Optical imaging of tumor vasculature and angiogenesis

Tomanič, Tadej (author), Milanič, Matija (mentor), Markelc, Boštjan (comentor)

Assessing human tumors and their vasculature in the head and neck region requires advanced imaging methods capable of capturing detailed physiological and morphological features. The main goal of the research presented in this thesis was to implement a multimodal optical imaging system combining hyperspectral imaging (HSI) and laser speckle contrast imaging (LSCI) into the clinical environment to assess human tumors and tumor vasculature in the head and neck region. Three fundamental research objectives were identified to achieve the main goal: 1) the development of the multimodal optical imaging system combining HSI and LSCI; 2) the implementation of the imaging system in the preclinical environment to monitor murine tumor models; 3) the translation of the imaging system into the clinical environment to human tumors. The multimodal system was first developed and characterized, integrating HSI and LSCI to enable comprehensive imaging of tumors and their vasculature. The HSI system captured the spectral signatures of tissue describing physiology, pathology, and morphology, while the LSCI system provided information about blood flow and tissue perfusion. The hyperspectral image analysis pipeline utilized an inverse adding-doubling (IAD) algorithm, which was tested for accuracy and robustness in extracting key tissue properties. Also, blood vessel segmentation algorithms and vascular metrics were employed to quantify vascular morphology. In preclinical studies, the system was employed to monitor murine tumor models, including subcutaneous solid tumors and dorsal skinfold window chamber (DSWC) models. These studies demonstrated the ability to capture tumor growth, vascular changes, and response to treatments. In subcutaneous tumor models, biological features of CT26 tumors and their vasculature were identified, while differentiation of B16-F10, MC38, MOC1, and MOC2 tumors during growth was achieved. Also, the response of MC38 tumors to radiotherapy was observed. Moreover, the DSWC model offered superior capabilities for monitoring blood vessels, providing detailed insights into blood vessel growth in B16-F10 and MC38 tumors. Also, a response of 4T1 tumors to GET treatment was studied. Ultimately, the system was adapted for clinical use to assess human skin cancer, namely basal cell carcinoma (BCC) and squamous cell carcinoma (SCC) in the head and neck region. A portable HSI system was utilized for imaging, and a machine learning (ML) algorithm was developed to differentiate between tumor types based on their tissue properties. This thesis provides critical insights into the use of multimodal optical imaging techniques to improve tumor diagnosis and growth monitoring, as well as tumor blood vessel development. Building on the knowledge gained from preclinical studies, we translated the technology into clinical practice, bridging the gap between preclinical and clinical applications.

• Language: Slovenian
• Keywords: hyperspectral imaging, laser speckle contrast imaging, optical profilometry, tumors, vasculature, murine models, angiogenesis, biomedical optics
• Work type: Doctoral dissertation
• Typology: 2.08 - Doctoral Dissertation
• Organization: FMF - Faculty of Mathematics and Physics
• Year: 2025
• PID: 20.500.12556/RUL-169196
• COBISS.SI-ID: 236187395
• Publication date in RUL: 16.05.2025

Secondary language

• Language: English
• Title: Optično slikanje tumorskega ožilja in angiogeneze

Abstract

Diagnoza tumorjev ter spremljanje njihovega ožilja zahteva napredne slikovne metode, ki so sposobne zajeti natančne fiziološke in morfološke značilnosti tkiv. Glavni cilj doktorske disertacije je vpeljati večmodalni optični slikovni sistem, ki združuje hiperspektralno slikanje (HSI) in lasersko slikanje s pegami (LSCI), v klinično okolje za ovrednotenje človeških tumorjev in tumorskega ožilja v predelu glave in vratu. Za uresničitev glavnega cilja smo opredelili tri specifične raziskovalne cilje: 1) razvoj večmodalnega optičnega sistema, ki združuje HSI in LSCI; 2) uporabo slikovnega sistema v predkliničnem okolju za spremljanje mišjih tumorskih modelov, 3) vpeljavo slikovnega sistema v klinično okolje za spremljanje človeških tumorjev. V začetku smo razvili in karakterizirali večmodalni slikovni sistem, ki združuje HSI in LSCI ter na ta način omogoča celovito slikanje tumorjev in tumorskega ožilja. Sistem HSI zajame spektralne podatke, ki opisujejo fiziološke, patološke in strukturne lastnosti tkiv, medtem ko LSCI podaja informacije o pretoku krvi v žilah in prekrvavitvi tkiv. Za analizo hiperspektralnih slik smo uporabili inverzni algoritem dodajanje-seštevanje (IAD), pri čemer smo testirali natančnost in ponovljivost določanja tkivnih lastnosti z algoritmom. Prav tako smo uporabili algoritme za obris ožilja s slik ter količine, ki ovrednotijo strukturo ožilja. V sklopu predkliničnih študij smo uporabili slikovni sistem za spremljanje mišjih tumorjev, vključno s čvrstimi tumorji podkožja in tumorji, ki rastejo znotraj dorzalnih oken (DSWC). S študijami smo pokazali, da lahko spremljamo rast in razvoj tumorjev ter ožilja, prav tako pa lahko zaznamo spremembe, povezane z odzivom na zdravljenje. Pri čvrstih tumorjih smo identificirali pomembne biološke lastnosti tumorjev CT26, prav tako pa smo pokazali, da lahko razlikujemo tumorje B16-F10, MC38, MOC1 in MOC2 med rastjo in razvojem. Obenem smo spremljali uspešnost zdravljenja tumorjev MC38 z obsevanjem (radioterapijo). V nadaljevanju smo ugotovili, da so dorzalna okna primernejša za opazovanje krvnih žil. Med drugim smo pridobili vpogled v rast novih žil (angiogenezo) pri tumorjih B16-F10 in MC38 ter spremljali odziv tumorjev 4T1 na zdravljenje z genskim elektroprenosom (GET). Naposled smo slikovni sistem vpeljali v klinično rabo za diagnozo kožnega raka pri ljudeh, in sicer bazalnoceličnih karcinomov (BCC) ter ploščatoceličnih karcinomov (SCC) na predelu glave in vratu. Pri tem smo za slikanje uporabili prenosno hiperspektralno kamero, za analizo podatkov pa smo razvili algoritem strojnega učenja (ML), ki je omogočil razlikovanje med različnimi tipi tumorjev na osnovi njihovih optičnih lastnosti. V doktorski disertaciji smo omogočili vpogled v uporabo večmodalnih optičnih slikovnih tehnik za izboljšanje diagnoze tumorjev, spremljanje rasti tumorjev ter razvoja ožilja. Na podlagi znanja, pridobljenega v predkliničnih študijah, smo tehnologijo uspešno prenesli v klinično prakso, s čimer smo premostili vrzel med predklinično in klinično uporabo.

• Keywords: hiperspektralno slikanje, lasersko slikanje s pegami, optična profilometrija, tumorji, ožilje, mišji modeli, angiogeneza, biomedicinska optika