Original scientific paper (Izvorni znanstveni članak)
Branko Kovalisko1,3, Damir Štimac2, Frane Mihanović3
1 “Dr. Juraj Njavro” National Memorial Hospital, Vukovar, Croatia
2 Josip Juraj Strossmayer University of Osijek – Faculty of Medicine, Osijek, Croatia
3 University of Split, Faculty of Health Sciences, Split, Croatia
Corresponding author: Branko Kovalisko, email:
DOI: https://doi.org/10.55378/rv.49.2.1
Abstract
Introduction: Hounsfield Units (HU) on CT represent a quantitative measure of X-ray attenuation in tissue – 0 HU is defined for water, and −1000 HU for air. On a 3D C-arm (cone-beam flat-panel detector imaging), the degree of attenuation is displayed by shades of Gray Scale Values (GSV) in the device’s DICOM image. If a correlation between HU and GSV could be proven, the in vivo or in situ application of mobile radiological devices could be expanded outside hospital settings (e.g., in field forensic investigations), while enabling quantitative image analysis.
Aim: To investigate the relationship between the GSV on a mobile 3D C-arm and the HU on a standard CT scanner when imaging the same object (a phantom). Correlation between the mean values of GSV and HU was tested, as well as the difference in the measured GSV and HU values in regard to the DICOM viewer used.
Methods: A standard CT calibration quality control phantom was used in this experimental study. The phantom was scanned using a 64-slice medical CT scanner (Siemens Somatom Perspective, 512×512 matrix, 12-bit scale) and a mobile 3D C-arm (Ziehm Vision RFD, 320×320 matrix, 16-bit scale). Scans were performed at 80 kV and 110 kV on the CT and 80 kV and 115 kV on the C-arm. DICOM images were stored and analysed on two independent DICOM software viewers (Onis 2.5 and RadiAnt) to obtain numerical GSV and HU values. Measurements were taken at eight defined ROI areas (P1 – P8) within the phantom (approx. 1 cm2 each), identically located on the CT and C-arm images. For each ROI, the mean GSV (C-arm) and HU (CT) along with the corresponding min, max, and standard deviation were determined. The obtained data were compared and statistically analysed using the Pearson correlation coefficient (with a significance level of p < 0.05).
Results: The analysis shows a positive correlation between GSV and HU, but with varying intensity depending on the device voltage. At the same tube voltage on the CT and C-arm (80 kV), a weak positive correlation was found (r = 0.450; p > 0.05). A similarly weak, but consistent, correlation (r = 0.434; p > 0.05) was found between the CT at 110 kV and the C-arm at 80 kV. In contrast, at higher voltages (CT 110 kV and C-arm 115 kV), the correlation was very low, non-significant positive (r = 0.185; p > 0.05). No significant difference was observed in the measurement results between the two different DICOM viewers, suggesting that the choice of software did not affect the obtained values. However, none of the correlations found were statistically significant.
Conclusion: The results suggest that a relationship exists between the shades of gray scale on the 3D C-arm and HU on the CT, but the correlation is of weak intensity. The observed correlation is insufficient for reliable quantitative calibration of the C-arm gray scale to the CT HU scale. Therefore, the hypothesis (of no correlation) cannot be confidently rejected. Further research is needed with different phantoms and improved standardization to determine whether mobile CBCT/C-arm systems with appropriate calibration can provide quantitative information comparable to CT. If a strong correlation or the possibility of converting GSV into pseudo-HU is confirmed, it would open up wider applications for 3D C-arms in clinical practice (e.g., implant planning, bone density determination) and field forensic analysis.
Keywords: C-arm; CT phantom; Hounsfield units; gray scale; quantitative analysis
Validacija sive skale na C-luku u korelaciji s HU na CT-u
Sažetak
Uvod: Hounsfieldove jedinice (HU) na CT-u predstavljaju kvantitativnu mjeru oslabljivanja rendgenskih zraka u tkivu – 0 HU definiran je za vodu, a -1000 HU za zrak. Na 3D C-luku (konusno snimanje s ravnim detektorom) stupanj atenuacije prikazan je nijansama sive skale (GSV) u DICOM slici uređaja. Ako bi se dokazala korelacija između HU i GSV, in vivo ili in situ primjena pokretnih radioloških uređaja mogla bi se proširiti izvan bolničkih uvjeta (npr. u forenzičkim istragama na terenu), uz omogućavanje kvantitativne analize slike.
Cilj: Istražiti odnos između vrijednosti sive skale (GSV) na pokretnom 3D C-luku i (HU) na standardnom CT uređaju pri snimanju istog objekta (fantoma). Istraženo je postoji li statistički značajna korelacija srednjih vrijednosti GSV-a i HU-a i ima li razlike u izmjerenim vrijednostima GSV i HU bez obzira na korišteni DICOM preglednik.
Metode: U okviru eksperimentalnog istraživanja korišten je standardni kalibracijski CT fantom za kontrolu kvalitete. Fantom je snimljen 64-slojnim medicinskim CT uređajem (Siemens Somatom Perspective, 512 × 512 matrica, 12-bitna skala) i 3D mobilnim C-lukom (Ziehm Vision RFD, 320×320 matrica, 16-bitna skala). Snimanja su provedena pri 80 kV i 110 kV na CT-u te 80 kV i 115 kV na C-luku. DICOM slike su pohranjene i analizirane na dva neovisna softverska DICOM preglednika (Onis 2.5 i RadiAnt) radi dobivanja numeričkih vrijednosti GSV i HU. Mjerenja su vršena na osam definiranih ROI područja (P1–P8) unutar fantoma (cca 1 cm² svako), identično lociranih na CT i C-luk slikama. Za svako ROI određena je srednja vrijednost GSV-a (C-luk) i HU-a (CT) zajedno s pripadajućim min, max i standardnom devijacijom. Dobiveni podatci uspoređeni su i statistički analizirani uz pomoć Pearsonovog koeficijenta korelacije (uz razinu značajnosti p < 0,05).
Rezultati: Analiza pokazuje pozitivnu korelaciju između GSV-a i HU-a, ali različitog intenziteta ovisno o naponu uređaja. Pri istim naponima cijevi na CT-u i C-luku (80 kV) utvrđena je slaba pozitivna korelacija (r = 0,450; p>0,05). Također slaba, ali dosljedna korelacija (r = 0,434; p > 0,05) nađena je i između CT-a na 110 kV i C-luka na 80 kV. Nasuprot tome, pri višim naponima (CT 110 kV i C-luk 115 kV) korelacija je bila vrlo niska, neznačajna pozitivna (r = 0,185; p > 0,05). Nije uočena bitna razlika u rezultatima mjerenja između dva različita DICOM preglednika, što upućuje da odabir softvera (Onis vs. RadiAnt) nije utjecao na dobivene vrijednosti. Međutim, sve utvrđene korelacije statistički nisu značajne.
Zaključak: Rezultati upućuju na postojanje određenog odnosa između nijansi sive skale na 3D C-luku i HU na CT-u, no korelacija je slabog intenziteta. Uočena povezanost nedovoljna je za pouzdanu kvantitativnu kalibraciju sive skale C-luka u HU skalu CT-a. Stoga se pretpostavka o nepostojanju korelacije ne može sa sigurnošću odbaciti. Potrebna su daljnja istraživanja s različitim fantomima i poboljšanom standardizacijom kako bi se utvrdilo mogu li pokretni CBCT/C-luk sustavi uz odgovarajuću kalibraciju davati kvantitativne informacije usporedive s CT-om. Potvrdi li se čvrsta korelacija ili mogućnost pretvorbe GSV-a u pseudo-HU, otvorila bi se šira primjena 3D C-lukova u kliničkoj praksi (npr. planiranje implantata, određivanje gustoće kosti) i forenzičkim analizama na terenu.
Ključne riječi: C-luk; CT fantom; Hounsfieldove jedinice; siva skala; kvantitativna analiza
