Avtonomna naprava za test devetih zatičev

magistrsko delo

Jaka Jemec (Author), Žiga Špiclin (Mentor), Žiga Bizjak (Co-mentor)

Abstract

Test devetih zatičev (angl. nine hole peg test; NHPT) je bil razvit za vrednotenje spretnosti prstov, znane tudi kot fina ročna spretnost. Osnovne meritve vključujejo čas izvajanja z dominantno in nedominantno roko. Uporablja se lahko pri širokem spektru populacij, vključno z bolniki z možgansko kapjo, multiplo sklerozo, Parkinsonovo boleznijo, ipd. Poleg tega je NHPT sorazmerno poceni in ga je mogoče enostavno izvesti z uporabo štoparice. Vendar pa zaradi povečanega obsega dela zdravstvenemu osebju primanjkuje časa za redno opravljanje NHPT. Da pa test lahko zazna spremembe v motoriki, ga je potrebno izvajati pogosto in pod istimi pogoji. Avtonomen test bi omogočil pogosto in ponovljivo spremljanje stanja bolnika brez dodatne obremenitve zdravstvenega osebja. Za avtonomen test devetih zatičev (ANHPT) smo razvili tako strojno kot programsko opremo. S 3D tiskalnikom smo natisnili testno ploščo s centralno posodico za zatiče in levo-desno simetrično postavljeni 3x3 mreži luknjic, s čimer smo odpravili potrebo po obračanju testne plošče. V ploščo smo vgradili elektronsko vezje, ki na osnovi parov svetleče in sprejemne fotodiode zaznava prisotnost zatiča v vsaki od 18ih luknjic. Na ploščo smo namestili s 3D tiskalnikom natisnjeno pravokotno stojalo za tri kamere. ANHPT se izvaja na osebnem računalniku in se upravlja preko grafičnega uporabniškega vmesnika. Ta na uporabniškem nivoju omogoča vnos podatkov o preiskovancu, izvajanje NHPT po standardnem protokolu in izpis meritev, v ozadju pa zagon in komunikacijo z mikrokrmilnikom ter branje stanj sprejemne fotodiode in pa kalibracijo, zajem in analizo videoposnetkov iz treh kamer. Meritev časa skupnega časa vstavljanja in pospravljanja devetih zatičev za dominantno in nedominantno roko ANHPT pridobi iz elektronskega vezja in na osnovi analize videoposnetkov. Poleg tega smo pridobili tudi meritve za fazi vstavljanja in pospravljanja zatičev. Pravilnost meritev z ANHPT smo naprej validirali z zdravimi osebami glede na ročne meritve s štoparico. Ugotovili smo, da je merjenje časa z elektronskim vezjem bolj točno kot zajemanje časa z avtomatsko analizo videoposnetkov. Povprečni napaki glede na ročne meritve sta bili in -0,14 in -1.01 sekunde. Ker pa lahko iz videoposnetkov izluščimo dodatno informacijo, kot so tresavica roke, položaj zapestja in 3D trajektorijo gibanja roke ter zaznamo morebitno napačno izvedbo testa, predlagamo da čas izvajanja testa merimo z elektronskim vezjem, izmerjene čase uporabljamo kot pomožno informacijo za analizo videoposnetkov, s čimer iz videoposnetkov lahko bolje izluščimo relevantno in dodatno informacijo o stanju bolnika in potrdimo pravilnost izvedbe testa. Tekom validacije smo pokazali, da so ročne meritve s štoparico posameznega in različnih ocenjevalcev obremenjene s povprečno absolutno napako 0,26 sekunde, maksimalna napaka pa je bila 0,71 sekunde. Glede na povprečni čas izvajanja testa za zdravo osebo, ki je okoli 20 sekund je to sicer relativno majhna napaka. Opravili smo tudi preliminarno validacijo ANHPT v kliničnem okolju na zdravih osebah in bolnikih z multiplo sklerozo, pri čemer smo želeli potrditi zmožnost razvite naprave, da zazna razlike v meritvi časa med zdravimi osebami in bolniki z multiplo sklerozo. Pridobljene meritve se ujemajo iz rezultati znanstvenih člankov, in sicer potrebujejo bolniki z multiplo sklerozo v splošnem več časa (povprečna razlika 7,42 sekund) za opravljanje testa od zdravih oseb. Razlike med osebami v skupini bolnikov so večje kot pri zdravih, npr. razlike v motoriki, razlike v stopnji bolezni, kar se izraža v večjem raztrosu meritev časa, in sicer 6,69 sekund za bolnike z multiplo sklerozo in 1,89 sekund za zdrave osebe. Analiza povprečnih časov in raztrosa meritev časa ločeno glede na fazi vstavljanja in pospravljanja zatičev kaže, da je razlika med bolniki in zdravimi osebami bolj izražena v fazi sestavljanja, kar se sklada z opažanji iz znanstvene literature. Naša ANHPT naprava se je izkazala za točno, zanesljivo in robustno pri meritvah časa, poleg standardne meritve celotnega časa testa pa ima še dodatne možnosti meritev časa vstavljanja in časa pospravljanja, diagnostike pravilnosti izvedbe testa, izrisa trajektorije premikanja roke v prostoru in njeno nadaljnjo analizo, itd. Bistvena prednost ANHPT je standardizacija testa, saj so meritve vedno opravljene po istem protokolu. V trenutni praksi s klasičnim testom se protokol lahko od ocenjevalca do ocenjevalca precej razlikuje, zato rezultati med ocenjevalci in/ali ustanovami pogosto niso primerljivi. Predvidevamo, da bi uporaba ANHPT vrnilo točnejše in medsebojno bolj primerljive meritve med različnimi ustanovami in medicinskimi timi, s tem pa bi se ANHPT lahko uveljavil kot orodje za redno spremljanje stanja bolnikov.

Keywords

test devetih zatičev;meritev časa;analiza videoposnetkov;detekcija roke v prostoru;grafični uporabniški vmesnik;kvantitativna validacija;klinična študija;magisteriji;

Data

Language:	Slovenian
Year of publishing:	2023
Typology:	2.09 - Master's Thesis
Organization:	UL FE - Faculty of Electrical Engineering
Publisher:	[J. Jemec]
UDC:	004(043.3)
COBISS:	142479363
Views:	36
Downloads:	7
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary title:	Autonomous nine hole peg test device
Secondary abstract:	Nine Hole Peg Test (NHPT) was developed to evaluate finger dexterity, also known as fine motor skill. The basic measurements include the time required to perform the test using the dominant and non-dominant hand. It can be used for a wide range of populations, including patients after stroke, multiple sclerosis, Parkinson's disease, etc., to monitor disease status and/or progression. In addition, the NHPT is relatively cheap and can be easily performed using a stopwatch. However, due to the increased workload, health professionals lack the time to regularly perform the NHPT. An autonomous test would allow frequent and repeatable monitoring of the patient's status without additional burden on the healthcare professionals. We have developed both the hardware and software for the Autonomous Nine Hole Peg Test (ANHPT). A 3D printer was used to print a test plate with a central peg container and left-right symmetrically aligned 3$\times$3 grid of holes, thus eliminating the need to rotate the plate when switching hands. The plate was equipped with an electronic circuit that detects the presence of the peg in each of the 18 holes based on pairs of light-emitting diodes and receiving photodiodes. A rectangular stand for three cameras was also printed using a 3D printer and mounted on the plate. The ANHPT is controlled through a graphical user interface on a personal computer. This interface allows the entry of patient data, performing the NHPT according to the standard protocol, and printing out the final measurements, while in the background the software runs and communicates with the microcontroller, reads the state of the receiving photodiodes, and calibrates, captures, and analyzes video footage from the three cameras. The measurement of the total time for inserting and removing the nine pegs for the dominant and non-dominant hand can be obtained from the electronic circuit or based on video analysis. In addition, we have obtained time measurements separately for the phases of inserting and removing the pegs. The accuracy of the ANHPT measurements was validated on healthy subjects and compared to manual measurements with a stopwatch. We found that measuring time with the electronic circuit was more accurate than capturing time with automatic video analysis. The average errors compared to manual measurements were -0.14 seconds with electronic circuit measurement and -1.01 seconds when measuring was based on recorded video analysis. However, since additional information can be extracted from the video, such as hand tremors, wrist position, and 3D hand movement trajectory, and incorrect test execution can be detected, we suggest the time to perform the test to be measured using the electronic circuit. These times can be used as auxiliary information for video analysis, thus enabling better extraction of relevant and additional information about the patient's status and to validate test execution. During evaluation, we showed that manual measurements with a stopwatch by individual and different evaluators are affected by an average absolute error of 0.26 seconds, with a maximum error of 0.71 seconds. Given the average time for a healthy subject to perform the test, which was around 20 seconds, this is relatively a small error. We also carried out a preliminary validation of the ANHPT in a clinical environment with healthy subjects and patients with multiple sclerosis, with the aim of confirming the ability of the ANHPT to detect differences in the time measurement the two groups. The obtained measurements matched the results of scientific articles; for instance, patients with multiple sclerosis generally took longer (average difference 7.42 seconds) to perform the test compared to healthy subjects. Differences among the patient group were larger than among healthy subject group, owing to differences in motor skills and in the variability of the stage of the disease. The latter reflected in a larger spread of time measurements, i.e. 6.69 seconds for patients with multiple sclerosis versus 1.89 seconds for healthy subjects. Analysis of the average times and standard deviation of measurements separately for the insertion and removal phases of the pegs showed that the difference between patients and healthy subjects was larger in the peg insertion phase, which is in line with observations from the scientific literature. Our ANHPT device has proven to be accurate, reliable and robust in measuring time. In addition to the standard measurement of the total test time, it also provides additional measurements of the insertion and removal times, option to diagnose the validity of the performed test trial, plotting of the trajectory of hand movement in space and further analyses. The essential advantage of the ANHPT is the standardization of the test, as measurements are always performed according to the same protocol. In current practice with the classical test, the protocol can vary greatly from evaluator to evaluator, so results between evaluators and/or institutions are often not comparable. We assume that the ANHPT would provide more accurate and comparable measurements across institutions and medical groups and thus further establish itself as a viable tool for regular monitoring of patients' status.
Secondary keywords:	Nine hole peg test;time measurement;video processing;hand detection;graphic user interface;quantitative validation;clinical study;
Type (COBISS):	Master's thesis/paper
Study programme:	1000316
Embargo end date (OpenAIRE):	1970-01-01
Thesis comment:	Univ. v Ljubljani, Fak. za elektrotehniko
Pages:	XXVIII, 82 str.
ID:	18027357