A customizable 32P hydrogel applicator for brachytherapy of skin hemangioma based on machine learning and 3D-printing

Abstract

Skin hemangioma is a tumor originating from skin blood vessels, which often occurs in infants and children. Brachytherapy with the ³²P-based radionuclide applicator is an effective non-invasive therapeutic method. However, the inordinance of lesions is still the main challenge for precise local treatment and radiation protection of normal skins. A radionuclide applicator possessing advanced shape adaptability, favorable radionuclide biodistribution, optimized stress feature, and convenient preparation method is highly required for clinical practice. Herein, we present a customizable polyacrylamide (PAAm) hydrogel-based radionuclide applicator, integrating automatic lesion recognition via machine learning and 3D printing technology. The machine learning algorithm achieved a geometric accuracy of 98.78% in automated lesion contour recognition, providing guaranteed data support for 3D printing. The optimized hydrogel exhibited excellent mechanical properties (elastic modulus: 228 kPa, fracture toughness: 4.51 MJ m⁻³), rapid curing (<10 min), and promising ³²P loading efficiency (>85%). Especially, this system greatly shortened the fabrication time while ensuring precise geometric matching for complex lesions. Through in vitro cell and in vivo tumor-bearing mouse models, the hydrogel loaded with ³²P (P-HG) demonstrated favorable biocompatibility and effective therapeutic efficacy. It is believed that the synergy of intelligent recognition, 3D printing, and enhanced hydrogel performance can establish a promising treatment method with great practical potential for precise fitting brachytherapy of skin hemangioma.