On June 11th and 12th medial students Benjamin Sjögren and Yousef Hamid successfully presented their master thesis.
Benjamin's thesis investigated non-invasive hyperspectral imaging in skin cancer diagnostics.
Skin cancer is one of the most common types of cancer worldwide. In order to improve patient survival through early diagnosis, and to minimize the risk of non-radical removal of tumors, several non-invasive imaging techniques have been developed during the past decades. However, no method, apart from dermatoscopy, has become widely established in the clinical setting. Hyperspectral imaging as a tool for skin cancer diagnostics has gained interest in the last ten years but it has yet to be applied clinically. This pilot study was conducted to explore the potential of hyperspectral imaging in skin cancer diagnostics. A hyperspectral imaging camera was used to obtain spectral information from four skin tumor samples ex vivo. The spectral information was used to obtain the average absorbance spectra for each pixel in order to detect specific spectral signatures for different skin tumors but also between different regions of the same skin tumor samples. Multivariate analysis was used to delineate the borders between healthy skin tissue and skin tumor tissue. A distinct difference in the spectral signature of basal cell carcinoma, squamous cell carcinoma and melanoma was found, especially in the wavelength range 600 - 900 nm. Within a single sample of melanoma in situ, spectra extracted from regions between healthy tissue and the tumor exhibited a gradual increase of absorbance, as well as changes in the spectral shape, in the ranges 600 - 900 nm and 1400 - 1500 nm. Lastly, the spectral information made it possible to develop an algorithm capable of delineating the tumor borders of three different skin tumor types. In coclusion, Benjamin's pilot study shows that hyperspectral imaging has the potential to non-invasively obtain spectral information to be used for pre-operative skin cancer diagnostics.
The incidence rate of cutaneous melanoma has increased substantially in recent years. The histologically measured thickness, or Breslow’s depth, is the single most important prognostic indicator and guides treatment. Several non-invasive imaging techniques have been developed to preoperatively assess Breslow’s thickness and thus facilitate the clinical management of melanoma. However, these have proven to be limited either by an underwhelming resolution, or an inability to visualize deep-lying structures. Photoacoustic imaging (PAI) is a novel, non-invasive method that combines laser with high frequency ultrasound (HFUS) to provide images at great depths and with impressive spectral and structural resolutions. The aim of the master thesis was to find a spectral signature unique for melanoma, and by this create 3-dimensional (3-D) images in order to compare Breslow’s depth measured with PAI to that determined by pathologists. PAI was performed on 14 histologically confirmed melanomas ex vivo using a Vevo LAZR-X multimodal imaging system, however, only nine were used for analysis due to poor signal quality in the remaining five lesions. The obtained spectral signature for melanoma was used in the spectral unmixing process to generate a 3-D image for each skin sample. Histological measurements were obtained for five of the nine melanomas analyzed. The 2-dimensional (2-D) cross section with the largest vertical spread of each sample was compared with histologically measured Breslow’s depth. A significant difference between the mean spectral signature for melanoma and healthy tissue was found in the 680-950 nm avelength range. Performing spectral unmixing allowed for 3-D reconstruction of an entire skin sample, distinguishing melanoma from healthy tissue. A strong correlation (m = 1.028; R2 = 0.923) was found between photoacoustic (PA) measured Breslow’s depth and histological Breslow’s depth, although without statistical significance. In coclusion, Yousef's study shows that PAI is a novel, non-invasive, non-ionizing imaging modality that shows great promise in the preoperative assessment of tumor margins and depth.