IMTOP, IMage-TO-Print Wound Dressings: From a Single Calibrated RGB Photograph to a Patient-Specific Patch

Patient-specific dressings for diabetic foot ulcers (DFUs) must follow the individual lesion, yet image-to-fabrication pipelines rely on depth sensing and seldom report the accuracy of the fabricated object. We present IMTOP, a low-cost pipeline that turns a single calibrated RGB photograph into a printable patient-specific patch, and we budget its area error stage by stage. Two terms were measured: on 18 polylactic acid phantoms printed by fused deposition modelling, geometry was reconstructed to 0.32% area error and printing deviated by 0.69%; on 149 public clinical DFU images (uncalibrated, single-operator reference), three reference-seeded segmenters gave median patch area errors of 3.9% to 6.0%. The scale term was simulated (assumed 3 px placement noise on a 117 px reference; standard deviation 7.3%). Monte Carlo propagation of the terms gave an estimated median absolute total error of 7.3% to 8.0%; segmentation and scale were comparable, and printing contributed under 0.1% of the variance. Mask overlap correlated with, but did not determine, the patch error. A normal perilesional offset of 14% to 21% of the wound equivalent diameter covered the wound in about 94% of cross-validated cases. The results support the feasibility of the approach as a proof of concept.