Modula aining esou ces o bioimage analysis
Ch is ian Tische 4*, An onio Poli i9*, Tim-Oli e Buchholz1, Elnaz Fazeli2, Nicola G i i3,
Aliaksand Hala a yi4, Sebas ián González-Ti ado5, Julian Hennies4, Toby Hodges6, A i Khan4,
Dominik Ku a4, S e ania Ma co i7, Bug a Oezdemi 8, Felix Schneide 4, Ma in Scho b4, Anniek
S okke mans10, Yi Sun4, Nima Vakili4
1Facili y o Ad anced Imaging and Mic oscopy, F ied ich Miesche Ins i u e o Biomedical
Resea ch, Fab iks asse 24, 4058 Basel, Swi ze land
2Biomedicum Imaging Uni , Facul y o Medicine and HiLIFE, Uni e si y o Helsinki, Finland
3Eu opean Molecula Biology Labo a o y, Ca e del D . Aiguade 88, 08003 Ba celona, Spain
4Eu opean Molecula Biology Labo a o y, Meye ho s aße 1, 69117 Heidelbe g, Ge many
5Ins i u e o Compu a ional Biomedicine, Heidelbe g Uni e si y, Im Neuenheime Feld 130.3,
69120 Heidelbe g, Ge many
6The Ca pen ies, c/o Communi y Ini ia i es, 1000 B oadway, Sui e #480, Oakland, CA 94607,
USA
7Randall Cen e o Cell and Molecula Biophysics, King’s College London, London, UK
8Eu o-BioImaging Bio-Hub, EMBL, Meye ho s aße 1, 69117 Heidelbe g, Ge many
9Facili y o Ligh Mic oscopy, Max Planck Ins i u e o Mul idisciplina y Sciences, Am Fassbe g
11, 37077 Gö ingen
10Hub ech Ins i u e, Uppsalalaan 8, 3584 CT U ech , he Ne he lands
* co esponding o [email p o ec ed] and [email p o ec ed]
Abs ac
Mode n mic oscopy enables us o measu e s uc u al and dynamical p ope ies o many
biological p ocesses and is he e o e an indispensable esea ch ool. Howe e , he amoun and
complexi y o he p oduced imaging da a is s eadily inc easing. Thus, handling he da a as well
as ep oducibly and au oma ically ex ac ing accu a e scien i ic in o ma ion equi es dedica ed
“bioimage analysis” expe ise. To acili a e he dissemina ion o his ubiqui ously equi ed
expe ise we de eloped an open-access bioimage analysis aining esou ce. The esou ce is
designed o help aine s o design and un cou ses on bioimage analysis o li e scien is s. The
ma e ial is modula whe e each module co e s one concise opic and p o ides co esponding
ac i i ies. The ac i i ies can be execu ed using a ious popula so wa e packages (e.g. ImageJ,
Py hon). The ma e ial is hos ed on a public so wa e eposi o y allowing he bioimaging
communi y o eadily con ibu e new aining modules o imp o e exis ing modules. Wi hin he
las h ee yea s, he ma e ial has been used by se e al aine s in nume ous cou ses and
con inuously imp o ed.
1
G aphical abs ac
In oduc ion
Mic oscopy is a key echnology d i ing biological esea ch and a la ge and di e se so wa e
ecosys em exis s o analyse mic oscopy images. So wa e solu ions can be open-sou ce o
comme cial, and a y om gene al o speci ic use-cases, ease o use, le els o documen a ion,
and accessibili y (Haase e al., 2022). Despi e hese a ailable ools, he complexi y and size o
he co esponding imaging da a can make i challenging o e icien ly ex ac bio-physically
meaning ul in o ma ion om he acqui ed images in a quan i a i e and eliable/ ep oducible
manne . As a consequence, he bio-compu a ional discipline “bioimage analysis” has eme ged
(GloBIAS, 2024, NEUBIAS, 2024). Scien is s pe o ming bioimage analysis ha e a di e se
backg ound and need o acqui e a heo e ical ounda ion on image o ma ion as well as image
p ocessing and analysis algo i hms. Fu he mo e, scien is s mus acqui e he necessa y skills o
e ec i ely and adequa ely use a ious so wa e ools o image analysis asks.
Du ing he las decade, online esou ces o sel -s udy ha e eme ged in he o m o ideos,
slides, in e ac i e books, and Massi e Open Online Cou ses (MOOCs) (e.g., (Bankhead, 2024,
Haase, 2020, Image.sc, 2019, Sei z e al., 2019, Jones e al., 2022)). In addi ion, nume ous
in-pe son o online bioimage analysis cou ses ha e been deli e ed and some imes eco ded
(e.g., (NEUBIAS, 2022, ZIDAS, 2024)). Li e online o in-pe son cou ses a e e y aluable as
hey p o ide a pla o m o ne wo king, in e ac ing wi h expe aine s, and add essing speci ic
2
ques ions wi h espec o lea ne s’ needs. We ound ha an open-access esou ce allowing
bioimage analysis aine s o e icien ly design and deli e such cou ses o a ious a ge
audiences would be help ul and is cu en ly missing (Haase e al., 2024, Si agu una han e al.,
2024). Such a esou ce should help and simpli y he much-needed dissemina ion o bioimage
analysis skills.
To add ess hose needs we s a ed in ea ly 2019 o c ea e an open-access online esou ce o
bioimage analysis aining accessible ia h ps://neubias.gi hub.io/ aining- esou ces/ (Tische e
al., 2024). Since hen, his esou ce has been used in a leas 25 cou ses wi h in o al nea ly 600
pa icipan s (Supplemen al Table 1). Wi h his publica ion we aim o u he sp ead he use o
his ma e ial and also kindly in i e he bioimaging communi y o con ibu e o any aspec o his
esou ce.
In he ollowing sec ions we will desc ibe he a ionale, con en , implemen a ion de ails and
usage ins uc ions.
T aining esou ce design and o ganiza ion
We aimed o a esou ce o acili a e deli e ing in-pe son o online li e cou ses. The ma e ial
should include ac i i ies o illus a e use cases. Gi en he la ge numbe o image analysis
so wa e, we wan ed o sepa a e an ac i i y om a pa icula so wa e implemen a ion. Finally,
he ma e ial should allow bioimage analys s o aylo cou ses o di e en a ge audiences. This
led o he ollowing design cons ain s: open-access online hos ing o he ma e ial, modula i y,
in e ac i i y, usabili y wi hin a li e con ex , and easy ex ension by he communi y.
Design p inciples
We ound ha exis ing online ma e ial is p ima ily concei ed as sel -s udy esou ces and does
no ye ul il all ou p e equisi es. While a simple collec ion o slides could be used wi hin a li e
eaching con ex , hey a e a he s a ic, o en lack in eg a ed ac i i ies, and a e ha d o
de elop/main ain as a communi y. P e- eco ded ideos a e inapp op ia e o a li e, in e ac i e
se ing and also lack in eg a ed ac i i ies. MOOCs p o ide comp ehensi e ma e ial o en
en iched wi h ideos, exe cises, and assessmen s ha a e bes sui ed o sel -paced s udy
(Sei z e al., 2019, Jones e al., 2022). The in e ac i e book by P. Bankhead (Bankhead, 2024)
con ains comp ehensi e explana ions, example images, and ac i i ies, ep esen ing a good
e e ence o bioimage analysis concep s. While all he a o emen ioned ma e ial a e g ea
esou ces, hey impose a p e-de ined o de o opics and lack modula i y. Modi ica ions o
addi ions o new concep s may he e o e equi e majo e isions; in addi ion, in ou iew, he
exis ing esou ces a e no designed o suppo agile con ibu ions by a communi y o bioimage
analys s.
3
We decided o base ou aining esou ce on he didac ically excellen aining ma e ial o The
Ca pen ies (h ps://ca pen ies.o g/) (Backhaus e al., 2024, B own e al., 2023). A Ca pen ies
expe has been a c i ical con ibu o o he ini ial de elopmen o he bioimage analysis aining
esou ce. Among o he hings, lessons by The Ca pen ies a e aimed a no ices as well as
expe s and g adually build knowledge, a oiding oo much in o ma ion a once. A lesson is
composed o se e al so-called episodes, las ing o 20 o 60 minu es wi h clea ly s a ed lea ning
objec i es. Each episode includes ac i i ies sui ed o li e demons a ion/li e-coding, so o
explain e e y s ep o he p ocess, and o ma i e assessmen s o iden i y misconcep ions. The
Ca pen ies p o ide a cou se on image-p ocessing wi h py hon ha can be used in a li e
con ex wi h a a ie y o example images and ac i i ies (Meysenbu g e al., 2023).
Ou ma e ial di e s om The Ca pen ies' p ima ily in i s inc eased modula i y and speci ic
applica ion o bioimaging da a. Unlike The Ca pen ies, which deli e s p ede ined lessons, we
designed small, independen aining modules ( eminiscen o Ca pen ies episodes) ha can be
lexibly assembled in o a ious cou ses — di e ing in leng h, a ge audience, and so wa e
packages co e ed (Fig. 1). To manage and op imize cogni i e load (Swelle , 1988), each
aining “module” ocuses on jus one o e y ew new concep s ha should be augh wi hin
30-45 min. This inhe en ly es ic s he complexi y o numbe o concep s included wi hin a
single module. A 'concep ' in his con ex e e s o a gene ic, commonly used p inciple o
me hod in bioimage analysis ha is agnos ic o speci ic so wa e implemen a ions, such as
'objec shape measu emen '. A module is designed o p o ide all he necessa y in o ma ion o
e ec i ely mo i a e, explain (e.g., h ough igu es, concep maps), and apply (e.g., using
example images and code) i s con en .
Figu e 1: T aining esou ce design: Hie a chy o he bioimage analysis aining cou se. A cou se is designed by
assembling se e al exis ing modules. Each module explains a concep /me hod and p o ides ac i i ies. An ac i i y has
gene al ins uc ions, an example image, and di e en so wa e implemen a ions.
4
Module s uc u e
The abo e design p inciples, e ined ia epea ed eaching o di e en modules, led o he
subdi ision o each module in o nine dis inc sec ions, as de ailed in Figu e 2 and Table 1. A e
he P e equisi es sec ion, he ollowing ou sec ions (Fig. 2A-C) se e o each he concep in a
lec u e s yle p esen a ion, whe e he Concep map and a Figu e a e used as a isual suppo o
he ins uc o o explain he concep o he lea ne s. Some addi ional de ailed explana ions could
be added a e he main igu e. The main sec ion o a module is Ac i i ies ha demons a es
di e en applica ions o he concep s ha a e augh in he module. Each ac i i y is desc ibed in
gene al so wa e-agnos ic e ms and has linked example da a ha can be eadily used. The
ins uc o and lea ne can hen selec ins uc ions o speci ic so wa e packages (Fig. 2D, E).
This sepa a ion o gene al concep s om speci ic implemen a ions was a majo design decision
o his esou ce. This makes i possible o con inuously add new so wa e implemen a ions o
an ac i i y, wi hou needing o change much o he gene ic eaching con en . In addi ion, his
design also inc eases he didac ic quali y o he ma e ial, because i o ces he aine and
s uden s o disen angle uni e sally applicable concep s and me hods such as, e.g. “connec ed
componen analysis” and “objec shape measu emen s”, om speci ic implemen a ions such as
ImageJ’s Pa icle Analyze (Schneide e al., 2012) o sciki -image’s egionp ops ( an de Wal
e al., 2014).
5
Figu e 2: S uc u e o a aining module: This igu e shows sc eensho s o a module websi e and ac i i y. (A)
Knowledge equi ed o ollow a gi en module is p o ided ia links o o he modules (P e equisi es). Lea ning
objec i es and Mo i a ion p o ide he why and wha will be lea ned du ing he module. The (B) Concep Map and (C)
Figu e isuals p o ide and o e iew o he key concep s. Addi ional explana ions may appea a e he main igu e
(no shown he e). In Ac i i ies (D) he ac i i y ins uc ion and i s implemen a ions p o ide he necessa y in o ma ion
o showcase a concep using a speci ic so wa e (e.g. py hon and sciki -image/napa i). (E) Example ou pu gene a ed
when execu ing an ac i i y. (F) Fo ma i e assessmen s a e p o ided a he end o a module o ein o ce wha has
been lea ned.
6
Module sec ion
Explana ion
P e equisi es
Con ains links o o he modules o ou ma e ial,
de ining he con en ha lea ne s should know
be o e s a ing
Lea ning objec i es
De ines he skills ha lea ne s will acqui e
Mo i a ion
Explains why i is impo an o acqui e hese skills
Concep map
A schema ic d awing ha de ines he co e new
concep s ha a e augh
Figu e
A igu e illus a ing he eaching con en
Explana ions (op ional)
Addi ional de ails ypically used o help sel -s udy
and add ma hema ical con ex
Ac i i ies
In e ac i e ins uc ions o applying he concep s
augh in a ious so wa e pla o ms
Assessmen
Ques ions and answe s o es ing he concep ual
unde s anding
Follow-up ma e ial
Links o sui able ollow-up aining modules and
ex a-cu icula ma e ial
Table 1: T aining module sec ions: A aining module is composed o di e ence sec ions o in oduce he concep ,
p o ide examples, and u he eading.
The subsequen “assessmen ” sec ion (Fig. 2F) comp ises sho quizzes ha allow he
pa icipan s o es hei unde s anding o he main concep s. Fo ins ance, a ques ion and
answe could be: ”Q: Wha is he ypical da a ype o a label mask image? A: Typically, label
mask images a e o posi i e in ege da a ypes (unsigned in ege 8 o 16 bi ) wi h ze o being he
backg ound label”. Finally, he Follow-up Ma e ial sec ion p o ides links o addi ional modules
ha can be lea ned a e he cu en module o ex e nal esou ces.
In ou expe ience, his p ede ined module s uc u e signi ican ly acili a es he addi ion o new
ma e ial. While i s ill o ces he con ibu o o ca e ully hink abou wha hey ac ually wan o
each in he module, i ees one om hinking oo much abou how o s uc u e and o ma he
ma e ial.
Resou ce con en
O e he las 5 yea s we c ea ed ~50 eaching modules (Supplemen al able 2). Supplemen al
ideo 1-3 illus a e he con en o he aining ma e ial and how o each a module.
The cu en ma e ial co e s con en ypically augh in in oduc o y and in e media e bioimage
analysis cou ses. Among o he s, he esou ce includes modules on image da a handling and
he unde s anding o he di e en da a o ma s, image da a inspec ion, image il e ing and
7
h esholding, segmen a ion and measu emen s, as well as basic sc ip ing, ba ch p ocessing,
and cloud based compu ing. We also included so-called wo k lows, whe e lea ne s, s a ing
om example images, apply a sequence o me hods o sol e an image analysis ask and ob ain
a esul able. Finding sui able example image da a o eaching a ce ain concep is a majo
e o . The eposi o y con ains small example image da a ha ha e been ca e ully selec ed o
ul il his need. The ac i i ies a e designed o di e en le els o unde s anding and complexi y.
8
Figu e 3: Module dependencies and cou ses: This igu e shows sc eensho s o he esou ce home page. (A)
Alphabe ical lis o modules. (B) Ne wo k o dependencies be ween modules. Each module is a node and he
P e equisi es de ine he edges o he g aph. By clicking on one o he nodes (a ow) he dependencies a e highligh ed
(g een). This in o ma ion can be used o de ine a se o modules o a cou se. (C) Link o cou se sugges ions ha can
9
Re e ences
Backhaus, A., Becke , E., Bellini Saibene, Y., B ooks-Kie e , J., Companjen, B., Cu y, R.,
Da ey, R., He weck, K., Hodges, T., Kam a , Z., Khan, B. J., Koh , F., Kuzak, M.,
McCla chy, S., Michonneau, F., Nenadic, A., S e ens, S., Takemon, Y., T izna, M.,
T usle , A., en de Bu g, S., Alleg a, V., Wilson, G. & Wo d, K. (2024) The Ca pen ies
Collabo a i e Lesson De elopmen T aining. So wa e Ca pen y Founda ion, doi:
10.5281/zenodo.8410625, u l: h ps://ca pen ies.gi hub.io/lesson-de elopmen - aining/.
Bankhead, P. (2024) In oduc ion o Bioimage Analysis. doi, u l: h ps://bioimagebook.gi hub.io/.
B own, S. M., Dennis, T., Pe ez-Sua ez, Po e , N., Wheele , J. & Wo d, K. (2023) The
Ca pen ies Ins uc o T aining. So wa e Ca pen y Founda ion, doi:
h ps://doi.o g/10.5281/zenodo.7612756, u l:
h ps://ca pen ies.gi hub.io/ins uc o - aining/.
EMBL-EBI (2024) BioImage A chi e. doi, u l: h ps://www.ebi.ac.uk/bioimage-a chi e/.
GloBIAS (2024) Global BioImage Anlays s's Socie y. doi, u l: h ps://www.globias.o g.
Haase, R. (2020) Lec u e BioImage Analysis. You ube, doi, u l:
h ps://www.you ube.com/playlis ?lis =PL5ESQN M5lc7SAMs Eu082i W4BDM d0U.
Haase, R., Fazeli, E., Legland, D., Doube, M., Culley, S., Bele ich, I., Joki alo, E., Scho b, M.,
Klemm, A. & Tische , C. (2022) A Hi chhike 's guide h ough he bio-image analysis
so wa e uni e se. FEBS Le , 596, 2472-2485, doi: 10.1002/1873-3468.14451, u l:
h ps://www.ncbi.nlm.nih.go /pubmed/35833863.
Haase, R., Tische , C., Bankhead, P., Miu a, K. & Cimini, B. (2024) A Call o FAIR and
Open-Access T aining Ma e ials o ad ance Bioimage Analysis. OSF P ep in s, doi:
10.31219/os .io/2zgmc, u l.
Image.sc (2019) Bioimage Analysis: Recommended Reading and Viewing - Image Analysis.,
doi, u l:
h ps:// o um.image.sc/ /bioimage-analysis- ecommended- eading-and- iewing/28051.
Jekyll (2024). 4.3.4 ed., doi, u l: h ps://jekyll b.com/.
Jones, S., Bu ns, I., A ms ong, B., Wang, A., Rich, M. & Thompson, M. (2022) Image
P ocessing o Enginee ing and Science Specializa ion. Ma hwo ks, Cou se a, doi, u l:
h ps://www.cou se a.o g/specializa ions/image-p ocessing.
Meysenbu g, M., Du ham B ooks, T., Ku a, D., Pape, C. & Becke , E. (2023) Image P ocessing
wi h Py hon. doi, u l: h ps://da aca pen y.gi hub.io/image-p ocessing/.
NEUBIAS (2022) NEUBIAS Academy. doi, u l:
h ps://eubias.o g/NEUBIAS/ aining-schools/neubias-academy-home/.
NEUBIAS (2024) Ne wo k o Eu opean BioImage Analys s. doi, u l:
h ps://eubias.o g/NEUBIAS/.
Sei z, A., Guie , R., Bu i, O. & Chia u ini, N. (2019) EPFLx: Image P ocessing and Analysis o
Li e Scien is s. EPFL, edX, doi, u l:
h ps://www.edx.o g/lea n/image-analysis/ecole-poly echnique- ede ale-de-lausanne-ima
ge-p ocessing-and-analysis- o -li e-scien is s.
Si agu una han, S., Ma co i, S., Nelson, C. J., Jones, M. L., Ba y, D. J., Sla e , T. J. A., Elicei i,
K. W. & Cimini, B. A. (2024) B idging imaging use s o imaging analysis - A communi y
su ey. J Mic osc, 296, 199-213, doi: 10.1111/jmi.13229, u l:
h ps://www.ncbi.nlm.nih.go /pubmed/37727897.
Swelle , J. (1988) Cogni i e Load Du ing P oblem Sol ing: E ec s on Lea ning. Cogni i e
Science, 12, 257-285, doi: h ps://doi.o g/10.1207/s15516709cog1202_4, u l.
Tische , C., Poli i, A., Buchholz, T., Fazeli, E., G i i, N., Hala a yi, A., Gonzalez Ti ado, S.,
Hennies, J., Hodges, T., Khan, A., Ku a, D., Ma co i, S., Oezdemi , B., Schneide , F.,
Scho b, M., S okke mans, A., Sun, Y. & Vakili, N. (2024) Modula aining esou ces o
16
bioimage analysis. 1.0.1 ed. Zenodo, doi: 10.5281/zenodo.14264885, u l:
h ps://gi hub.com/NEUBIAS/ aining- esou ces/.
ZIDAS (2024) Swi ze land's Image and Da a Analysis School. doi, u l: h ps://www.zidas.o g/.
Supplemen al ma e ial
The supplemen al ma e ial is a ailable a :
P ep in
Supplemen al Table 1: Lis o cou ses 2021-2024
The able lis s cou ses ha ha e being held in he las 4 yea s using he bioimage analysis
aining esou ces.
Supplemen al Table 2: Lis o modules
The able lis s all he consolida ed modules in he esou ce.
17
Supplemen al Video 1: Bioimage analysis aining esou ce: How o
A ideo illus a ing how o na iga e h ough he websi e o he aining esou ce.
Supplemen al Video 2: Digi al image basics: In oduc ion
Video showing he eaching o he module Digi al image basics.
Supplemen al Video 3: Digi al image basics: 2d py hon
Video schowing he eaching o he i s ac i i y in he module Digi al image basics.
18
