Image Features of LSCI and Thermography for Prediction of Healing Trajectory in Diabetic Foot Patients Affected by the First (or Successive) Ulcer: (MY-FOOT-B)

The study aims to predict healing at 26 weeks after the first visit in patients affected by the first ulcer, by means of combined data monitoring of Laser Speckle Contrast Imaging (LSCI) and temperature measurements during patient visits in hospital.

In order to achieve this objective, the study aims to produce a logistic regression model and then evaluate its prognostic ability by means of the area under the curve (AUC) of the receiver-operating-characteristics (ROC) curve.

The variables evaluated for the model will include demographic and baseline characteristics of the patients, location and seriousness of the ulcer, alongside LSCI and temperature measurements at baseline and different timepoints. The development of the model will include a rigorous variables selection in order to produce a parsimonious model.

In the method proposed by Mennes, LSCI measurements at baseline, biological zero, post occlusion peak, and other parameters like non-invasive blood pressure measurements were individually evaluated as possible prognostic factors of healing trajectory at 26 weeks. All these parameters, when assessed at their highest possible value of sensibility and specificity, produced AUCs always inferior to 0.65. The highest value of AUC (0.625) was reached by toe pressure parameter when calculated using a threshold value of 54 mmHg.

From diagnostic literature, an AUC of ≥ 0.8 is considered excellent result, while an AUC < 0.7 is considered less than acceptable. Since the Mennes's method showed these results, study investigators set the AUC resulting from the model under the null hypothesis (H0) to be equal to 0.65. By including all these parameters, alongside the other mentioned variables, in a single prognostic model, study investigators expect to increase this prognostic ability. Therefore, in this study, investigators would like to detect an AUC ≥ 0.8 (H1). With these hypotheses, a power of 80%, a one-sided I type error of 5% and a prevalence of 43% of healed patients at 26 weeks, a total of 82 patients should be enrolled.