Diagnostic image analysis of malignant melanoma in in vivo confocal laser-scanning microscopy: a preliminary study

Background/purpose: In this study we assessed the applicability of image analysis and a machine learning algorithm on diagnostic discrimination of benign and malignant melanocytic skin tumours in in vivo confocal laser-scanning microscopy (CLSM).
Methods: A total of 857 CLSM tumour images including 408 benign nevi and 449 melanoma images was evaluated. Image analysis was based on features of the wavelet transform. For classification purposes we used a classification tree software (CART). Moreover, automated image analysis results were compared with the prediction success of an independent human observer.
Results: CART analysis of the whole set of CLSM tumour images correctly classified 97.55% and 96.32% of melanoma and nevi images. In contrast, sensitivity and specificity of 85.52% and 80.15% could be reached by the human observer. When the image set was randomly divided into a learning (67% of the images) and a test set (33% of the images), overall 97.31% and 81.03% of the tumour images in the learning and test set could be classified correctly by the CART procedure.
Conclusion: Provided automated decisions can be used as a second opinion. This can be valuable in assisting diagnostic decisions in this new and exciting imaging technique.     

Diagnostic applicability of in vivo confocal laser scanning microscopy in melanocytic skin tumors

In vivo confocal laser scanning microscopy (CLSM) represents a novel imaging tool that allows the examination of skin morphology in real time at a resolution equal to that of conventional microscopes. The aim of the study was to test the applicability of CLSM to the diagnostic discrimination of benign nevi and melanoma. five independent observers without previous experience in CLSM received a standardized instruction about diagnostic CLSM features. Subsequently, 117 melanocytic skin tumors (90 benign nevi and 27 melanoma), imaged using a commercially available, near-infrared, reflectance confocal laser scanning microscope, were evaluated by each observer. Overall, sensitivity of 88.15% and specificity of 97.60% was achieved by the five observers. Logistic regression analysis revealed that mainly cytomorphology, architecture and keratinocyte cell borders should be taken into account for diagnostic decisions. Remarkably, using the presence or absence of monomorphic melanocytes as a single diagnostic criterion, the classification results with a sensitivity of 98.15% and a specificity of 98.89% were superior to the intuitive, integrative judgement of the observers. This first sensitivity and specificity study with CLSM has yielded promising results. CLSM provides new and useful information to the clinician diagnosing melanocytic skin tumors.     

In vivo confocal laser scanning microscopy of melanocytic skin tumours: diagnostic applicability using unselected tumour images

BACKGROUND: In vivo confocal laser scanning microscopy (CLSM) represents a novel imaging tool that allows the noninvasive examination of skin cancer morphology in real time at a 'quasi-histopathological' resolution viewing microanatomical structures and individual cells. OBJECTIVES: To validate diagnostic confocal examination of melanocytic skin tumours using unselected tumour images. METHODS: In the present study, we used a total of 3709 unselected CLSM tumour images obtained from 20 malignant melanomas and 50 benign naevi. The entire set of images derived from each tumour was evaluated by independent observers. Classification tree analysis based on a subsample of 857 tumour images was performed to develop a diagnostic algorithm. RESULTS: Overall, sensitivity and specificity of 97.5% and 99% could be achieved by the independent observers (positive predictive value 97.5%, negative predictive value 99%). Classification tree analysis yielded a three-step algorithm based on only three morphological CLSM features, facilitating a correct classification in 92.4% of the benign naevus images and 97.6% of melanoma images. CONCLUSIONS: In vivo CLSM augurs a sea change in the way we will view skin tumour processes clinically at the bedside and merits application for use as a screening tool in skin oncology.     

Diagnostic tissue elements in melanocytic skin tumors in automated image analysis

In tissue counter analysis, digital images are divided into subregions (elements), and the digital information in each element is used for statistical analysis. In this study, we assessed the morphologic details of tissue elements that have turned out to be of diagnostic significance in the discrimination of benign common nevi and malignant melanoma. After creation of a data set based on a total of 12,000 cellular elements obtained from 100 benign common nevi and 100 malignant melanomas, classification and regression tree (CART) analysis was performed to differentiate between cellular elements of nevi and melanoma. In a second step, the slides were re-evaluated by the decision tree; cellular elements suggestive either for benign common nevi or for malignant melanoma were highlighted on zoomed images of the whole sections, and the individual elements were displayed in galleries. Eight groups of elements (so-called terminal nodes) seemed to indicate benign common nevi, whereas seven terminal nodes were suggestive for malignant melanoma. The elements of nodes suggestive for benign nevi largely contained nevus cells with amphiphilic cytoplasm intermingled with fibrillary material, whereas the elements of the nodes suggestive for malignant lesions often showed hyperchromatism, perinuclear halos, heavy pigmentation, or a lymphohistiocytic infiltrate. Tissue counter analysis automatically detects tissue elements that are in accordance with morphologic criteria used in conventional histopathology for diagnostic discrimination.     

Expression of heat shock protein 105 and 70 in malignant melanoma and benign melanocytic nevi

Background:  Heat shock proteins (HSPs) restore immature proteins or denatured proteins, thus protecting cells. Also, the expression of some HSPs is elevated substantially in malignant tumors, but the expression of HSPs in association with melanoma has yet to be studied. Therefore, we examined the expression patterns of HSP 70 and 105 in melanoma, benign melanocytic nevi and normal human skin.
Methods:  Two specimens of malignant melanoma, two of benign melanocytic nevi and six of normal human skin were analyzed using Western blot analysis for expression of HSP 70 and 105. In another set, 16 specimens of malignant melanoma, 24 of benign melanocytic nevi and eight of normal human skin were analyzed for the expression of HSP 105 using immunohistochemical studies.
Results:  The Western blot analysis showed that HSP 70 was overexpressed in all three types. But, the HSP 105 was hardly expressed in normal human skin and benign melanocytic nevi. However, in malignant melanoma, the HSP 105 was overexpressed, and immunohistochemical examination of HSP 105 showed a result similar to that of Western blot analysis.
Conclusions:  In our study, HSP 105 is thought to be a more relevant tumor-associated antigen in malignant melanoma than is HSP 70.     

Automated epiluminescence microscopy—tissue counter analysis using CART and 1-NN in the diagnosis of Melanoma

Background/purpose: In tissue counter analysis, digital images are overlayed with regularly distributed measuring masks (elements) of equal size and shape, and the digital contents (grey level, colour and texture parameters) of each element are used for statistical analysis. In this study we assessed the applicability of tissue counter analysis and machine learning algorithms on tumour segmentation and diagnostic discrimination of benign and malignant melanocytic skin lesions.
Methods: A total of 369 standardised dermatoscopic images (93 melanomas, 276 benign nevi) were evaluated. The Classification and Regression Tree (CART) analysis was performed in order to differentiate between melanocytic skin lesions and surrounding skin. Instance-based learning (1-NN) was tested for differentiating between benign and malignant tumour elements. For diagnostic assessment, only the percentage of elements suggestive for malignancy in each lesion was used.
Results: Evaluation of a total of 369 melanocytic skin lesions showed a suitable segmentation of the tumour portion in 97.6%. When instance-based learning was applied to an independent test set, a threshold value of 27.4% of elements suggestive for malignancy recognised 35 out of 35 melanomas and 100 out of 101 nevi (sensitivity 100%, specificity 99%, positive predictive value 97.2%, negative predictive value 100%).
Conclusion: Tissue counter analysis combined with machine learning algorithms turned out to be a useful method for diagnostic purposes in epiluminescence microscopy.     

In vivo reflectance confocal microscopy: automated diagnostic image analysis of melanocytic skin tumours

Background In vivo reflectance confocal microscopy (RCM) has been shown to be a valuable imaging tool in the diagnosis of melanocytic skin tumours. However, diagnostic image analysis performed by automated systems is to date quite rare. Objectives In this study, we investigated the applicability of an automated image analysis system using a machine learning algorithm on diagnostic discrimination of benign and malignant melanocytic skin tumours in RCM. Methods Overall, 16 269 RCM tumour images were evaluated. Image analysis was based on features of the wavelet transform. A learning set of 6147 images was used to establish a classification tree algorithm and an independent test set of 10 122 images was applied to validate the tree model (grouping method 1). Additionally, randomly generated ‘new’ learning and test sets, tumour images only and different skin layers were evaluated (grouping method 2, 3 and 4). Results The classification tree analysis correctly classified 93.60% of the melanoma and 90.40% of the nevi images of the learning set. When the classification tree was applied to the independent test set 46.71 ± 19.97% (range 7.81–83.87%) of the tumour images in benign melanocytic skin lesions were classified as ‘malignant’, in contrast to 55.68 ± 14.58% (range 30.65–83.59%; t‐test: P < 0.036) in malignant melanocytic skin lesions (grouping method 1). Further investigations could not improve the results significantly (grouping method 2, 3 and 4). Conclusions The automated RCM image analysis procedure holds promise for further investigations. However, to date our system cannot be applied to routine skin tumour screening.     

Diagnostic imaging of melanocytic skin tumors

BACKGROUND: In tissue counter analysis, digital images are dissected into subregions (elements), and the digital information in each element is used for statistical analysis. The aim of this study was to test the applicability of tissue counter analysis and CART (Classification and Regression Tree) to the diagnostic discrimination of benign common nevi and malignant melanoma in dermatopathology. METHODS: Two hundred cases each of benign nevi and malignant melanoma were consecutively sampled. CART analyses of background versus tissue elements, cellular versus 'other' tissue elements and benign versus malignant cellular elements were performed. For diagnostic assessment, only the percentage of cellular elements suggestive for malignancy in each case was used. RESULTS: CART analysis led to a correct classification of 99% of background versus tissue elements, 96% of cellular versus 'other' tissue elements and 79.1% of benign versus malignant cellular elements. When the percentage of cellular elements suggestive for malignancy in each case was evaluated, 29.5 +/- 14% (range 4.1-62.4) 'malignant' elements were found in benign nevi (n = 200), in contrast to 75.9 +/- 13.9% (range 32.8-97.3) in melanoma (n = 200; z =-16.72, p < 0.001). It turned out that a threshold level of 52.51% provides a correct classification of 192 nevi and 186 melanoma out of 200 each (specificity 96%, sensitivity 93%, positive predictive value 95.9%). CONCLUSIONS: Tissue counter analysis combined with CART may be a useful method for diagnostic purposes in histopathology.     

Limitations of dermoscopy in the recognition of melanoma

OBJECTIVE: To compare dermoscopic features of melanocytic nevi with those of early melanomas that were not excised initially because of their uncharacteristic clinical and dermoscopic appearance. DESIGN: Retrospective study of the baseline images of 325 melanocytic skin lesions that were observed by digital dermoscopy and finally excised because of changes over time. SETTING: A dermatologic clinic and a dermatologic department at a university hospital. MAIN OUTCOME MEASURES: Comparison of baseline images of melanomas and melanocytic nevi by pattern analysis, the ABCD rule of dermoscopy, and the 7-point checklist. RESULTS: Baseline dermoscopic images of 262 melanocytic nevi and 63 melanomas from 315 patients were included in the analysis. The patterns of dermoscopic features observed in the baseline images of melanocytic lesions finally diagnosed as melanomas during follow-up did not differ substantially from the patterns observed in the baseline images of melanocytic nevi. Pattern analysis, the ABCD rule of dermoscopy, and the 7-point checklist failed to achieve adequate diagnostic accuracy for melanoma. In retrospect, no dermoscopic feature or pattern of features could be identified that reliably differentiated between melanomas and melanocytic nevi at the time of the first presentation. CONCLUSION: Dermoscopy depends on the appearance of classic dermoscopic features and is therefore limited in the diagnosis of very early and mainly featureless melanomas.     

Colour analysis of skin lesion regions for melanoma discrimination in clinical images

Background: Skin lesion colour is an important feature for diagnosing malignant melanoma. Colour histogram analysis over a training set of images has been used to identify colours characteristic of melanoma, i.e., melanoma colours. A percent melanoma colour feature defined as the percentage of the lesion pixels that are melanoma colours has been used as a feature to discriminate melanomas from benign lesions.
Methods: In this research, the colour histogram analysis technique is extended to evaluate skin lesion discrimination based on colour feature calculations in different regions of the skin lesion. The colour features examined include percent melanoma colour and a novel colour clustering ratio. Experiments are performed using clinical images of 129 malignant melanomas and 129 benign lesions consisting of 40 seborrheic keratoses and 89 nevocellular nevi.
Results: Experimental results show improved discrimination capability for feature calculations focused in the lesion boundary region. Specifically, correct melanoma and benign recognition rates are observed as high as 89 and 83%, respectively, for the percent melanoma colour feature computed using only the outermost, uniformly distributed 10% of the lesion's area.
Conclusions: The experimental results show for the features investigated that the region closest to the skin lesion boundary contains the greatest colour discrimination information for lesion screening. Furthermore, the percent melanoma colour feature consistently outperformed the colour clustering ratio for the different skin lesion regions examined. The clinical application of this result is that clustered colours appear to be no more significant than colours of arbitrary distribution within a lesion.     

WT1 marker is not sufficient for distinguishing between melanoma and melanocytic nevi

Background: The heterogeneous histological features of melanoma may often overlap with melanocytic nevi. For this reason, pathologists have sought after immunohistochemistry to assist with difficult cases. Recently, Wilms' tumor 1 protein (WT1) has been suggested to differentiate between melanoma and melanocytic nevi.
Objective: Our objective was to determine whether immunohistochemistry analysis of WT1 expression is a reliable tool in differentiating cutaneous melanoma from melanocytic nevi.
Methods: Forty-five melanoma and 43 melanocytic nevi were immunostained with anti-WT1 monoclonal antibody (clone 6F-H2).
Results: Forty of the 45 cutaneous melanoma (89%) and 22 of the 43 melanocytic nevi (51%) stained (> 10% cells) for WT1. The highest sensitivity for WT1 was expressed by nodular melanoma (19/20), superficial spreading melanoma (8/10) and Spitz nevi (9/11). At the threshold of above 75% WT1-stained cells, the specificity for melanoma was 95% but the sensitivity was only 31%. At the threshold of 10%, the sensitivity increased to 89% but the specificity decreased to only 49%. Finally, at the threshold of 25% and 50%, the sensitivity and specificity were 71%, 61% and 64%, 77%, respectively.
Conclusions: Our data suggest that melanoma is associated with increased WT1 expression. However, as a single immunostaining marker, WT1 is not sufficient for distinguishing melanoma from melanocytic nevi.     

Dermoscopy of an acral congenital melanocytic nevus

Congenital melanocytic nevi carry a risk for malignant transformation into melanoma, therefore early detection of suspicious features is crucial to reduce mortality rates. Dermoscopy improves the early detection of melanoma while reducing the number of unnecessary excisions of benign pigmented skin lesions. Dermoscopically, congenital melanocytic nevi are often characterized by the presence of a cobblestone pattern, but to date, little is known about the dermoscopic features of acral congenital melanocytic nevi. We report an acral congenital melanocytic nevus typified by the presence of three different dermoscopic patterns that are commonly seen in acquired melanocytic nevi of palms and soles.     

Follow-up of melanocytic skin lesions with digital epiluminescence microscopy: patterns of modifications observed in early melanoma, atypical nevi, and common nevi

BACKGROUND: Digital epiluminescence microscopy (DELM) has been reported to be a useful technique for the follow-up of melanocytic nevi. One of the promises of this technique is to identify modifications over time that indicate impending or incipient malignancy and to facilitate surveillance of melanocytic skin lesions, particularly in patients with multiple clinically atypical nevi. OBJECTIVE: Our purpose was to report on patterns of modifications over time observed in benign melanocytic skin lesions and melanoma. METHODS: A total of 1862 sequentially recorded DELM images of melanocytic lesions from 202 patients (mean age, 36.1 years; 54.0% female patients) with multiple clinically atypical nevi were included in the analysis. The median follow-up interval was 12. 6 months. Melanocytic lesions with substantial modifications over time (enlargement, changes in shape, regression, color changes or appearance of ELM structures known to be associated with melanoma) were excised and referred to histopathologic examination. RESULTS: A total of 75 melanocytic skin lesions (4.0%) from 52 patients (mean age, 33.3 years; 63.5% female patients) showed substantial modifications over time and were excised and referred to histopathologic examination. Eight changing lesions were histologically diagnosed as early melanomas. These lesions frequently showed focal enlargement associated with a change in shape as well as appearance of ELM structures that are known to be associated with melanoma. In contrast, the majority of benign changing lesions (common and atypical nevi) showed symmetric enlargement without substantial structural ELM changes. Six of the 8 patients in whom melanoma developed were unaware of the fact that the lesion had changed over time. CONCLUSION: We demonstrate that follow-up of melanocytic lesions with DELM helps to identify patterns of morphologic modifications typical for early melanoma. DELM may therefore serve as a useful tool to improve the surveillance of patients with multiple atypical nevi.     

Factors leading to the biopsy of 1547 pigmented lesions at Mayo Clinic, Scottsdale, Arizona, in 2005

Background  Both physician-driven and patient-driven factors influence biopsy decisions. We sought to determine the ratio of benign to malignant melanocytic biopsy findings in our general dermatology practice and to characterize the reasons for biopsy.
Methods  A retrospective review of institutional records (1 January to 31 December 2005) was undertaken.
Results  We identified 1398 nevi, 147 invasive and in situ melanomas, and two lesions interpreted as atypical melanocytic proliferations. Prior histories of melanoma, atypical nevi, or nonmelanoma skin cancer were common. Patient concerns about changes or symptoms drove about one-third of the biopsies. Physician concerns more commonly drove biopsies in men and older patients (> 60 years). Physician-directed biopsies more commonly yielded atypical nevi, but there was no difference in the likelihood of melanoma. The ratio of removed nevi to melanomas was 9.2 : 1.
Conclusions  Both patient-driven and physician-driven indications lead to skin biopsies. We found no standard method of documentation of dermoscopic evaluation, which prevented us from making definitive conclusions about the role of dermoscopy in this cohort.     

Automatic differentiation of melanoma from melanocytic nevi with multispectral digital dermoscopy: a feasibility study

BACKGROUND: Differentiation of melanoma from melanocytic nevi is difficult even for skin cancer specialists. This motivates interest in computer-assisted analysis of lesion images. OBJECTIVE: Our purpose was to offer fully automatic differentiation of melanoma from dysplastic and other melanocytic nevi through multispectral digital dermoscopy. METHOD: At 4 clinical centers, images were taken of pigmented lesions suspected of being melanoma before biopsy. Ten gray-level (MelaFind) images of each lesion were acquired, each in a different portion of the visible and near-infrared spectrum. The images of 63 melanomas (33 invasive, 30 in situ) and 183 melanocytic nevi (of which 111 were dysplastic) were processed automatically through a computer expert system to separate melanomas from nevi. The expert system used either a linear or a nonlinear classifier. The "gold standard" for training and testing these classifiers was concordant diagnosis by two dermatopathologists. RESULTS: On resubstitution, 100% sensitivity was achieved at 85% specificity with a 13-parameter linear classifier and 100%/73% with a 12-parameter nonlinear classifier. Under leave-one-out cross-validation, the linear classifier gave 100%/84% (sensitivity/specificity), whereas the nonlinear classifier gave 95%/68%. Infrared image features were significant, as were features based on wavelet analysis. CONCLUSION: Automatic differentiation of invasive and in situ melanomas from melanocytic nevi is feasible, through multispectral digital dermoscopy.     

Seborrheic keratoses, solar lentigines, and lichenoid keratoses. Dermatoscopic features and correlation to histology and clinical signs

Evaluation of the three benign lesions discussed here form the basis for dermoscopic evaluation of other pigmented skin lesions. The features of seborrheic keratosis, including [figure: see text] the various forms of fissures, comedo-like openings, and milia-like cysts, often allow easy interpretation of seborrheic keratosis; however, similar structures are commonly associated with melanocytic neoplasms, notably congenital nevi. Understanding solar lentigo and its dermoscopy features allows for the appreciation of pigment networks common in lentiginous melanocytic nevi and melanoma. The lichenoid keratosis is the model for lichenoid inflammation elsewhere, notably in halo nevi, regressing melanoma, and other melanocytic neoplasms with significant host inflammatory reactions.     

Melanoma and seborrheic keratosis differentiation using texture features

Purpose: To explore texture features in two-dimensional images to differentiate seborrheic keratosis from melanoma.
Methods: A systematic approach to consistent classification of skin tumors is described. Texture features, based on the second-order histogram, were used to identify the features or a combination of features that could consistently differentiate a malignant skin tumor (melanoma) from a benign one (seborrheic keratosis). Two hundred and seventy-one skin tumor images were separated into training and test sets for accuracy and consistency. Automatic induction was applied to generate classification rules. Data analysis and modeling tools were used to gain further insight into the feature space.
Result and Conclusions: In all, 85–90% of seborrheic keratosis images were correctly differentiated from the malignant skin tumors. The features correlation_average, correlation_range, texture_energy_average and texture_energy_range were found to be the most important features in differentiating seborrheic keratosis from melanoma. Over-all, the seborrheic keratosis images were better identified by the texture features than the melanoma images.     

Detection of atypical texture features in early malignant melanoma

Background: The presence of an atypical (irregular) pigment network (APN) can indicate a diagnosis of melanoma. This study sought to analyze the APN with texture measures.
Methods: For 106 dermoscopy images including 28 melanomas and 78 benign dysplastic nevi, the areas of APN were selected manually. Ten texture measures in the CVIPtools image analysis system were applied.
Results: Of the 10 texture measures used, correlation average provided the highest discrimination accuracy, an average of 95.4%. Discrimination of melanomas was optimal at a pixel distance of 20 for the 768 × 512 images, consistent with a melanocytic lesion texel size estimate of 4–5 texels per mm.
Conclusion: Texture analysis, in particular correlation average at an optimized pixel spacing, may afford automatic detection of an irregular pigment network in early malignant melanoma.     

Displacement of dermal solar elastosis in malignant melanoma

BACKGROUND: Ultraviolet radiation (UVR) is a major environmental causal factor for skin malignancy. In this study, we investigated the morphology of the solar elastosis (SE) band in benign and malignant melanocytic lesions. METHODS: We measured the SE band in perilesional and lesional skin of 13 melanomas (9 invasive and 4 in situ) and 11 melanocytic nevi (5 usual intradermal nevi, 4 blue nevi and 2 desmoplastic nevi) occurring in sun-exposed areas. RESULTS: The melanoma and nevus groups had similar age range, gender ratio and anatomic distribution. The mean SE thickness was 0.35 mm in melanomas and 0.29 mm in nevi (p = 0.56), indicating similar UVR exposure. There was a mean downward SE displacement (SED) of 0.43 mm in melanomas and essentially no displacement (-0.02 mm) in nevi (p < 0.005). Tumor cells and inflammatory host response were responsible for SED in melanoma. CONCLUSIONS: SED may help in the differential diagnosis of melanocytic lesions in sun-exposed areas. In melanoma, the new lesion depresses the pre-existing SE band. Conversely, the long-standing nevus co-exists with the SE band without significant displacement. Evaluation of the SE band may help to differentiate melanoma with chronic sun-induced damage as they have a distinct set of molecular alterations.     

Immunohistochemical expression of cyclooxygenage-2 in melanocytic skin lesions

Background Several reports have shown expression of cyclooxygenase‐2 (COX‐2) in malignant skin tumors. COX‐2 has also recently been reported as a marker of malignant melanoma (MM). Objective Our aim was to investigate whether there is a difference in the immunohistochemical expression of COX‐2 between malignant and benign melanocytic lesions of the skin. Methods We selected 40 archival cases of MM including 10 cases of superficial spreading melanoma, 10 of lentigo maligna melanoma, 10 of nodular melanoma, and 10 of acral lentiginous melanoma. For comparison, we also selected 35 benign melanocytic lesions, which included 15 nonatypical nevi and 10 atypical nevi. The remaining 10 cases were Spitz nevi. COX‐2 immunohistochemical staining was performed, and intensities were assessed quantitatively. Results The MM group and the benign melanocytic nevi group showed a highly statistically significant difference in the intensity of COX‐2 expression (P < 0.0001). Staining intensity in the dermal component of MM cases also showed a tendency to increase with increasing tumor depth. By contrast, the intensity of the dermal component in the melanocytic nevi group decreased with increasing depth as the nevus cells matured from type A to type C cells. No statistical difference was noted between the MM and Spitz nevi cases (P = 0.20). Conclusions Malignant melanoma shows stronger immunohistochemical expression of COX‐2 than benign melanocytic nevi. Although COX‐2 cannot be used alone to differentiate MM from melanocytic nevi, it may serve as an aid in the differential diagnosis of melanocytic skin lesions.