Localized myxedema histologically mimicking spindle cell lipoma.

In this report, a 55-year-old woman with Graves disease and exophthalmos had a recurrent nodule on the foot. Her initial biopsy and excision specimens were believed to be consistent with spindle cell lipoma, which aligned with her early tumor-like clinical morphology. Her tumor recurred after excision, which is not consistent with spindle cell lipoma. As her condition progressed, her clinical morphology became more consistent with localized myxedema and her biopsies were congruent, securing clinicopathologic correlation. With standard treatment for localized myxedema, she improved significantly. This case emphasizes how clinicians need to have high suspicion for localized myxedema in patients with history of Graves disease and exophthalmos. It also emphasizes how localized myxedema should be included in the histologic differential diagnosis for spindle cell lipoma with prominent myxoid stroma, particularly in those not responding to treatment as anticipated.

Iris pigmented lesions as a marker of cutaneous melanoma risk: an Australian case-control study.

BACKGROUND: Iris naevi and iris freckles have a frequency of 4% and 50% in the European population, respectively. They are associated with dysplastic naevi, but few studies have examined their link to cutaneous melanoma. OBJECTIVES: To assess whether iris pigmented lesions are a predictive indicator for cutaneous melanoma. METHODS: This is a melanoma case-control study of 1254 European-background Australians. Sun exposure and melanoma history, a saliva sample for DNA analysis and eye photographs taken with a digital camera were collected from 1117 participants. Iris images were assessed by up to four trained observers for the number of iris pigmented lesions. The data were analysed for correlations between iris pigmented lesions and melanoma history. RESULTS: Case participants over the age of 40 had similar numbers of iris pigmented lesions to age matched controls (mean 5·7 vs. 5·2, P = 0·02), but in younger case and control participants there was a greater difference (mean 3·96 vs. 2·19, P = 0·004). A logistic regression adjusted for age, sex, skin, hair and eye colour, skin freckling and naevus count found that the presence of three or more iris pigmented lesions increases the melanoma risk 1·45-fold [95% confidence interval (CI) 1·07-1·95]. HERC2/OCA2 rs12913832 and IRF4 rs12203592 influenced both eye colour and the number of iris pigmented lesions. On the HERC2/OCA2 A/A and A/G genotype background there was an increasing proportion of blue eye colour when carrying the IRF4 T allele (P = 3 × 10-4 ) and a higher number of iris pigmented lesions with the IRF4 T/T homozygote (P = 3 × 10-9 ). CONCLUSIONS: Iris pigmented lesion count provides additional predictive information for melanoma risk above that from conventional risk factors.

Quantifying acceptable artefact ranges for dermatologic classification algorithms.

Background: Many classifiers have been developed that can distinguish different types of skin lesions (e.g., benign nevi, melanoma) with varying degrees of success.1-5 However, even successfully trained classifiers may perform poorly on images that include artefacts. While problems created by hair and ink markings have been published, quantitative measurements of blur, colour and lighting variations on classification accuracy has not yet been reported to our knowledge. Objectives: We created a system that measures the impact of various artefacts on machine learning accuracy. Our objectives were to (1) quantitatively identify the most egregious artefacts and (2) demonstrate how to assess a classification algorithm's accuracy when input images include artefacts. Methods: We injected artefacts into dermatologic images using techniques that could be controlled with a single variable. This allows us to quantitatively evaluate the impact on the accuracy. We trained two convolutional neural networks on two different binary classification tasks and measured the impact on dermoscopy images over a range of parameter values. The area under the curve and specificity-at-a-given-sensitivity values were measured for each artefact induced at each parameter. Results: General blur had the strongest negative effect on the melanoma versus other task. Conversely, shifting the hue towards blue had a more pronounced effect on the suspicious versus follow task. Conclusions: Classifiers should either mitigate artefacts or detect them. Images should be excluded from diagnosis/recommendation when artefacts are present in amounts outside the machine perceived quality range. Failure to do so will reduce accuracy and impede approval from regulatory agencies.