Assessment of pigmented skin lesions in terms of blood perfusion estimates

Background/aims: Cutaneous malignant melanoma is a disease of increasing clinical and economical importance. The prognosis is good with early diagnosis. The chief differential diagnosis is benign melanocytic naevus, a common lesion in Caucasians. Attempts have been made to use bioengineering techniques to aid in the initial diagnosis. The present study proposes a method of extracting possibly discriminative blood perfusion properties in pigmented skin lesions by combining information on the lesions' blood perfusion with optical or visual information of their spatial extent. Methods: A total of 46 blood perfusion measurements were performed on 22 pigmented skin lesions, the ultimate diagnosis of which was three histologically proven malignant melanomas, four histologically proven benign naevi and fifteen naevi assessed by two specialist dermatologists as being benign. Laser Doppler perfusion imaging gave two different types of two-dimensional data sets (64x64 pixels), one representing the total backscattered light intensity at each measurement point (TLI image) and the second corresponding to perfusion values. The boundaries of each examined lesion were derived from the TLI image employing greyscale thresholding, thus resulting in an estimated region of interest (ROI) approximating the optical extent of the lesion. The ROI was superimposed on the perfusion image and extraction of perfusion features was then performed. Results: The processing of the TLI images was successful in delineating the lesions' boundaries. The first hypothesis that the mean perfusion quotients in MM and benign naevi are equal could not be rejected at the chosen 5% level of significance. The second hypothesis that the mean percent-age of elevated perfusion values (image pixels) within the ROI shows no difference between MM and benign naevi could be rejected at a 5% level of significance. Conclusions: This study has presented a method of extracting blood perfusion parameters of pigmented skin lesions by combining blood perfusion information with information on the lesion's optical extent. The proposed method of presenting data could prove to be a useful discriminative adjunct in the assessment of pigmented skin lesions.