Usefulness of in vivo reflectance confocal microscopy and automated videomosaics in the treatment and management of skin cancers = Ús de la microscòpia confocal de reflectància in vivo i dels videomosaics automatitzats en el tractament i seguiment dels càncers cutanis

Abstract

[eng] Skin cancer is major health problem due to its increased incidence in the last decades. Standard treatment of skin cancer is still surgery, which can be challenging to perform since most skin cancers are located in visible areas such as the face. Hence, methods to improve skin cancer treatment and minimize scarring are crucial. In this sense, non-invasive imaging technologies can be of great help in determining the margins of skin cancers prior to their treatment, as well as they offer a great opportunity to monitor its treatment and detect early relapses. Reflectance confocal microscopy (RCM) is one of this non-invasive imaging technologies which has cellular resolution and has shown to have high diagnostic accuracy for different types of skin cancers such as keratinocyte carcinomas, melanomas or Paget disease. Currently two RCM microscopes are commercially available: 1) a wide-probe microscope which requires attaching a metal/plastic ring on the skin to obtain image mosaics up to 8 x 8 mm; 2) a handheld RCM (HRCM) microscope which allows free imaging along the skin with a field of view which ranges from 0.75 to 1 mm. HRCM has the advantage that permits imaging areas larger than 8 x 8 mm by navigating over the skin but lacks mosaicking capabilities in its native software, making orientation complicated. However, it is intuitive to think that HRCM is the ideal tool to predefine the margins of skin cancers and to monitor its responses to treatment due to its versatility, especially when the size of tumor is larger than 8 mm. The goal of the current thesis was to develop a computer algorithm to obtain static 2D mosaics obtained from dynamic videos obtained with HRCM, and to apply this technique to address some of the challenges that physicians encounter when managing skin cancers, such as defining the skin cancer margins prior to surgery (particularly important in some skin cancers such as lentigo maligna [LM] which have ill-defined margins), identifying residual tumor directly in the surgical wound, and identifying small foci of tumor in recurrent or persistent skin cancers which present in a patchy fashion such as extramammary Paget’s disease (EMPD). The research for this thesis resulted in 5 publications summarized herein. The first publication describes the videomosaicking algorithm as well as clinical to showcase how this algorithm can be used not only in RCM but in other microscopic imaging systems. The second publication applies the videomosaicking algorithm to delineate the preoperative margins in a consecutive series of 23 LM/LM melanoma, showing that the HRCM with videomosaicking can correctly predict the surgical defect after staged excision with minimal variations. In the third publication, HRCM was used to identify persistence or recurrence of EMPD after different treatments. Here, the videomosaicking algorithm was used in some of the patients and showed that the diagnostic accuracy improved when videomosaics were added to the conventional HRCM examination. In the fourth and fifth publications HRCM and videomosaics were used to image directly in the surgical wound after treating keratinocyte carcinomas (after laser ablation in the 4th publication, and after Mohs surgery in the 5th publication), showing it is possible to identify residual tumor using HRCM and videomosaics directly inside the surgical wound. In summary, this thesis shows how videomosaics expand the usefulness of HRCM in skin cancer management and overcome its main challenges by expanding the field of view and obtaining architectural information which is crucial when evaluating microscopic images.

[cat] El càncer de pell representa un important problema de salut pública ja que la seva incidència no ha parat d’augmentar en les darreres dècades. El tractament estàndard del càncer de pell segueix essent la cirurgia, que sovint pot ser complexa i potencialment morbida ja que la majoria dels càncers de pell aparèixen en sones zones visibles com la cara. Per tant, qualsevol mètode que permeti millorar-ne el tractament i minimitzar-ne les cicatrius és crucial. En aquest sentit, les tecnologies d'imatge no invasives poden ser de gran ajuda per determinar els marges tumorals abans del seu tractament, controlar la resposta al tractament i detectar la presència de persistència o recurrència de forma precoç. La microscòpia confocal de reflectància és una d'aquestes tecnologies d'imatge no invasiva que posseeix una resolució cel·lular i ha demostrat tenir una alta precisió diagnòstica per als diferents tipus de càncers de pell. Actualment dos microscopis confocals de reflectancia estan disponibles al mercat: 1) un microscopi de sonda àmplia que requereix adherir un anell de metall/plàstic a la pell per tal obtenir mosaics d'imatges de fins a 8 x 8 mm; 2) un microscopi de mà que permet obtenir imatges de forma lliure amb un camp de visió que oscil·la entre 0,75 i 1 mm. El microscopi de mà té l’avantatge que permet avaluar àrees superiors a 8 x 8 mm navegant pel damunt de la pell segons les necessitats de l’operador, però no té la capacitat de generar mosaic en el seu programari natiu, de manera que l’orientació és complicada. No obstant, és lògic pensar que el microscopi de mà és l’eina ideal per predefinir els marges dels càncers de pell i controlar-ne la resposta al tractament gràcies a la seva versatilitat, especialment quan la mida del tumor és superior a 8 mm. L’objectiu d’aquesta tesi doctoral ha estat desenvolupar un algoritme informàtic per obtenir mosaics 2D estàtics obtinguts a partir de vídeos dinàmics obtinguts amb el microscopi confocal de mà, i aplicar aquesta tècnica per abordar alguns dels reptes que els metges es troben en fer front a càncers de pell, com ara definir els marges dels càncers de pell abans de la cirurgia (especialment important en alguns d’ells com el lentigen maligne que tenen marges mal definits), identificar el tumor residual directament a la ferida quirúrgica, així com identificar petits focus de tumor recurrents o persistents en entitats com la malaltia de Paget extramamària o carcinomes queratinocítics.