Development and Implementation of a Telemedicine System for Wound Care in Different UE Environments.

The project focused on evaluating the introduction of a telemedical device across various European Countries. The Wound Viewer (WV) is a device designed to support the diagnosis and monitoring of skin ulcers by capturing an image of the lesion and using an integrated AI algorithm to detect and segment the area of interest, providing classification, area and depth of the wound. With a view to potentially integrating the WV into the healthcare system in the field of wound care, an analysis was conducted on the current costs dedicated to chronic wound care in different European countries. Based on this analysis, an evaluation was made of the hypothetical cost reduction that could occur if the device were adopted at a national scale. The calculation was based on the results obtained from a trial conducted at a hospital in Asti and on the percentage of people affected by skin ulcers each year worldwide (2% of the population). Specifically, the focus was on Denmark, where there is currently an integration of the device's functionalities with the national telemedical platform Plaja.net for the treatment of skin ulcers. However, one of the biggest challenges is the volume of data generated daily in healthcare. The significant proliferation of new technologies and the digital transition in healthcare have led to the generation of a vast amount of data, often incompatible with the available data transmission and storage infrastructures. For this reason, to introduce a device like the WV at a national level, an innovative method for data compression was evaluated to significantly reduce data size, facilitating transmission and lowering storage costs. The implemented method is Compressed Sensing (CS), a technique designed for signal acquisition at a sampling frequency lower than that established by the Nyquist Theorem, followed by digital reconstruction of the original signal. In the current application context, CS was applied to WV images at various compression levels to assess the relationship between the extent of compression and the loss of relevant features for ulcers classification. To evaluate the compression quality, a Convolutional Neural Network (CNN) was used for automatic detection and classification of lesions within the image. This approach allowed quantification of the loss of relevant information for ulcers classification following images compression and reconstruction by analyzing the classification performance on input images at different compression levels. The work resulted in images that were up to 13 times smaller in size compared to the originals, while maintaining reasonable classification performance (Accuracy: 0.7). Thus, promising results were obtained that could be further investigated by expanding the dataset for network training. The developed method produced promising results for the introduction of a device like the Wound Viewer at a national level, significantly reducing the costs of data transmission and storage generated by the device. At the same time, it presents a potential method for image encryption, enhancing data protection by limiting sharing to only authorized individuals.