Yapay zekâ algoritması ile gerçekleştirilen anteroposterior sefalometrik analizin bilgisayar destekli dijital analiz ile kıyaslanması
[ X ]
Tarih
2024
Yazarlar
Dergi Başlığı
Dergi ISSN
Cilt Başlığı
Yayıncı
Kırıkkale Üniversitesi
Erişim Hakkı
info:eu-repo/semantics/openAccess
Özet
Tez çalışmamızın amacı; yapay zekâ destekli bir sefalometri çizim programı olan WebCeph programının, anteroposterior sefalometrik ölçümlerindeki güvenilirliğinin ve doğruluğunun manuel olarak dijital ortamda Dolphin programı (v.12.0) yardımıyla gerçekleştirilen ölçümlerle karşılaştırılarak değerlendirilmesidir. Bu amaçla arşivde bulunan tedavi öncesinde teşhis ve tedavi planlaması amacıyla alınmış toplamda 50 hastaya ait anteroposterior sefalometrik radyografiler dahil edilme kriterlerine uygun bir şekilde seçilmiştir. Çalışmamız, toplamda 23 adet anatomik landmark ve 21 adet ölçümün yer aldığı Grummons analizine göre gerçekleştirilecek şekilde planlanmıştır. Öncelikle Dolphin programıyla ölçümleri yapılan radyografiler daha sonra web tabanlı WebCeph programına yüklenip kalibrasyonu el ile gerçekleştirilmiş ve yapay zekâ dijitalleştirmesi aracılığıyla analizleri tamamlanmıştır. Daha sonra elde edilen analizler dijital ortamda çakıştırılarak Dolphin ve WebCeph programları aracılığıyla belirlenen aynı anatomik landmark arasındaki mesafe x ve y koordinatlarına göre ImageJ programı yardımıyla ölçülerek kaydedilmiştir. Ardından gözlemci içi güvenilirliğin araştırılabilmesi amacıyla 20 ölçüm tekrarlanmıştır. Radyografilerin kalibrasyonunun işaretlenmesinden analiz ekranı ortaya çıkana kadarki süre kaydedilmiştir. İstatistiksel incelemede eşleştirilmiş t-testi, Wilcoxon işaretli sıralar testi ve gözlemci içi tutarlılığın belirlenmesi amacıyla ICC testi kullanılmıştır. Eşleştirilmiş örneklem t-testi sonuçlarına göre Sağ Z uzunluğu, Sol Z uzunluğu, Sağ J'den A6 vertikal mesafesi ve Sol Me-Ag mesafesi parametreleri için Dolphin ve WebCeph gözlemleri arasındaki fark istatistiksel olarak anlamlı bulunmuştur (p<0.05). Wilcoxon işaretli sıralar testine göre ise x düzleminde Z (sağ ve sol), CG, Co (sağ ve sol) ve B6 (sol) noktalarında, y düzleminde ise ZA (sol), NC (sağ ve sol), Me, B1 (sol), A6 (sağ ve sol) ve A1 (sağ ve sol) noktalarında istatistiksel olarak anlamlı farklılık bulunmuştur (p<0,05). Dolphin programına göre yapay zekâ tabanlı WebCeph programında çalışma süresinin 10 kat daha düşük olduğu tespit edilmiştir. Çalışmamız sonucunda hipotezimiz doğrulanmış, WebCeph programı otomatik anteroposterior radyografi analizinde %71,6 oranında başarılı bulunmuştur.
The aim of our thesis study is evaluate the reliability and accuracy of the WebCeph program, an artificial intelligence-supported cephalometry drawing program, in anteroposterior cephalometric measurements by comparing it with measurements made manually in a digital environment with the Dolphin program (v.12.0). For this purpose, anteroposterior cephalometric radiographs of 50 patients, taken for diagnosis and treatment planning purposes before treatment, were selected in accordance with the inclusion criteria. Our study was planned to be carried out according to Grummons analysis, which includes a total of 23 anatomical landmarks and 21 measurements. First of all, the radiographs were measured with the Dolphin program, then they were uploaded to the web-based WebCeph program, their calibration was carried out manually, and their analysis was completed through artificial intelligence digitization. Then, the obtained analyzes were superimposed and the distance between the same anatomical landmark determined through Dolphin and WebCeph programs was measured by ImageJ program and recorded in x and y coordinates. Then, 20 measurements were repeated to investigate intraobserver reliability. The time from marking the calibration of the radiographs until the analysis screen appeared was recorded. In statistical analysis, paired t-test, Wilcoxon signed-rank test and ICC test were used to determine intra-observer consistency. According to the paired sample t-test results, the difference between Dolphin and WebCeph observations for the parameters Right Z length, Left Z length, A6 vertical distance from Right J and Left Me-Ag distance were found to be statistically significant (p<0.05). According to the Wilcoxon signed-rank test, on the x plane, Z (right and left), CG, Co (right and left) and B6 (left) points, on the y plane, ZA (left), NC (right and left), Me, B1 (left), A6 (right and left) and A1 (right and left) points were found to be statistically significant (p <0.05). It has been determined that the working time in the artificial intelligence-based WebCeph program is 10 times lower than the Dolphin program. As a result of our study, our hypothesis was confirmed and the WebCeph program was found to be 71,6% successful in automatic anteroposterior radiography analysis.
The aim of our thesis study is evaluate the reliability and accuracy of the WebCeph program, an artificial intelligence-supported cephalometry drawing program, in anteroposterior cephalometric measurements by comparing it with measurements made manually in a digital environment with the Dolphin program (v.12.0). For this purpose, anteroposterior cephalometric radiographs of 50 patients, taken for diagnosis and treatment planning purposes before treatment, were selected in accordance with the inclusion criteria. Our study was planned to be carried out according to Grummons analysis, which includes a total of 23 anatomical landmarks and 21 measurements. First of all, the radiographs were measured with the Dolphin program, then they were uploaded to the web-based WebCeph program, their calibration was carried out manually, and their analysis was completed through artificial intelligence digitization. Then, the obtained analyzes were superimposed and the distance between the same anatomical landmark determined through Dolphin and WebCeph programs was measured by ImageJ program and recorded in x and y coordinates. Then, 20 measurements were repeated to investigate intraobserver reliability. The time from marking the calibration of the radiographs until the analysis screen appeared was recorded. In statistical analysis, paired t-test, Wilcoxon signed-rank test and ICC test were used to determine intra-observer consistency. According to the paired sample t-test results, the difference between Dolphin and WebCeph observations for the parameters Right Z length, Left Z length, A6 vertical distance from Right J and Left Me-Ag distance were found to be statistically significant (p<0.05). According to the Wilcoxon signed-rank test, on the x plane, Z (right and left), CG, Co (right and left) and B6 (left) points, on the y plane, ZA (left), NC (right and left), Me, B1 (left), A6 (right and left) and A1 (right and left) points were found to be statistically significant (p <0.05). It has been determined that the working time in the artificial intelligence-based WebCeph program is 10 times lower than the Dolphin program. As a result of our study, our hypothesis was confirmed and the WebCeph program was found to be 71,6% successful in automatic anteroposterior radiography analysis.
Açıklama
Diş Hekimliği Fakültesi, Ortodonti Ana Bilim Dalı
Anahtar Kelimeler
Diş Hekimliği, Dentistry
