AI-assisted 3D-printed transvaginal template guidance for interstitial brachytherapy in patients with cervical cancer with parametrial invasion.
Authors
Affiliations (4)
Affiliations (4)
- Department of Gynecology, Clinical Oncology School of Fujian Medical University, Fujian Cancer Hospital, Fuzhou, Fujian, China. Electronic address: [email protected].
- Department of Gynecology, Clinical Oncology School of Fujian Medical University, Fujian Cancer Hospital, Fuzhou, Fujian, China.
- Department of Radiation Physics, Clinical Oncology School of Fujian Medical University, Fujian Cancer Hospital, Fuzhou, Fujian, China.
- Department of Obstetrics and Gynecology, Putian First Hospital, Putian, Fujian, China.
Abstract
Three-dimensional (3D) interstitial brachytherapy (BT) can improve bulky cervical cancer treatment efficacy. BT technology guided by 3D-printed templates can assist radiation oncologists in accurately inserting needles and ensuring optimal dose coverage. Artificial intelligence (AI) holds promise for enhancing the accuracy, precision, efficiency, and overall quality of radiotherapy (RT) for cancer patients. This retrospective study aimed to evaluate the safety and efficacy of using AI-assisted transvaginal 3D-printed (AI/3D-printed) templates for guiding interstitial BT as a part of definitive RT for patients with cervical cancer with parametrial invasion. Localization data from computed tomography scans of 17 patients were gathered and imported into the software. Using AI-assisted technology, individualized 3D-printed templates were automatically configured based on the specific anatomical morphology and volume of the target. To ensure the precision and consistency of the AI/3D-printed template, the patient's clitoris and anus were used as localization markers. Needle positions in the template were adjusted to maintain a designated distance from the insertion needles to the tip of the tailbone. The difference in the distance between the actual interstitial needle and the designed needle to the tailbone was aimed to be less than 2 mm. The target dose dosimetric parameters were evaluated and compared between AI/3D-printed applicator guidance and free-hand (FH) insertion methods, with one fraction of AI/3D-printed BT paired with one fraction of FH BT per patient for direct comparison. No instances of severe bleeding or infection associated with puncture were observed. Compared with FH BT, the AI/3D-printed approach required no intraoperative needle adjustment. The mean (with standard deviation) values of the dose-volume histogram (DVH) parameters of AI/3D-printed BT showed significantly higher high-risk clinical target volume (HR-CTV) V100 (90.19 ± 0.72% vs. 88.89 ± 0.11%, p = 0.011) and HR-CTV D98 (5.67 ± 0.14 Gy vs. 5.55 ± 0.13 Gy, p = 0.044), but no significant difference in D90 (6.80 ± 0.10 Gy vs. 6.79 ± 0.10 Gy, p = 0.163). For organs at risk (OARs), the AI/3D-printed approach resulted in lower doses to the bladder (4.92 ± 0.06 Gy vs. 5.10 ± 0.06 Gy, p = 0.004) and rectum (3.95 ± 0.10 Gy vs. 4.24 ± 0.05 Gy, p = 0.01), but no significant difference for the sigmoid (2.99 ± 0.13 Gy vs. 3.23 ± 0.14 Gy, p = 0.125). The AI/3D-printed approach exhibited significantly superior conformal index(CI), dose homogeneity index (HI), and overdose volume index (OI) compared with the FH approach. AI/3D-printed applicator-guided BT for cervical cancer with parametrial invasion demonstrated successful implementation, significant dosimetric benefits to organs at risk (OARs) and better plan quality indicators (CI, HI, and OI), and minimal treatment-related complications. This method represents a promising advancement in BT for cervical cancer treatment.