Analiza vpliva ekspozicijskih pogojev na kakovost slik pri računalniški radiografiji : diplomsko delo

Vodopivec, Lea (author), Žakelj, Tadeja (author), Žibert, Janez (mentor), Mekiš, Nejc (co-mentor)

Uvod: Kvaliteta rentgenske slike temelji na dobrem prikazu struktur in je posledica dobre kontrastne in prostorske ločljivosti. Z računalniško obdelavo lahko sliki dodamo večjo diagnostično uporabnost in boljšo kakovost. Namen: Zanimalo nas je, kako ekspozicijski pogoji in obdelava slik vplivajo na njihovo kakovost ter ali so slike z višjim razmerjem signal-šum bolj kakovostne od tistih z nižjim. Metode dela: Pregledali smo literaturo in pridobili rentgenske slike prsnih in trebušnih organov fantoma z različnimi stopnjami počrnitve. Slike smo obdelali z glajenjem oziroma izboljševanjem kontrastne ločljivosti v programu ImageJ. S pomočjo treh ocenjevalcev in novega programa namenjenega tovrstnem ocenjevanju slik smo pridobili subjektivne ocene slik na podlagi zastavljenih kriterijev ocenjevanja. Slikam smo v programu ImageJ izračunali razmerje signal-šum. Naredili smo statistično analizo podatkov in tako pridobili rezultate. Rezultati: S statistično analizo ANCOVA smo pri trebušnih organih ugotovili, da ekspozicijski pogoji enako vplivajo na kakovost slike pri kontrastni izboljšavi in glajenju slik (p=0,189) ter drugače pri neobdelanih slikah (p<0,001). Z enako analizo smo pri prsnih organih ugotovili, da obdelava slik statistično značilno ne vpliva drugače na kakovost slik (neobdelane slike, p=0,816; glajenje, p=0,641). Pri trebušnih organih smo pri odvisnosti kakovosti prikaza slike od razmerja signal-šum in obdelave slike z ANOVO izračunali, da obstajajo razlike pri obdelavi slik (p<0,001). S post-hoc testi smo pokazali statistično značilno razliko med neobdelanimi in kontrastno izboljšanimi slikami (p<0,001) ter glajenimi in kontrastno izboljšanimi slikami (p<0,001). Med neobdelanimi in glajenimi slikami ni bilo razlike v razmerju signal-šum (p=0,068). Enako analizo smo ponovili tudi pri prsnih organih, kjer smo izračunali, da obdelava slik statistično značilno ne vpliva na spremembe razmerja signal-šum (p=0,057). Razprava in sklep: Ugotovili smo, da obdelava ne vpliva na kakovost prikaza slike. Ugotovili smo tudi, da ekspozicijski pogoji pozitivno vplivajo na kakovost prikaza slike. Pri originalnih slikah kakovost narašča z naraščanjem razmerja signal-šum, pri glajenih je kakovost ves čas enaka, pri kontrastnih pa kakovost slik z naraščanjem razmerja signal-šum pada. Pri primerjavi med originalno in glajeno sliko je boljša glajena, saj je mediana pozitivna. Mediana pri primerjavi originalne in kontrastne slike pa je negativna, zato je originalna slika boljša.

• Language: Slovenian
• Keywords: računalniška obdelava slike, razmerje signal-šum, prsni organi, trebušni organi
• Work type: Bachelor thesis/paper
• Organization: ZF - Faculty of Health Sciences
• Year: 2017
• PID: 20.500.12556/RUL-94576
• COBISS.SI-ID: 5341291
• Publication date in RUL: 05.09.2017

Secondary language

• Language: English
• Title: Analysis of impact of exposition conditions on image quality in computed radiography : degree paper

Abstract

Introduction: The quality of a radiograph is based on a good display of structures and is a consequence to good spatial resolution. The computer treatment can add a greater diagnostic usefulness and better quality to the radiographic image. Purpose: We were interested how the exposition conditions and treatment of radiographs influence on their quality and whether the radiographs with the higher signal-to-noise ratio possess better quality to those with the lower ratio. Methods: We have examined some literature and gained radiographs of chest and abdominal organs of a phantom with various levels of blackening. The images were treated with smoothing or improvement of contrast resolution in the ImageJ programme. With the help of three assessors and a new programme intended for such assessment of images, we have gained subjective evaluations of images on the basis of prepared assessment criteria. We have calculated the signal-to-noise ratio of the images in the ImageJ programme. The statistical analysis of data was also performed in order to get results. Results: In the abdominal organs, the statistical analysis ANCOVA has shown the exposition conditions equally influence the quality of image as the contrast improvement and smoothing of images (p=0.189) and differently in untreated images (p<0.001). The same analysis in chest organs established that treatment of images does not statistically significant influence differently on the quality of images (untreated images, p=0.816; smoothing, p=0.641). In abdominal organs in the dependence of the quality of image display to the signal-to-noise ratio and treatment of image with ANOVO, we have calculated there are some differences in the image treatment (p<0.001). The post-hoc tests have shown statistically significant difference between untreated and contrast improved images (p<0.001) and smoothed and contrast improved images (p<0.001). There was no difference between untreated and smoothed images in the signal-to-noise ratio (p=0.068). The same analysis was repeated also in chest organs, where we have calculated that treatment of images does not statistically significant influence on the change of the signal-to-noise ratio (p=0.057). Discussion and conclusion: We have established that treatment does not influence the quality of image display. We have also determined that exposition conditions positively influence on the quality of image display. In original images, the quality increases with the increasing of the signal-to-noise ratio, the quality in the smoothed images remains the same; however, the contrast images show decrease of the quality of image display with the increasing signal-to-noise ratio. In comparison between original and smoothed image, the smoothed is proved better, since the median value is positive. On the other hand, the median value in comparison between the original and contrast images is negative; therefore, the original radiographic image is better.

• Keywords: computer treatment of radiographic images, signal-to-noise ratio, chest organs, abdominal organs