放射線光束角度優化是現代放射線治療的關鍵問題,在非共面設置下這可能具有挑戰性。雖然非共面放射治療技術具有潛在的劑量學優勢,但會使得發生碰撞的風險增加。 碰撞檢測在外部放射線光束治療中很重要,有助於降低確認醫師所設計的光束角度可用性所需的時間並防止病患與機器之間的碰撞。我們的目標是建立一個適用於各種病患的平台,以便在設計治療計劃之前,先計算碰撞空間,防止放射線治療機器因發生碰撞而損壞,並提高病患的安全性。如此可幫助醫師聰明地選擇安全和適用的放射光束角度,以優化劑量的分佈。 我們建立並驗證了兩種不同的模型。與我們所開發的應用程式預測結果相比,碰撞角度的預測誤差範圍都在7度以下。混淆矩陣驗證於兩種不同的線性加速器模型,準確率為0.989(分別為0.989和0.973)。真陽性率和真陰性率分別為0.978(0.978和0.969)和0.998(0.998和0.980)。 我們所開發的應用程式是一個簡單、有效、方便的工具,不需要額外的設備即可模擬碰撞檢測,並進一步幫助放射治療計劃聰明地進行放射線光束角度優化。
Beam angle optimization is a critical issue for modern RT, and this could be challenging under non-coplanar settings. Non-coplanar radiation therapy techniques have potential dosimetry advantages but with the increase in the risk of collision. Collision detection is important in external beam radiation therapy to help eliminate the needed time to confirm the usability of selected beam angles and prevent patient machine collisions. Our goal is to establish a platform which is patient-specific to calculate the collision space before treatment planning, prevent hardware damage, and improve patient safety to help physicians choose the safe and applicable beam angles wisely for enhancing dosimetric distributions. Two different models were built and validated. Compared with our application prediction, the collision angles were predicted within the different range of 7 degrees. The confusion matrix was applied for two different linear accelerators models, and the accuracy was 0.989 (0.989 and 0.973 respectively). True positive rate (TPR) and true negative rate (TNR) were 0.978 (0.978 and 0.969) and 0.998 (0.998 and 0.980) respectively. Our application is a simple, effective, and convenient tool to simulate the collision detection without the extra equipment and further help radiation therapy treatment planning smarty for the beam angle optimization.