Monitoring the response to radiotherapy in cancerous skin lesions: a quantitative analysis of vascular changes using OCTA

Skin cancer represents the most prevalent form of malignancy worldwide, primarily categorized into cutaneous carcinomas—basal cell carcinoma and squamous cell carcinoma—and melanomas. Among the treatments for cutaneous carcinoma, radiotherapy is considered an effective option, particularly kilovolt radiotherapy. Superficial kV-based radiotherapy uses low-energy X-rays, depositing most energy in the first few millimeters of tissue. This makes it suitable for treating superficial lesions, minimizing damage to underlying tissues and enhancing treatment precision. Angiographic optical coherence tomography (OCTA) is employed to monitor treatment response. This non-invasive technique enables volumetric rendering of the tissue's vascular network without contrast agents. OCTA provides insights into tissue morphology and vascular structure, with an imaging depth of up to 1 mm, facilitating detailed visualization of epidermal and dermal microstructures. This is key for lesion analysis, as changes in vascularization can indicate angiogenesis related to tumor growth, providing important information on therapy effectiveness. The primary goal of this thesis is to assess skin superficial lesions at three distinct time points: before radiotherapy,one month after, and three months post-treatment.The quantitative analysis of vascular network parameters aims to identify significant variations between these time points and correlate them with treatment response and effects on skin tissue. The study involves five patients with a total of nine lesions, analyzing 22 volumes overall—12 lesion volumes and 10 volumes of healthy skin.The image processing pipeline includes preprocessing the raw data into OCT matrices, applying an intensity-based method for OCTA reconstruction, and performing artifact removal. Emphasis is placed on deep color coding and semi-automatic volume segmentation with Amira software. The vascular parameters extracted through the skeletonization of the segmented volume include vascular density, vessel radius, distance metric, sum of angles metric, number of trees, number of branches, and inflection count metric. The results focus on the trends of extracted parameters for each lesion of each patient across three acquisition sessions. Maximum and average values of the parameters are computed across the patients at the different time points to observe changes over time. This approach facilitates comparisons among the three sessions, highlighting any statistically significant differences. The results indicate significant changes in some of the analyzed parameters,such as vascular density vessel radius, distance metric, number of branches and inflection count metric, suggesting an evolution of the skin lesions during time. Additionally, comparing the extracted vascular parameters of healthy skin and each patient's lesions reveals that the vascular network is crucial for distinguishing healthy from diseased skin. Finally, this thesis highlights the possibility of performing therapeutic follow-up non-invasively through the use of OCTA, presenting a significant opportunity for future development. However, limitations include the possibility that some lesions may not show significant improvements in vascular parameters due to acquisition artifacts and inaccuracies in positioning across sessions. Additionally, post-radiotherapy tissue inflammation may elicit immune responses that can affect metrics and delay observable treatment effects.