Evaluation of a Virtual Reality-Based Eye Tracker for Neuro-Ophthalmic Assessment: A Feasibility, Reliability and Reproducibility Study

Pa r tic ipa nt s’ feed b ac k (n= 27 ) hi ghl ighted hi gh ac cept ability f or Bulbi CA M : 89% foun d th e 38 te st com for table , 9 2.6% fe l t t h e te sting duration w a s app ropri at e, an d 81.5% rep o rt e d no ey e strai n 39 or fa tigue . Int er -visi t r e p r od ucibil ity o f B ulbi cam t e st s s how ed hig h r eproducib il it y for pur suit and 40 pupil te s t s (I CC = 0.8 8-0 .76) , w hile s ac cad ic task s s how ed lo we r rep roduci bility ( be st IC C a t 0.62 ). 41 Pai red a sse s s me n t s be t w e en de vice s sho we d clo s e a g reem ent for ke y pupi llome t er metr ics : b a seli n e 42 dia mete r (bi as : -0 .48 ± 0.47 mm), p eak cons tr ic ti on dia me ter (bia s: -0.56 ± 0 .36 mm), cons tr i cti on 43 ve locity (bia s: 0.22 ± 0.58 mm/s), and d u r a t i on o f con s t riction (bi a s : - 0 .052 ± 0 .15 s) . 44 Conc lus ions: Thi s stu dy hi ghligh t s the c li nic al fea s ibi li ty of Bul biC AM, with hi gh p atie n t acc ept abili t y 45 and r e produ cibil it y f or pur s ui t a nd pupil t e s t s . Pai red a sse s s men t s con f i r me d it s a cc ur a cy f o r k ey 46 pupil lomet ric pa r a me te rs, v alida t i ng it s re liabil ity f or cli nica l and r e sea rc h use . 47 48 49 50 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 3 51 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 4

52 Ey e-t r ac king devi ces a re w idely utiliz ed i n bo t h neu ro- ophth almic re s e a r c h an d cl in ical s e tting s , 53 providi ng va luable in sight s int o ocul ar an d ne ur ol ogic al func t i o n (1 -3). In rec ent y ea rs, Virtu al Re ali ty 54 (VR ) hea d set e ye - tr a cki ng dev ice s ha ve ga ined i ncrea sing pop ulari ty a s po rta ble and e ffic ien t 55 alter nativ e s to conv enti onal e ye -t rack ing sys tem s (4 ). By int eg r a t i ng adv ance d ey e -trac king 56 ca pabili t i es wit hin VR h e ad se ts, th e se de v ice s of fe r en ha nce d fl exi bility a nd con venien ce (5) . Unli ke 57 co nventiona l s e t u p s w ith s e p ara te di spla y s , p r oje c tor s , a n d co mpute r s , the s e al l -i n-on e sy ste ms 58 of fer a spac e- s a ving s olu tion f or clini cs w hile providin g ac curat e ey e -t r a cki ng data . 59 Both subjec t i ve a n d objec t i ve me t hod s a re employe d to eva lua te ey e mov eme nt s and pupi llary l ight 60 refl ex e s . H ow ever , the re i s a growin g ne ed for c linic all y ad a pt a ble ey e t rack ers u t ilizing objec t i ve 61

s (6 ). Pa rtic ularly, VR hea ds et ey e - t rac king devi ce s equipp e d with a u tom ated so ft w a re, o f fer 62 signific ant adv an tage s by providin g preci se a n d s t a nd ardize d mea sur ement s, en s ur i ng ac curac y an d 63 relia bili ty (7, 8) . Bulbi CAM i s an examp le of a V R head s et with eye trac king capa bil itie s aime d a t 64 obtaini ng obje c tive metr i c s for oculomo tor be haviour a nd pupill ome te r; thu s h avi ng the pot enti al 65 for a s s e s sing a ra nge o f n euro - o p ht h almi c disea s e s a nd dys func t i on s (9 ). 66 VR ey e t rack ing sy ste ms a r e und er -expl o r e d compa re d to t r a di tional sy stem s (10 ) and to d at e, th er e 67 are no re pr od ucibil i ty/rel iabili ty studie s using V R head s et -ba sed eye tra cke rs. Th e r e for e, in thi s 68 st u d y, we aim to a sse s s th e r e p r odu cibil i t y of Bulbi C A M te st s an d relia bil ity of Bul biCAM com pa red 69 to oth er wea ra ble ey e track e r with pupi ll ometry such a s PupilL ab s Neo n gla s se s , o ffe r i ng new 70 insight s in to th eir po ten tia l cli nica l appli c abil ity. 71 72 73 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 5

s (Subjec t s) and Methods 74 Parti cipan t s 75 Healthy par ticipa n ts (n =39 ) wer e rec r u i te d for t h e s t udy and succ e s sfull y co mple ted bo th v is i ts . 76 Healthy s ubj ect s wer e de f i ned a s pa rtic ip ant s who ha d no neu rolo gica l di se as e s or clini cal s ig ns o f 77 oc ular motili ty dysfu nction . The dem ogr aphic charac t eri s tic s o f t h e part icipa nts a re pr e sent ed in 78 Table 1. T heir ag e s r ang ed from 21 t o 59 yea rs, a nd th e mal e -to- fema l e ra tio w a s 18:22. Re fr a c tive 79 erro r s rang ed from -8 .00D to +5 . 50D . All subjec t s a chi ev ed a be s t -c o r rec t e d visu a l a cuity (BC VA) o f 80 0.20 logMAR or bet ter . Colo ur v is i o n, a s ses s ed wi t h t h e I s h ihar a p seudoi s ochrom a t i c pla te s, w as 81 normal ( 1 7 / 17 ) in mo st pa rticipa nts. Ho we ver, one indivi dual r ec or d ed score s o f 5/ 1 7 in the r ig h t 82 ey e and 6 / 17 in th e le f t e ye . 83 None o f th e p artic ipant s ha d a hi stor y o f neurolog ical di so r de rs . Oph thalmolog ica l histo r ie s w er e 84 large ly unrema r k able, exc ept in fiv e indi v idual s: thr ee ha d cornea l re frac tive s u rge r y , one h ad 85 previ ou s ly undergon e strabismu s s urge r y and a lso had c olour b lind ne ss, and on e had lat tice 86 d e ge n e ra t i on a nd p os t e r i or vi t r eou s det ac h m e n t in bot h e y e s . 87 Written in forme d co nsent w a s obtaine d f rom each p art i cip a nt. 88 89 90 91 92 93 94 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 6 Tabl e 1: Demog r a phic info rmati on of p a rtic ipan ts 95 Va r ia ble Age, M ea n ± SD 30. 2±9.7 S ex, No. (%) Fem ale 2 1(54) Male 1 8(46) E t hnic ity, No. (%) White-Br i t ish 2 (5) W hi t e-Other 8 (18) As ia n-O riental 21 (54) A sian-India n S ubcontin ent 7 (18) A fro-Ca ribbean 1 (2.5) Other 1(2.5) Vis ua l Acuity (c or rec t e d, logMAR), Mean ± SD L eft e y e -0.01±0 .06 Right e ye -0.02± 0.07 Col our Vision N or mal N o. (% ) 38 (97.4) 96 Study D e s ig n 97 39 p art i cipan t s co mplet ed two te sting se s s i on s , s c hed uled a t s im ila r time s o f th e da y. At ba s eline , all 98 partic ipan t s unde rwent be st -corr ecte d v isual ac ui ty a ss e ssment s (ETDRS c har t at 4m) a nd I shihar a 99 te sts for col ou r vision scr eeni ng. 100 Fi gure 1 il lustra te s t he par tici pan t s ’ fl ow through th e s tudy and t he o r d er o f te s tin g pe r fo rmed. 101 Furthe r de tai l s reg a rding the s timulus p a rame te rs u sed in Bulbi CA M are p rovi ded belo w. 102 In the s eg men t o f the s tudy inv olvin g Neon gl asse s , part ici pant s w or e the gl a sse s whil e tr a ck ing 103 st i muli di spl ay ed on t h e Bulbi CAM scre e n. D uring the s e me as urem ent s , pa rticipa nt s ’ he ad s wer e 104 posi tione d slig h t l y behi nd th e s t a nd ard p os tur e t o acc ommoda te t h e glas s e s a nd ens ure ac cur at e 105 data cap ture . 106 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 7 107 Figure 1 : Flow cha r t ex plai n in g th e s tudy de s i g n . B: B ulb icam, N : P upilL a bs N eo n e ye t rac king gla sses 108 109 Part i cipan t Ex perien c e 110 P ar ticipa nts w e r e a sked t o c omplet e a q ues tionn aire a fter c omple ting t w o vi s i t s. The q ue s tion na ir e 111 s c or ed (1 ) C om for t, (2 ) Le ngth o f th e ex amina t i on, (3) Brigh tne s s , (4) D o uble vi s io n, (5) Gr ad ing th e 112 fatigu e/ ey e strain s p ec ific ally for Bul bica m on a 5 - point Lik er t s c ale a s : 1 = S trongl y Dis a gr ee, 2 = 113 D i s ag r e e, 3 = Neu t r al, 4 = Ag ree , 5 = S tr o ng ly Agree (Figur e 3 ). 114 115 BulbiCAM 116 BulbiC A M (BulbiT ech AS , Oslo , Norwa y ) is a C E - c er t i f i ed and F DA- r eg ister ed e ye - t r a cki ng dev ice t ha t 117 us e s A I - int eg r a te d s of twa re t o a sse s s the data ob tain ed by camer a s. I t ac comm od ate s an 118 interpu pilla r y di st a nce o f 49mm to 75m m . The di s pl a y mode hardwar e u t i li se d is th e Sha r p 119 LS0 55R1SX0 4, whi ch c ons i sts of 5.5” LCD pane l (2560× 1440 pi xel s @60 fp s; 1920× 1080 pi xels @60 120 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 8 fp s ), M IPI boa rd, Whi te 60 - pi n fl ex cable , HD MI - MI P I b oard (ba sed on T C358870XBG M silic on) . The 121 came r a i s config ure d to ca p t ur e tw o dif f eren t type s o f ima ge s alt erna tely , D a r k Pupil (DP ) an d Bright 122 P upil ( B P). I f the ini tia l frame i s B P , th e n ext one will be D P , f ollo wed by BP ag ai n, a nd s o o n. The 123 came r a ope rat e s at 400 frame s pe r s ec o nd (FP S ), cap turing 200 DP a nd 200 BP i mage s pe r s ec ond, 124 r e s ulting in a to t a l o f 400 pupil ima ge s p er s econd . The s timulu s s y nchr oniza tion typi call y has an 125 acc uracy of ±18 ms, with a w or s t - ca se d e viation of ±38 ms (9 ). 126 Expe r im ental se t-up is s how n on F igur e 2 A, a nd Fi gure 2B show s r eal time re cordi ng s c r e en. 127 128 Figure 2 : (A ) Bul biCAM ey e track ing de v i ce s e t up an d a par ticip a nt bei ng a sse s s e d ( p l e as e c o nt ac t 129 t h e c or re s pon ding au tho r to req ue s t a c cess to F i gure2A) ( B ) B ulbic am screen d ur i ng a s s e ss men t (C ) 130 origin al pur s ui t te s t da ta . Whi te l ine s s h o w the s tart i ng and e ndi ng o f th e stim ulu s. Verti cal a r row is 131 a s c a l e of w a v e am pl it ud e a nd h o r i z ont al a r r ow r e p r e se n ts t wo s econd s ( D) ori gin al s ac cad e s te st 132 dat a (E) origin al pu pi l te st da t a 133 134 PupilLab s N e on g la s s e s 135 P upilLab s N e on Gla s ses (Pupil Labs Gmb H, Berlin, Ge rmany ) ar e ey e - t r ac king glas se s e quipped with 136 s c en e c amera s in fron t of the g la s s e s and t w o eye cam er a s w ith 850 nm infr are d ill uminator LE Ds . 137 The de vic e is conn e cted to th e c ompani o n devi ce (Andr oid phon e) whi ch en abl e s the gl a sse s to r u n 138 and proc e ss th e da ta coll ecti on. P upil C lo u d websi te i s th e rec ommend ed t o o l fo r analy s in g th e da ta 139 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 9 and pr o ce s s in g th e reco rdin gs. PupilL abs N e on gla sse s en ab le s r e sea rche rs t o co llect g aze, sacc ad e s, 140 pupil diame ter s dat a a long side r ea l -worl d fron t ca mera foo tage . The c amera op e rate s a t a 141 freq uenc y of 20 0 f rame s pe r se cond ( F PS ) (11) . 142 Data Set 143 Pur suit s, s ac cade s a nd pupill ary lig ht r e f l ex tes t val ue s wer e obta in ed by Bulbi CA M. The s mo oth 144 pursu it te s t ev alu ate s how well th e ey e t r a cks a vis ual ta rget smoo thly . The gain v alues w ere 145 provide d f or pur s ui t te s t s i n re s pon s e to 10 degree 0.5 Hz fr e qu ency stimul i. Gain, w hich measur es 146 track ing a ccurac y , is d efin ed a s th e r a t i o of the s t im ulu s ' s v eloc ity t o the ey e's m oveme nt veloc ity. 147 Per fec t t rac king i s indic at ed by a gai n val ue of 100 %, but un d er-t rac king is sugg e sted by va lue s le s s 148 than 1 00%. T hr o ugh ex amina tio n o f thi s variabl e a t dif fer ent s timulu s fre quenci e s, the acc uracy and 149 ef fec t i ve ne s s of the smoo t h pur s ui t sy st e m ca n be e valua t e d in va r i ou s s c enari o s ( 9,12). S t i muli and 150 track ing rec ordi ngs o f bo th right an d l ef t eye is s how n in Figure 2 C. 151 Sa cca de s are rapid eye mo veme nt s th at s h i ft gaz e betwe e n dif f eren t ta rget s . Duri ng a visual s earc h 152 ta sk, the sac cad e t est de live rs both ho rizo nt a l and ver tical s timul i . Pe a k ve locity, l atenc y, an d 153 ac curac y pa r a me ters a re p r ov ide d to a ss ess sacc adic ey e move me nt s . Ve rtic al 9 d egree a n d 154 horizon tal 20 degre e sac cade pa rame ter s were a ss e ssed u sing BulbiC AM’ s sta nd ard te st pro toc ol. 155 Fi gure 2D show s exa mple s o f horizon tal and vertica l s a cc ad e s. 156 Pupil ba selin e -pe ak c on s t ric t i on di ame te rs and con s t ric ti on vel ocity value s w er e obtain e d fo r 157 pupil lomet ry te st i ng . Da ta f rom onl y one e ye w as anal y s e d to a s s e s s re pr od ucibil i ty and r e liabil i t y of 158 pupil lomet ry te st i ng, since all the p art i ci pa nts we r e h eal thy and w e ex pected a high degree o f 159 co r rela t i on be twee n the two e ye s. The di agram in Fig ur e 2E show s th e pu pil dia m ete r increa sing and 160 dec r e a sing th rough th e t ri al s . I n ad dition t o Bulbic am, th e s ame variabl e s fo r pupi l te s t w er e 161 obtain e d fro m PupilL ab s N eon gla sse s Pu pil C loud pla t f orm w here gaz e a nd pupil diamete r da ta is 162 av ailable. 163 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 10 Statist i c al Analy si s 164 A sampl e s i ze c alc ula tion w a s per f o r m ed bas ed on a n exp ec ted i ntrac la s s c orre la ti on c oef ficie n t 165 (I CC ) of 0. 675 , deriv ed from sacc adic gai n r e liabil i ty (I CC s ra nged from 0 .52 – 0 .8 3) data in h eal thy 166 co nt rol s r epor ted in a Hun tington' s di sea se stud y (13) . A s s umi ng a null hy pothe s i s ICC o f 0.50, a 167 signific anc e lev el o f 0.0 5, an d 80% pow er, the r equire d samp le siz e wa s e stim ated to be 27 168 partic ipan t s . To ac coun t fo r an an tici pat e d 10% dropout ra te, the fina l adju st ed sa mple s iz e w as 169 dete rmined to be 31 par t i cipa nts . 170 The IC C wa s u s ed to a sse s s i n t e r-vi s it r ep roduc ibility o f BulbiC AM t e sts (14 ). At th e end o f the two 171 vi s it s, da ta from n= 39 pa rtici pan t s w as a nal ys e d for the pu rsui t te st, n =37 fo r t he vertica l 9 de gre e 172 sacc adic t e s t, n = 35 fo r the h orizon tal 20 degree sac cad ic te st , and n= 35 for th e p upil te s t , 173 re spectiv ely . F or pur s ui t s, gai n val ue s we re an aly sed in re s pon se to 10 degre e 0.5 Hz freq ue ncy 174 st i mulu s. For sacc ad es , pea k ve locity , la t enc y and acc uracy val ues o f ver tica l 9 de g r e e and ho r i z on tal 175 20 d egree w er e anal y sed . For th e pupil t es t; ba s eline pupil di a met er, pe ak cons tri ction di ame ter an d 176 co nstriction vel oc ity w ere an al yse d. Te st reprod ucibi lity is c a tegor ized a s ex cell e nt when t he I CC 177 ex ceed s 0.9 0, good when it fall s be twee n 0.75 a nd 0.90 , moder at e fo r va lue s rangi ng fr o m 0.50 to 178 0.75 , and po o r w hen i t i s be low 0 .50 (15 ) . 179 We obt ain ed pai red pupill ome try data fr om Pupil Lab s Neo n gl ass e s in n= 23 par tic i pant s and u sed 180 Blan d-Al t ma n te st to e valua t e t h e agr ee me nt in pupi llary li ght r ef lex mea sur e m e nts betwe en 181 Bulbi CAM a nd Pu pil Lab s Neo n g las ses (1 6). 182 Blan d-Al t ma n anal ysi s w a s pe rf or me d u si ng Graph Pad P r i s m, a nd I CC a n alysi s u sin g R. 183 Ethic al app roval 184 This study rec eived eth ica l approva l from the Ea s t Midl a nd s – N o tting h am 2 Re sea r c h Ethic s 185 Committ ee ( RE C ) (RE C re fer ence : 21/EM / 00 17, pro tocol ver sion : 0772 , and th e UK Hea lth Re s earch 186 Autho rity (HR A) app roval ( IRAS I D: 279 30 9). 187 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 11

188 Part i cipan t Ex perien c e 189 A tota l of n =27 par ticipa nts pr ovi ded f ee dba ck on their Bu lbiCAM e xamina ti on ex perienc e us i n g a 5 -190 point Li kert s c al e, achi eving an ex cel len t re s po n s e r at e, with ma jor i t y r epor t i ng posi tive e xperi enc e s. 191 Spec ifica lly, 89.0% fo und th e te s t comfo r t a ble , 9 2.6% c onsid ere d the ex amina t i on leng t h 192 appropri at e, a nd 92.5% wer e sa ti s fied w i t h the b rightne s s level s. F urthermo re, 6 2 .9% repor ted no 193 occ ur renc e o f do ubl e vi s ion , and 81 .5% i nd icated t he y ex peri enc ed no ey e s t rai n or fa tigue (F igu re 194 3). 195 196 197 Figure 3 : Patie nt Expe r ie n c e A c r o ss D i ffe rent Va riable s. Bar c harts ill ustra ting the frequ ency of 198 r es pons es fo r v ar ious pat ie nt e xp e r i e n ce f a c t o rs : ( A) C o mf o r t d u r i ng th e pr o ce du re , ( B) 199 A p p r op r i at e ne ss of t he e xa m i na t io n l en gt h , ( C) B r i gh tn e s s l ev e ls us e d du r i ng t he ex am i n at i o n, (D ) 200 Inci denc e o f do ub le vi s i on , a nd (E) E ye str ain or f ati gue expe r i ence d d uring o r a fte r the exam i nati on . 201 202 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 12 Inter -vi sit Rep rod u c ibil it y of B ulbi cam T est s 203 The pursuit t e st demo ns tra t e d high repr oduc ibili t y , with a n I CC v alu e o f 0 .81 (Fi g ure 4A) . Fo r the 204 ve r tic al 9 degre e sac cade s t e st, the I C C v alue s fo r lat enc y, pe ak v eloc ity, and acc uracy were 0 .62, 205 0.50 , and 0 .49, r e spectiv e ly. S imila r l y, f or the horiz ontal 20 d egree sac cad e s te s t, t he IC C v alue s f or 206 the se me tric s we r e 0 .52, 0 .46, a nd 0 .61, re spec tively ( Figu r e 4A ). 207 In the pupil te st , high r e produ cibil it y wa s obse rved f or th e ba selin e p upil dia me te r (I CC : 0.88), for 208 the pe ak c onstric t i o n diame ter (I CC: 0 .83 ), a s w ell a s fo r the c on stric t i on veloc ity ( IC C: 0.76 ) (Fi gur e 209 4A). 210 The se re sul ts in dic at e vary ing le vels o f re produc ibility acro s s t he t e ste d pa r a m ete r s, w it h part icula r l y 211 strong con si st e ncy in the pur sui t and pu p il te st me tr ic s . 212 213 214 215 216 217 218 219 220 221 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 13 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 Fig ur e 4. In tra -Cla ss Cor relati on (ICC) Analy s i s an d Blan d - Alt man Pl ots: (A ) ICC v a lues wi th 95% 239 co nfide nc e inte rv al s fo r sac cade s , p ursui t and p upil te st s. Col our s repre se nt di f fere nt te s t gro up s : 240 ve r tic al 9° (de g r e e ) sa cca dic task (b lue ), h orizon tal 20 ° ( de gr e e ) sacc adic t a s k (gr e e n), p ur suit te st 241 (pur ple) , an d pu pi l te sts (red ). (B-E ) Bla nd -Al t m a n plo t s for the p upi l te s t m e tr ic s f o r pupil d i ame ter 242 (ba seli ne a nd p e ak c ons t ric tio n), c on stric t io n ve l ocity (mm /s) a nd d ur a t i on of c on strictio n . Avera ge 243 repres en ts t h e bi a s betw e en tw o m ea sur emen ts , wh ile da s hed lin es i nd icat e the 9 5% limits o f 244 agree m en t (LoA ). 245 246 247 (A) (B) (C) (D) (E) . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 14 Paire d As se s sment Bet ween D e vic e s 248 The Bland -Altma n plot s i llu s t rat e th e a gr eemen t be t w e en Bul bi CAM and Pu pilLa b s N eon gla s s e s 249 mea s u remen t s ac r o s s variou s pupil te st pa r a me ters : Pa i r e d a s s e s smen t be twee n devi ces showed 250 cl os e ag reem en t for k ey pu pillome te r metric s : ba sel ine di a met er (bi a s: -0 .48 ± 0. 47 mm LoA:-1.4 t o 251 0.45 ) , pe ak cons tr i cti on dia me ter (bia s: - 0.56 ± 0.36 m m LoA: -1. 3 to 0.15 ), c on str ic t i on vel ocity 252 (bia s: 0.2 2 ± 0.5 8 mm/ s L oA: -0.9 to 1 . 3), and du r a t io n o f cons tric tion (bi as: - 0 .052 ± 0.15 s LoA : - 0 .34 253 to 0.23 ). Mo s t d ata po int s fe ll wi t h in th e limi t s o f a gre emen t (Lo A), demon s t ra tin g agreeme nt 254 betwe e n sy stem s (F igure 4B , 4 C, 4 D, 4E). These re s u lts sugge st goo d agreem e nt b etween t he t w o 255 sys t e m s , w ith mino r, c on s i s t ent bia se s ac ross s t atic and dynami c pupi l mea su rem ent pa ram ete rs. 256 Dis c us sion 257 This study ev alua ted t h e fea sibili ty of u si ng the Bul bi CAM devi ce i n a c linic al se tti ng, foc u s in g on the 258 reprod uci bility of pu rsuit, sac cad e, and p upi llometry t e st s . In addi tion, a pai red a sse s s me n t be tween 259 Bulbi CAM a nd th e Pup ilLa b s Neon e ye- tr ack ing gl as s e s ex amined t h e agre eme n t betwee n devic e s. 260 Overall, pa rticip a nt f e ed bac k indi cate d a high leve l of ac cep tab ili ty, w ith favou ra bl e rating s f or 261 co mfort, e xamina ti on leng t h , brightn e ss, a nd minim al repor t s of do uble v is i on or eye strain . S mooth 262 pursu it and pup ill omet ry te st s d e mon s tr ated st r ong rep roducibili ty, w hile s ac cad e te s t s s h owe d 263 more v ariabl e r e sul ts . The se findi ng s sup port th e po ten t i a l ev er y day clini cal appli c abili t y of 264 Bulbi CAM a s a f ea sible an d we ll- tole rat e d ey e- t rac king so lution . 265 266 Inter -vi sit Rep rod u c ibi lit y of B ulbiCAM T est s 267 The pursuit t e st s ho wed g ood repr oduci bi lity at 1 0-d egre e, 0 .5 Hz t e st freq uenc y . The r e prod ucibil ity 268 of th e Bul bi CAM wa s a s se ssed u sing t h e gain value of th e pur sui t te st a t a s t imul u s f requ e ncy of 0.5 269 Hz , as pu rsuit ga in i s know n to dec rea se at fr equ enc ie s above 0.5 Hz (17) . Alth ou gh pr e vious s t udi e s 270 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 15 hav e shown t ha t g ain v alue s a t 0. 25 Hz a r e higher t han th ose at 0 .5 Hz in ch ildren , w ith per for m anc e 271 dec lining as targe t vel ocity incre a se s, our finding s c on firm the hig h rep roduci bili ty of the pu rsuit te st 272 at 0.5 Hz . T hi s sugg e s t s th at 0.5 H z i s an i d eal pa ra met er f or r e produc ibili t y as se ss ment s (18 ,19 ). 273 Inte r-vi sit r epr oducibility f or pupi lla ry li ght reac tio n te s t s s how ed g ood r epro ducibility for b a s eli ne -274 pea k co nstric tion diame ter s a nd c on s tric tio n velo city pa ra met er s . I n con tr as t , th e latenc y, pe ak 275 ve locity , and a ccu r a cy p aram et er s o f t he vertica l 9 a nd hori z on ta l 20 sac cad e te st s s how ed onl y 276 modera te t o p oor re pr od ucibil ity. Thi s di scre pa ncy in repro duci bility b e tween the s a cca dic a nd 277 pupil lomet ry te sts c an be a ttrib u ted t o f und ament al d if fe r e nce s in t he phy siologi c al mecha nis m s 278 underly ing the s e ta sk s. Pupil lary li ght r efl exe s a r e pr i ma rily me diat ed by the au to nomi c ne r v ous 279 sys t e m, whic h i s rela tiv ely s tabl e and le s s inf luenc ed by highe r -o rde r c ognitive pr oces se s . 280 In c ontra st, sac cadi c ey e move men t s a re compl ex, i nvolv ing the i nt e gra tion o f s en sor y , c ogni tive , 281 and motor sy stem s . Fac to r s such a s a t t e ntion , le arning e ff ect s, an d deci sion -mak ing a r e know n to 282 influe nc e s ac cadi c pe rformanc e, po ten ti al ly in t roduc ing v ariabil i t y betwe e n visi ts (20-22) . Al though 283 previ ou s s tudie s h av e repo rte d high r e prod ucibil ity fo r sa cc ade t e st s und e r contr olled conditi on s 284 (15), the mode rat e to po or re produc ibili t y o bs erve d in o ur study ma y re flec t indivi dual di ffe renc e s in 285 att entio n a nd enga gem ent a cr oss v i s it s. Notab ly, t he pu rsuit t ask s ex hibi ted le ss varia bility, de s pi t e 286 also b e ing infl u enced by a tten tion and e ng ageme nt . Thi s di ff ere nce may be at tri buted t o the pu r suit 287 sys t e m’ s r eli ance o n co nt i nuo us, r e f l exiv e track ing me chani sm s, whi ch a r e l ess su s c ep tible t o 288 tran s i e nt f luctua tion s in at t e n t i on , co gnit ive engag ement , or phy s i ol ogic al sta te s c ompared t o 289 sacc adic move men t s . A d di tionally , the s tudy de sign, whic h requir ed tw o s e pa rat e visit s, coul d h ave 290 introduc e d varia bili ty du e to le a r nin g e ff ec ts or di ff erenc e s in partici p ant famili ari t y with the ta s k . 291 The us e of a VR -ba sed te sti ng sy stem , su ch a s th e Bulbi CAM, mig ht al s o play a rol e. W hi l e ef fec t i ve 292 for ce r tain me a sure s, it may n ot be op ti mized for c ap tur i ng the fi ne - g r a in ed dy namic s o f s a cca dic 293 ey e move ment s w ith high r e p r o ducibi li ty. Fu t ur e s tudie s c ould ex plore modi fic ati ons to th e te s t i ng 294 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 16 protocol o r alt erna tive s et up s to improv e the r eliabili ty of sac cade me a surem ent s u s in g VR -b a s ed 295 sys t e m s . 296 Paire d As se s sment Bet ween D e vic e s 297 This study de mon stra te s agr eeme n t bet ween devic e s in pupil lome try te s t variabl e s, indica t i ng 298 relia bl e mea s ur ement s ac ro ss the ob s erv ed pa r a me te rs. Bul bi CAM con si s t e n t l y mea s ured l o wer 299 pupil diame ter s f or bot h ba selin e and pe a k c ons t ric t io n stat e s, c ompar ed to Pupi l Labs N eon g la s s e s. 300 The se d if f e r e nce s ca n be a ttribu ted t o v ariation s in pa r tic ipa nt s' head po s i t io ning during the 301 as s e ssme nt s . D uring Ne on gla sse s te s tin g, participa n ts' he ad s were po s i tion ed s l i ghtly ba ck, leadi ng 302 to va r i able b rig ht n e ss l ev el s. In c on t ra s t, BulbiC AM a sse s s men ts en sure d more c o ns i st ent h ea d 303 posi tioning , re s ultin g in brig h t e r l ight ex pos u re and c on s e qu en tly small e r pupil dia meter s, 304 partic ula rly a t t h e peak con st r i ction ph a se. The ba s eli ne pupil diam ete r s were al so meas ured low e r 305 wi t h BulbiC AM, lik ely due t o simila r f a cto rs. Alth oug h the va r i able i s t ermed a s " b as eline " dia met er, 306 repea ted exposu re t o bright l ight stim uli during t h e t r i al s ( as me ntion ed pr ev iousl y on m ethodology 307 sec tion) i s li kely to in flu enc e th e ba s el in e dia mete rs. T hi s ef fec t, kn own a s the p o st-illumi n ation pupi l 308 re spon s e , r e flect s the del a y in r etu rning to ba selin e dilati on a f t e r e xpo sure to a st imulus, whi ch c an 309 take s e ver al s ec ond s ( 23) . Con sequ ently, diff erenc e s in stimulu s ill uminati on l evel s c ont r ibute d to 310 the v aria tion in ba sel ine p upil dia m ete r i n this study . 311 Di screpa nc ie s in ba seli n e mea sur ement s betwee n BulbiC AM a nd PupilL ab s N eon gla sse s a re 312 primarily at tr i bu t e d t o varia t i on s i n head pos i tion ing an d light i llumina n ce, w ith a prev iou sly 313 publi s h ed compa r i son s tudy simila r l y hig hlighting the e f fec t o f ta r g e t illumi nance variab ility on 314 diff er ent mea s u remen t m eth od s (24 ) . De s pi te t he s e fa ct or s , th e obs e r v ed dif fe re nc es in b a s eline 315 and peak con str ic tion pupi l di ame te rs st i l l fall within a cc ep tabl e li mit s i n our s tud y . Sc hmitz e t al 316 (2003 ) pre viously c ompar ed th ree di f fere nt met hod s of pu pil d iame ter me a sure ment a nd s u gge st ed 317 that d ev iation s o f up to 1 mm may be ac ce pt a bl e if c linic ally in signifi can t (25 ) . Con s i der in g t h e 318 findin g s o f t hi s s tudy a long side p rev iou s r e sea r c h and t h e pot ential e ffe c t s o f var ying s timu lu s 319 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 17 ill uminanc e, th e a gre ement betw e en Bul b i CAM an d Pupi lLabs N eon g la s s e s for pu pil diamete r 320 mea s u remen t s remai ns robu st a nd c linic ally applic able. 321 The BulbiC AM sy ste m mea sur e s a high er veloc ity of pu pil c on s t riction compa red to N e on gla s se s. 322 Pupil velo city i s de te r mi ne d by identi fying a c onsisten t c on s t ric t i on o r dilatio n ph as e and calc ulat ing 323 the c hange in pupi l di ame te r or a r e a div i ded by t h e pha s e dura tion (12 ) . Veloc i ty, defin e d a s the 324 ch ange in pupil diamet er pe r s ec ond for this te s t, i s grea ter in Bulbi CA M bec au se the dia me te r 325 ch ange ( a bs ol u t e valu e of b a s eline -pe ak c ons t ric tion di f fer enc e ) is l arger , a nd th e time to r ea ch pea k 326 c on s t r i c t i o n is f a s te r . N e o n g las s es r e c o rd e d a l o n ge r t im e t o c ons t r i c t io n , l i k e ly du e t o l o w e r 327 brightne s s leve l s d u r in g the a s se ssme nt, w hich m ay ha ve c aused a slow er an d more gradual pupi l 328 co nstriction comp ared t o t h e s h a r p e r r e s pons e observe d with Bulb i CAM. There fo re, w hile 329 Bulbi CAM sh ow s hi ghe r vel ocity mea sure ment s , con sid ering t h e diame ter c hang e and time to 330 co nstriction var iab l e s u se d to calc ula te v eloc ity, Bulbi CAM’ s v el oci t y mea sur eme nt is sti ll ac cept able 331 and r e liabl e f or pupil t e sting. 332 The se re sul ts r e in for c e the r el iabili ty of b oth dev ice s f or pupil lome try a s s e ssmen t s w hile 333 ac know ledgin g the infl u ence o f expe r im e ntal c ondi tion s on mea s u reme nt ou tc o mes . 334 L i m i ta ti o n s 335 As a lim iti ng fac to r o f t hi s study, r eli abili t y was as se ss ed solely f or t h e pair ed c ompari son o f the pu p il 336 te st betwee n Bulb iC AM and P upilL abs Neon glas s e s . No e valua t i on s i nvolv ing oth er dev ice s or t e st s 337 we r e conduc t ed, a s r eplic ating t h e exa ct st i mulu s wa s no t fe a sible with a t radi t i on al infra red ey e 338 track er. 339 Addition a lly, the r e cen t upda t e s to the d evic e so ft w a re, r equir e fu rthe r te s t i ng to as s e ss th eir impa c t 340 on repro duc ibility and r eli abili ty. 341 342 Conc lus ion 343 This study de mon stra te s th e f ea sibility , r eproducibi li ty an d relia bil ity of the Bu l bi CAM a s a tool f or 344 both c linic al a nd r e s ea r c h app lica t i o n s . P a tien t f eedba ck o n t he Bulbi C A M e xamin ation wa s hi ghly 345 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 18 posi tive. In te r -v i s i t repro duc ibili t y wa s hig h for pur s ui t a nd pupil te s t s, s up p or ting the dev ice' s 346 co ns i st ency . Whil e s ac cad e s show ed poo r to moder at e r ep roduc ibili t y , thi s highli ghts a r e as for 347 poten tial re fin emen t an d inve stiga tin g fu rt h e r . The pa i r e d a s s e s smen t be twee n d evi ces furthe r 348 va lidated the acc uracy o f key pupill omet r i c pa r a me ter. The se r e sult s und e rsco r e t he po ten tial o f 349 Bulbi CAM a s a reli abl e and p atie nt -f r ie nd ly de vic e for ocul omo tor and pup il lomet r ic ev alua t i on s i n 350 both c linic al a nd r e s ea r c h s etti ngs. 351 The finding s of thi s s t ud y provide a foun da t i on for fut u re res earc h in to th e a pplic a t i on o f VR -ba s ed 352 ey e- t rac king s y s t em s ac r o s s variou s di sor der s a nd c onte xt s . In fut ur e , we f o re see por tabl e V R b a sed 353 hea d s e t e ye tr ack ing de vic es pl ayin g a si gni ficant r ole in mee t in g the inc rea sing n ee d f or imagi ng 354 ac r o ss dive rse se tting s, con firmin g the p oin t-of -ca r e appr oac h. 355 356

s 357 1.Oyam a A, Ta ked a S, It o Y, N a kajima T, T akami Y, Ta key a Y, e t al. Novel M etho d for Ra pid 358 Ass es s m e nt o f Cogni tiv e Impa i r me nt Usi ng Hig h-Pe rfo r ma nce E ye -Trac king Tec hnology . Sc i Rep 2 019 359 Se p 10;9(1 ):129 32–x. 360 2.Ki m M, Le e J, Le e SY , H a M, Pa rk I, J ang J, et al . D ev elopmen t of a n eye -t r ac king s y stem ba sed on a 361 dee p le arning model to a sse s s e xec ut i ve func tio n in pa ti ent s w ith men tal illne s se s. S ci Rep 2024 Aug 362 6;14(1 ):1 8186–2. 363 3.L eigh RJ , Kenn ard C . Using s ac cad e s a s a re s ea r c h tool i n the cl inica l neu ro scien c es . Brain 364 2004 ;127(3) :460– 477. 365 4.Xu Y, Zha ng C, P an B, Y uan Q , Zha ng X. A p or t abl e and e f f i cien t de men tia scre en i ng tool usi ng eye 366 track ing ma chine lea rning and v irt u a l rea lity . N P J D ig i t Med 202 4 Aug 22 ;7(1 ):219 –5. 367 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 19 5.Mor eno -Arjonill a J, Ló pez-Rui z A, Jim é n ez - Pé rez J R, C al leja s- Aguile ra JE, Ju rado JM. Ey e -t r ac king on 368 vi r tual r e ality: a s u r v ey. Vir tual Re ality 20 24 -02-0 5;28 (1) . 369 6.Hirot a M, Ka to K, Fuk us hi ma M , Ik eda Y, H a yashi T, Miz o ta A. Analy si s of smoo th purs ui t e ye 370 move ment s in a c linic al c ont ext by tr a cki ng the targ et a nd eye s . S ci Rep 2022 May 19 ;12(1):8501 – 6. 371 7.THE PO WER OF S IG H T: U S I N G EY E TRACKIN G TO ASSE SS L EARNI NG EXPE RIEN CE (LX) IN V IRT U AL 372 REAL ITY ENVIR ON ME NTS . 11th Int erna ti ona l Tec hnology , Educ atio n a nd D evel op ment Con f erenc e; 373 6-8 Ma r c h, 2017 ; INTED2017 Proc e eding s : 2 017. 374 8.Ba r k evic h K, Ba il ey R, Diaz G J. Usi ng De ep Lea r ni ng to Inc r e ase Eye-T rac king Robust ne ss, Ac cur acy , 375 and Pre cisi on in Virtua l R eal it y . Proc AC M C omput G rap h Int erac t Te ch 20 24 -05- 17;7(2 ):1 . 376 9.Bul bit ech us er man u al. 20 23; Avai lab le at: ht t p s: // bu l bit ech.c om/ . 377 10.Ma han ama B, J aya wardana Y, R eng ar ajan S, Jay award ena G , Chuko skie L, Snid er J, e t a l. Ey e 378 Move ment an d Pu pil M ea sur es: A Revie w. F r onti e rs in Co mpu t e r Sci e nce 2022;3. 379 11.P upil L abs. Pupil L abs Ne on. 202 4; Av ai lab le a t: ht t p s: //pupi l-l abs.com /produc ts / ne on . 380 12.Holmq vis t K. Eye tr a ckin g : a compreh ensiv e guid e to me thod s and me asure s . Oxford : Oxfor d 381 Univer sity Pr e s s ; 2011. 382 13.Bl ekhe r T, W e ave r MR, C a i X, Hui S , M arsh all J, Jac ks on J G, e t al. Te st-r et e st re li abili t y of sacc adic 383 mea s u re s in subjec ts a t ri sk for Hu ntingt on di sea s e . Inve st Oph t h a lmol Vi s Sc i 20 0 9 384 Dec;5 0 (12):5707 – 5711. 385 14.Wei r JP . Q uan tify ing te s t - re te st reliabi lity using th e intra cla s s correl atio n coef fi c ient and t h e SEM. 386 J Str ength C o nd Re s 2005 Feb;19(1) :23 1– 240. 387 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 20 15.Nij B ijvank JA, P etzol d A, Bal k L J , Tan HS, U itde ha ag BM J, T heodo rou M, et a l . A s t a nda rdized 388 protocol for q u anti fica ti on o f sa cc adic eye moveme nts: DEMoNS . PL oS One 20 18 J ul 389 16;13 (7) :e020069 5. 390 16.Ma rtin Bl and J, Al tman D. STATIS TI CA L MET H ODS F O R AS SESS ING AG REEMEN T BET W E EN TWO 391

S O F CLI NIC AL MEASUREMENT . The Lanc et 198 6;327(8476 ) :3 07–310 . 392 17.Ma rtin s AJ , Kowl er E, Pa lm er C . Sm oo t h pursuit o f small - a mplitud e sinu soidal motion. J Opt S oc 393 Am A 198 5 Feb;2(2 ):234–24 2 . 394 18.S alman MS, Sh arpe JA, Li llaka s L , D en nis M, S tei nbach MJ . Smo oth pu rs uit eye moveme nts in 395 ch ildren. Ex p Brain R e s 20 06 Fe b;169(1 ): 139– 143. 396 19.S chröde r R, K eidel K, Tra utne r P, R a dbr u ch A, E t ting er U . N eu ral mech ani sm s of b ack ground and 397 ve locity ef fect s in smoo th pur s u i t eye m o vement s. Hum Br ain Ma pp 2023 Feb 15 ;44(3) :1002– 1018. 398 20.Ge r sch TM, Kowler E , Sc hnitzer BS , Do s he r BA . A t ten tion d ur i ng sequ enc e s o f sac ca de s along 399 marke d and me morized pa th s . Vis i o n Re s 2009 Jun;4 9 (10) :1256– 1266. 400 21.Hutt on SB. Cognitiv e con trol o f sac ca di c ey e mov ement s . B r a in C ogn 2008 De c ;68(3):327 – 340. 401 22.P ie rce, J .E ., Cleme n t z, B .A ., Mc D owell , J.E. Sa cc ad e s: Fund a ment al s and Ne ural M e chani s ms . In: 402 Klei n, C. , E t ting e r, U , e dito r. Eye Movem en t Re sea r c h : Spring er , Cham ; 2 019. 403 23.Ma t h ôt S . Pupil lom etry: P sych ology, P hysiolog y, and Func ti on. J Cogn 2018 F eb 21;1(1):1 6 . 404 24.Wic krema singhe SS , Smith GT, S teven s J D. Compari so n of dy namic digita l pupil l ometr y and st atic 405 mea s u remen t s o f pupil s i ze i n de termini ng s c otopic pupil size be for e re frac tiv e su r g er y . J Cat arac t 406 Ref rac t Surg 2005 Jun ;31(6 ):1171 –1176 . 407 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 21 25.S chmitz S, Kr um menau er F , He nn S, Dic k HB. Co mpari s o n of t hr e e dif fe ren t tec hn ologie s fo r pupil 408 dia mete r mea sure men t. G r a e fe s Arc h Cli n E xp Ophthalmo l 2003 J un ;241 (6) :472– 4 77. 409 410 Ack nowledge ment s: 411 We tha nk Vince n t D w ian ta a nd Gur t ek Sam r a fo r t he i r in s i ght on P upi lLabs Ne on g las s e s se tup. We 412 thank De nni s H e n s a nd t he Bulbi CAM t e c hnic al tea m for troub l e shooti ng techni ca l chal lenge s a s 413 requir ed du r in g th e s tudy. 414 Author contr i but ions 415 MT, IK : study de sign; MT: s up ervi s i on of the projec t ; IK, R M: da ta colle cti on; MT, I K, HY: da t a 416 ex t rac tio n ana ly sis; IK, RS: d r a f tin g th e or igin al man us c r i p t ; M T , EML, Q A : fun di ng ac quisi tion; MT, 417 IK, E ML, QA , VR , HY, RS , RM, RS , ZT: c r i t i cal r e view of t he manu s c r ip t. 418 All a utho rs hav e gi ven thei r a pproval for the publi sh ed ve rsion o f t he man usc ript . 419 Conflic t of Int e r est 420 No con f l ic ting r e la tion ship ex i s t s for any a uthors . 421 Fundi ng 422 This study w as s up por ted by the M edic al Re sea r c h Counc il (M RC ), L ondon , UK (gr ant num ber: 423 MR/X502777/1 ) an d th e Ulver scro ft F ou ndatio n . 424 425 426 Summa r y 427 What wa s k nown be for e ? 428 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint 22 • Ey e-t r ac king devi ces a re e s se ntial i n n eur o-oph thalmic re s earch a nd cl inica l p racti c e, 429 of fering k ey i n s ig ht s in t o ocul a r a nd n eur olog ical functi on. 430 • VR hea dset -ba s ed e ye t rack er s o ff er ob je ctive as se ssm ent me thod s in a porta bl e, al l-in -on e 431 co mpact sy st em. 432 • VR ey e t rack ing sy ste ms a r e und er -expl o r e d compa re d to t r a di tional sy stem s and no 433 reprod uci bility or r eli ability s tu die s ha ve bee n con duct ed on VR head s et -ba se d ey e tr a cke rs 434 to dat e. 435 What t his study add s? 436 • Par tic ipant fe edba ck highlig hted hig h ac c eptabil i ty; whi le pa ire d a s s e s s m en t be twee n 437 dev ices (Bulbi CAM an d Pupi lLab s N eon e y e track ing g las s e s ) showed st rong ag r e e ment 438 co nfir m ing Bulbi CAM’s relia bili ty. 439 • VR hea dset e ye trac king d evic e Bulbi cam de monstra ted h igh r eprod uc ibility f o r p ursui t 440 and pupil te s t s ; w hile sac cad e te s t s h ad l ower rep roduc ibility ou t c ome s. 441 • In fu ture , port able VR he ad set s coul d pl a y a ke y role i n expa nding point -o f- c a re 442 as s e ssme nt s in va r io u s se tting s. 443 444 445 446 447 448 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2025. ; https://doi.org/10.1101/2025.06.18.25329788doi: medRxiv preprint