ions
econs uc ed wi h 120 upda es
wi h no pos - il e (uppe se o
images) and a 10mm Gaussian
pos - il e (lowe se o images).
The co esponding eco e y
cu es a e gi en below o
maximum oxel ACR
987Eu opean Jou nal o Nuclea Medicine and Molecula Imaging (2023) 50:980–995
1 3
The need andme hods o ha monisa ion
Accu a e and ep oducible measu emen s o adioac i i y
a e essen ial o quan i a i e SPECT-CT imaging, enabling
he compa ison o esul s om mul i-cen e s udies [40].
This can be achie ed by es ablishing aceable esul s [41]
ha can be ela ed o a e e ence h ough a documen ed
unb oken chain o calib a ions, each con ibu ing o he
measu emen unce ain y [42]. Howe e , he unce ain-
ies a all s ages o he imaging acquisi ion and p ocessing
chain a e no p esen ly known o measu ed, pa ly due
o he a ia ions in a ailable co ec ion me hods, econ-
s uc ion algo i hms and he black-box na u e o he a ail-
able so wa e. Full aceabili y is he e o e no p esen ly
achie able, o en s opping wi h he ac i i y measu emen s
in he adionuclide calib a o p io o use in phan om o
pa ien s udies.
Ha monisa ion o 18F-FDG PET-CT has been add essed
by he EANM Resea ch L d EARL acc edi a ion p og amme
[43], and a simila app oach is cu en ly in de elopmen o
SPECT-CT. The ini ia i e in SPECT-CT was s a ed ol-
lowing he ecen in e es in absolu e quan i ica ion o his
modali y, mainly d i en by he need o dosime y ollowing
adionuclide he apy. Many in es iga o s ha e explo ed he
a iabili y o ac i i y quan i ica ion in in e labo a o y and
mul i-cen e s udies o 99mTc [44–47], 123I [48, 49], 131I [49,
50], 133Ba [51], 177Lu [24, 52, 53], 223Ra [54], highligh ing
he need o ha monisa ion p o ocols.
Accoun ing o di e ences in a ailabili y o local
esou ces, a se o minimum equi emen s o ha monise
SPECT-CT imaging ac oss cen es is ecommended:
• A adionuclide calib a o aceable o a na ional s and-
a ds labo a o y.
• A sui able and accessible phan om se o calib a e he
scanne and o assess pa ial olume e ec s h ough ACR
cu es ollowing he ecommenda ions desc ibed ea lie .
• S anda d ope a ing p ocedu es (SOP) o aceable phan-
om p epa a ion, image acquisi ion and econs uc ion,
assessmen o pa ial olume e ec s, ou lining olumes
o in e es and epo ing o esul s [19, 55].
• A e i ica ion phan om s udy is ecommended o assess
he quan i a i e accu acy ac oss cen es. This can be a
mo e ealis ic geome y, e.g. a ci cula o ellip ical cylin-
d ical phan om o ma ch he clinical condi ion unde con-
side a ion [24, 49].
All he scans mus be acqui ed and econs uc ed wi h he
same p o ocols used o he speci ic clinical condi ion. Fo
dosime y ollowing adionuclide he apy and o enable a
quan i a i e compa ison be ween cen es, i is ecommended
o calcula e he unce ain ies in he eco e y coe icien s ol-
lowing EANM guidelines [56].
The pa h oclinical use
Quan i a i e SPECT-CT is an eme ging imaging ech-
nology, bu , as wi h any new echnology, i s success
depends on whe he ou ine clinical applica ions can be
iden i ied. No e e y echnical e olu ion is au oma ically
ansla ed in o wide clinical accep ance, and depends on
issues including impac , ease-o -use, cos , a ailabili y
and an adequa ely ained wo k o ce [57]. I quan i a i e
SPECT-CT is he e o s ay, i should answe clinically el-
e an ques ions and impac pa ien ea men and ou come.
Recen de elopmen s in nuclea medicine in associa ion o
new he anos ic app oaches [58] suppo he use o SPECT-
CT o some adionuclides, such as 177Lu and 131I o isual-
ise he e icacy o ea men s. Quan i a i e SPECT-CT can
quan i y how much ac i i y is deli e ed o each umou lesion
and o gan a isk, and consequen ly, he abso bed dose. This
may be o high clinical impo ance o pe sonalised medicine,
e en hough con i ma ion in u u e s udies is needed.
The es ima ion o absolu e ac i i y concen a ion may
also be a ac i e o se e al o he pu poses: (1) o deli e
a eliable diagnosis, (2) o accu a e he apy esponse
moni o ing, (3) o p ognosis and o guide pa ien manage-
men decisions, (4) o imp o e he ep oducibili y o in e -
p e a ions, (5) o allow compa ison o da a be ween cen-
es (6) and o acili a e (semi)au oma ic analysis. Ini ial
applica ions epo ed in he li e a u e include he assess-
men o skele al condi ions (e.g. bone me abolism, de ec-
ion o bone me as ases, mandibula condyle asymme y),
co ona y a e y disease, amyloidosis and pa kinsonism.
Fo quan i a i e SPECT-CT, i will he e o e be essen ial
o de ine he limi a ions o he measu es p oduced by his
echnique. As di e en applica ions ha e di e en accu acy/
p ecision equi emen s, unde s anding he echnology’s
limi a ions will help guide he ocus owa ds a eas wi h he
highes likelihood o success ul clinical implemen a ion.
Op imisa ion o a ange o applica ions, adionuclides,
geome ies and ac i i ies will also be necessa y, as will
he ans e abili y o esul s. The cos -e ec i eness o he
echnique wi h ega ds o he humanis ic and socie al ou -
comes mus also s ill be p o en, which is beyond he scope
o hese guidelines. Howe e , hese guidelines should aid
in he design o clinical ials wi h he app op ia e me hod-
ology equi ed o demons a e he alue o he echnique.
Clinical use cases
Dosime y
Un il ecen ly, adionuclide he apies we e domina ed
by he use o adioac i e iodine, which has been used
988 Eu opean Jou nal o Nuclea Medicine and Molecula Imaging (2023) 50:980–995
1 3
87. Chan BH, Leung YY. SPECT bone scin ig aphy o he assess-
men o condyla g ow h ac i i y in mandibula asymme y: is
i accu a e? In J O al Maxillo ac Su g. 2018;47(4):470–9.
88. Kim J, e al. Maximum s anda dised up ake alue o quan i a i e
bone SPECT/CT in pa ien s wi h medial compa men os eoa -
h i is o he knee. Clin Radiol. 2017;72(7):580–9.
89. B aun M, Kacho an M, Vija A, Pagens e G, Wild D, K e zsch-
ma M. Quan i a i e 99mTc-DPD-SPECT/CT o he de ec ion
o p os he ic loosening in pa ien s wi h hip- and knee join
eplacemen —an in e im analysis o a p ospec i e s udy. J Nucl
Med. 2019;60(supplemen 1):92 LP – 92.
90. Beye T, e al. Wha scans we will ead: imaging ins umen a-
ion ends in clinical oncology. Cance Imaging. 2020;20(1):38.
91. Liu J, Yang Y, We nick MN, P e o ius PH, King MA. Deep
lea ning wi h noise- o-noise aining o denoising in SPECT
myoca dial pe usion imaging. Med Phys. 2021;48(1):156–68.
92. Ahn S, e al. Quan i a i e compa ison o OSEM and penalized
likelihood image econs uc ion using ela i e di e ence penal-
ies o clinical PET. Phys Med Biol. 2015;60(15):5733–51.
93. Ryden T, Ma in I, an Essen M, S ensson J, Be nha d P. Deep
lea ning gene a ion o in e media e p ojec ions and Mon e
Ca lo based econs uc ion imp o es 177Lu SPECT images
econs uc ed wi h spa se acqui ed p ojec ions. J Nucl Med.
2019;60(supplemen 1):528–35.
94. Shao W, Rowe SP, Du Y. SPECTne : a deep lea ning neu al
ne wo k o SPECT image econs uc ion. Ann T ansl Med.
2021;9(9):819–819.
95. Sei e R, Webe M, Kocaka uk E, Rischple C, and Ke s ing D.
AI and machine lea ning in nuclea medicine: u u e pe spec i es,
Semin Nucl Med. 2020;51(2):170–7.
96. A abi H, Akha anAlla A, Sanaa A, Shi i I, Zaidi H. The p omise
o a i icial in elligence and deep lea ning in PET and SPECT
imaging. Phys Medica. 2021;83:122–37.
Publishe 's no e Sp inge Na u e emains neu al wi h ega d o
ju isdic ional claims in published maps and ins i u ional a ilia ions.
995Eu opean Jou nal o Nuclea Medicine and Molecula Imaging (2023) 50:980–995
1 3