fix hyper-links

README.md

@@ -57,7 +57,7 @@ The ASSL module was developed using the [active_learning_validator](https://gith
 Using the obtained segmentation mask, we developed a second step of processing for the automated estimation of the PWAT score from the image.
 This second workflow of analysis is described by the scheme proposed in the Figure.
 
-| <img src="https://github.com/Nico-Curti/Deepskin/blob/master/img/pipeline.png" width=600> |
+| <img src="https://github.com/Nico-Curti/Deepskin/blob/main/img/pipeline.png" width=600> |
 | -- |
 | Schematic representation of the pipeline. The image is acquired by the smartphone (Deepskin dataset) during clinical practice. Two expert clinicians performed the manual annotation of the PWAT score associated to each wound, considering the status of the lesion and peri-lesion areas. The neural network model trained on the Deepskin dataset performs the automated segmentation of the wound area. Focusing on the wound and peri-wound areas (obtained by image processing analyses), a set of features for the quantification of textures and morphology of the lesion are extracted. A regression model based on the features extracted from the images is tuned for the automated prediction of the PWAT scores. While Step 1 and Step 2 requires the human intervention, by definition, the second half of the pipeline automatically performed the analysis. We would like to stress that the first two steps are mandatory for the training of the automated solution but are discarded during real clinical applications. |
 
@@ -67,7 +67,7 @@ we implemented a second step of automated image processing for the identificatio
 Using a combination of erosion and morphological dilation operators, keeping fixed the size of the structuring element (kernel) involved, we extracted for each image the associated peri-wound mask.
 An example of the resulting processing is showed in the Figure.
 
-| <img src="https://github.com/Nico-Curti/Deepskin/blob/master/img/wound_mask.png" width=600> |
+| <img src="https://github.com/Nico-Curti/Deepskin/blob/main/img/wound_mask.png" width=600> |
 | -- |
 | Example of segmentation masks used for wound identification. (a) Raw image extracted from Deepskin dataset. (b) Wound segmentation mask generated by automated neural network model. (c) Peri-wound segmentation mask obtained applying morphological operators on wound mask. |
 
@@ -77,7 +77,7 @@ In this way, we aimed to maximize the informative power of the features extracte
 A set of fully interpretable features will be extracted from the wound and peri-wound ROIs, and given as input for a regression model trained on a large dataset of annotated samples.
 The resulting regression model is able to determine the PWAT score of the wound, as showed in the Figure.
 
-| <img src="https://github.com/Nico-Curti/Deepskin/blob/master/img/pwat_results.png" width=600> |
+| <img src="https://github.com/Nico-Curti/Deepskin/blob/main/img/pwat_results.png" width=600> |
 | -- |
 | Example of the predictions obtained by the regression model on three test images. We report the assigned PWAT score and the predicted one for each image using our model. We highlighted the wound areas identified by our automated segmentation model with the green lines. |